MACHINE LEARNING-BASED SYSTEM FOR ESTIMATING GLUCOSE VALUES
An optimized population model that estimates blood glucose values for a population of users is generated by mapping received data for the population of users over a time period to a sequence of estimated blood glucose values for the population of users over the time period. Discrete blood glucose measurement data for each user, user activity data for each user, and other contextual data for each user can be processed via a supervised machine learning model to learn a transfer function for a population model that estimates blood glucose values for the population of users. One or more parameters of the learning model can be adjusted to generate the optimized population model.
The present technology is generally related to insulin delivery systems and, more specifically, to a machine learning-based system for estimating glucose values based on, for example, blood glucose measurements and contextual activity data.
BACKGROUNDPortable medical devices are useful for patients that have conditions that must be monitored on a continuous or frequent basis. For example, diabetics are usually required to modify and monitor their daily lifestyle to keep their blood glucose (BG) in balance. Individuals with Type 1 diabetes and some individuals with Type 2 diabetes use insulin (and other blood sugar lowering medications) to control their BG levels. To maintain glucose levels within a recommended range, patients with diabetes are advised to routinely keep strict schedules, including ingesting timely nutritious meals, partaking in exercise, monitoring BG levels daily, and adjusting and administering insulin dosages accordingly.
Infusion pump devices and insulin pump systems are relatively well known in the medical arts. Infusion pump devices and insulin systems are designed to deliver accurate and measured doses of insulin via infusion sets. An infusion set delivers the insulin to a patient through a small diameter tube that terminates at, e.g., a cannula inserted under the patient's skin. Use of infusion pump therapy has been increasing, especially for delivering insulin for diabetics.
In one type of system, the infusion pump can be programmed to inject insulin on a set schedule. For instance, a doctor or other healthcare administrator can program an insulin pump according to a set schedule so that it delivers insulin into the patient's bloodstream according to that schedule throughout the day. In addition, the patient can also simply activate the insulin pump to administer an insulin bolus as needed, for example, in response to the patient's high BG level. One drawback to this approach is that the user's lack knowledge of their blood glucose levels in any given point, and it is difficult for patients to determine when it may be necessary to administer insulin outside their scheduled administrations.
To address this problem, a patient can monitor BG levels using a BG meter or measurement device. For instance, a patient can utilize a blood glucose meter to intermittently measure their instantaneous blood sugar levels at any given time via a “finger prick test.” Information from that measurement can be processed and used to determine whether insulin should be administered to regulate the patient's blood sugar level. A problem with this approach is that monitoring of blood glucose levels and the administration of insulin is done by the user on an ad hoc basis based on when they think insulin should be administered. This can be problematic in certain situations, such as with patients who experience hypoglycemia unawareness, which is a condition that can occur where a patient cannot perceive symptoms that might be indicators that their blood sugar is low.
To address this issue, continuous glucose monitoring (CGM) systems can be used to help diabetic patients continuously monitor their blood glucose levels. Continuous glucose monitoring systems employ a continuous glucose sensor to monitor a patient's blood glucose level in a substantially continuous and autonomous manner. In many cases, a continuous glucose monitoring sensor is utilized in conjunction with an insulin infusion device or insulin pump as parts of a digital diabetes management system. The digital diabetes management system can determine the amount of glucose in the patient's blood at any given time so that an appropriate amount of insulin can be automatically administered to help regulate the amount of glucose in the patient's blood. This way insulin can be automatically administered depending on the patient's specific needs such that their blood sugar levels do not go too high or too low at any particular time. This way, an equilibrium or balance can be achieved between entry of glucose into the body (e.g., the amount of glucose that is being introduced into the patient's blood stream via meals) and how much glucose is being consumed or utilized by the patient.
While existing continuous glucose monitoring systems that employ a continuous glucose sensor work well, such continuous glucose monitoring systems are often too expensive for potential users to afford. In some cases, a patient's insurance does not cover such therapy or the patient's insurance may refuse to pay for the patient to use such a solution if the patient does not fall within a high enough risk category to be covered for such therapy. In addition, some patients may simply choose not to utilize continuous glucose monitoring systems for lifestyle reasons. For instance, a particular user may choose not to wear a continuous glucose monitor or sensor arrangement because it could be uncomfortable, or they choose not to wear it all the time for other reasons. This can prevent a large group of users from using a digital diabetes management solution.
Today, a large segment of the diabetes market does not utilize CGM-type therapy devices. A substantial proportion of these users include, for example, users who rely on discrete blood glucose measurements several times a day to monitor glycemia (e.g., those using blood glucose meters), or users who intermittently wear a CGM under supervision of a health care provider (HCP). Without using a CGM device on a regular basis, these approaches can make it difficult for users to manage glycemia without actively making meaningful behavioral adjustments to improve their individual management.
Accordingly, it is desirable to provide digital diabetes management solutions that are less expensive and can thus be an option for users who are unable to afford more expensive continuous glucose monitoring systems. It would be desirable to provide users with alternative solutions that can achieve the same or similar benefits of full CGM-type therapy solutions without requiring users to wear a CGM device on a regular basis and without requiring users to absorb the costs associated with wearing a CGM device on a regular basis. Furthermore, other desirable features and characteristics will become apparent from the subsequent detailed description and the appended claims, taken in conjunction with the accompanying drawings and the foregoing technical field and background.
BRIEF SUMMARYA method for estimating glucose values of a user and an insulin infusion and management system are provided. The insulin infusion and management system can include an insulin infusion device configured to deliver insulin to a user; a blood glucose meter; a source of user activity data; and a processor-based computing device that supports data communication with the insulin infusion device. The computing device can include a processor device and a non-transitory processor-readable medium operatively associated with the processor device. The processor-readable medium comprises executable instructions configurable to cause the processor device to perform a method for estimating glucose values of a user. In some embodiments, the processor device is implemented at a backend server system, and the estimation model is executed as part of a web application at the backend server system. The backend server system is configured to communicate results of processing the estimated glucose values to one or more of a client device and an insulin infusion device. In another embodiment, the processor device is implemented at a client device, and the estimation model is executed as part of a mobile application at the client device. In another embodiment, the processor device is implemented at an insulin infusion device. In accordance with the method, a first set of inputs can be received and processed via an estimation model for a user to generate a set of estimated glucose values that track actual glucose values. The first set of inputs can comprise contextual activity data collected by a source of user activity data. For example, in some embodiments, the contextual activity data can include metabolic data about the user, and/or physical activity data about the user.
In some embodiments, the estimation model is an ensemble model that comprises one or more machine learning algorithms; and a physiological model. The machine learning algorithms can include one or more of: one or more artificial neural networks; one or more regression models; one or more decision tree models; one or more support vector machines; one or more Bayesian networks; one or more probabilistic machine learning models; one or more Gaussian processing models, one or more Hidden Markov models; and one or more deep learning networks.
In accordance with other embodiments, the estimation model can receive and process other information in addition to the first set of inputs to generate the set of estimated glucose values. For example, in accordance with another embodiment, the estimation model can also receive and process a second set of inputs comprising intermittent glucose measurements provided from a blood glucose meter to generate the set of estimated glucose values. In this embodiment, the processor device that executes the estimation model for the user processes the first set of inputs and the second set of user inputs to generate the set of estimated glucose values without using information from a continuous glucose sensor arrangement. In accordance with another embodiment, the estimation model can also receive and process a third set of inputs comprising nutritional information input by a user, and a fourth set of inputs comprising insulin delivered by an insulin infusion device of the user to generate the set of estimated glucose values. In this embodiment, the processor device that executes the estimation model for the user processes the first set of inputs, the second set of inputs, the third set of inputs and the fourth set of inputs to generate the set of estimated glucose values.
In some embodiments, the processing system can regularly determine whether the estimated glucose values are within an acceptable accuracy range. When the estimated glucose values are not within the acceptable accuracy range, the processing system can cause a first notification that is perceptible to the user to be generated. When the estimated glucose values are within the acceptable accuracy range, the processing system can utilize the estimated glucose value for diabetes management including controlling delivery of insulin by an insulin infusion device based on the estimated glucose values.
This summary is provided to introduce a selection of concepts in a simplified form that are further described below in the detailed description. This summary is not intended to identify key features or essential features of the claimed subject matter, nor is it intended to be used as an aid in determining the scope of the claimed subject matter. The details of one or more aspects of the disclosure are set forth in the accompanying drawings and the description below. Other features, objects, and advantages of the techniques described in this disclosure will be apparent from the description and drawings, and from the claims.
The following detailed description is merely illustrative in nature and is not intended to limit the embodiments of the subject matter or the application and uses of such embodiments. As used herein, the word “exemplary” means “serving as an example, instance, or illustration.” Any implementation described herein as exemplary is not necessarily to be construed as preferred or advantageous over other implementations. Furthermore, there is no intention to be bound by any expressed or implied theory presented in the preceding technical field, background, brief summary or the following detailed description.
It should be understood that various aspects disclosed herein may be combined in different combinations than the combinations specifically presented in the description and accompanying drawings. It should also be understood that, depending on the example, certain acts or events of any of the processes or methods described herein may be performed in a different sequence, may be added, merged, or left out altogether (e.g., all described acts or events may not be necessary to carry out the techniques). In addition, while certain aspects of this disclosure are described as being performed by a single module or unit for purposes of clarity, it should be understood that the techniques of this disclosure may be performed by a combination of units or modules associated with, for example, a medical device.
In one or more examples, the described techniques may be implemented in hardware, software, firmware, or any combination thereof If implemented in software, the functions may be stored as one or more instructions or code on a computer-readable medium and executed by a hardware-based processing unit. Computer-readable media may include non-transitory computer-readable media, which corresponds to a tangible medium such as data storage media (e.g., RAM, ROM, EEPROM, flash memory, or any other medium that can be used to store desired program code in the form of instructions or data structures and that can be accessed by a computer).
Instructions may be configurable to be executed by one or more processors, such as one or more digital signal processors (DSPs), general purpose microprocessors, application specific integrated circuits (ASICs), field programmable logic arrays (FPGAs), or other equivalent integrated or discrete logic circuitry. Accordingly, the term “processor” as used herein may refer to any of the foregoing structure or any other physical structure suitable for implementation of the described techniques. Also, the techniques could be fully implemented in one or more circuits or logic elements.
Exemplary embodiments of the subject matter described herein are implemented in conjunction with medical devices, such as portable electronic medical devices. Although many different applications are possible, the following description focuses on embodiments that incorporate an insulin infusion device (or insulin pump) as part of an infusion system deployment. For the sake of brevity, conventional techniques related to infusion system operation, insulin pump and/or infusion set operation, and other functional aspects of the systems (and the individual operating components of the systems) may not be described in detail here. Examples of infusion pumps may be of the type described in, but not limited to, U.S. Pat. Nos. 4,562,751; 4,685,903; 5,080,653; 5,505,709; 5,097,122; 6,485,465; 6,554,798; 6,558,320; 6,558,351; 6,641,533; 6,659,980; 6,752,787; 6,817,990; 6,932,584; and 7,621,893; each of which are herein incorporated by reference.
Generally, a fluid infusion device includes a motor or other actuation arrangement that is operable to linearly displace a plunger (or stopper) of a fluid reservoir provided within the fluid infusion device to deliver a dosage of fluid, such as insulin, to the body of a user. Dosage commands that govern operation of the motor may be generated in an automated manner in accordance with the delivery control scheme associated with a particular operating mode, and the dosage commands may be generated in a manner that is influenced by a current (or most recent) measurement of a physiological condition in the body of the user. For example, in a closed-loop or automatic operating mode, dosage commands may be generated based on a difference between a current (or most recent) measurement of the interstitial fluid glucose level in the body of the user and a target (or reference) glucose setpoint value. In this regard, the rate of infusion may vary as the difference between a current measurement value and the target measurement value fluctuates. For purposes of explanation, the subject matter is described herein in the context of the infused fluid being insulin for regulating a glucose level of a user (or patient); however, it should be appreciated that many other fluids may be administered through infusion, and the subject matter described herein is not necessarily limited to use with insulin
In one non-limiting embodiment, a method and system are provided for estimating glucose values based, for example, on data from a BG meter and an activity tracker. In other embodiments, other contextual information, such as food (including carbohydrate and macronutrient intake) can also be used to estimate glucose values (e.g., to reconstruct continuous glucose data). Machine learning-based estimation/prediction models can be developed off-line for each patient, or updated online for continuous learning, particularly during retraining periods of continuous glucose monitor (CGM) taking in current contextual data. Generic estimation/predication models can be built using patterns/instances from data collected from a similar cohort as the patient which provides a bigger data pool and hence larger scope for patient characterization. Because every person is different the estimation/prediction models are uniquely trained for each patient.
For example, a customized machine learning-based estimation/prediction model can be built for each patient during a training process using CGM data. The customized machine learning-based estimation/prediction model can then be invoked at run time based on finger-stick measurements entered by the user via a blood glucose meter in addition to other contextual information such as steps, heart rate, exercise, metabolic equivalents and other parameters collected from an activity tracker, and insulin and meal information from insulin pumps (or insulin pens) or user entries, while also taking into account their medical history and demographic information so that the predictions are not only personalized based on the individual's diabetes profile and behavior, but also equipped to account for deviations from a patient's regular routine. The customized machine learning-based estimation/prediction model can thus provide continuous current and predictive estimates of glucose levels in real-time using finger-stick measurements entered by the user via a blood glucose meter and including any of the other contextual information mentioned above. Also, because the contextual data for a patient can change over time, the estimation/prediction models can be updated on a regular basis. In other words, because a patient's condition can change over time, a patient can wear a continuous glucose sensor periodically (e.g., once or twice a month) to acquire data that can be used to retrain the estimation model so that it can generate more accurate estimates for that particular patient.
The functionality can reside, for example, both within a web-based application as well as a mobile application. The estimation models can be implemented/deployed at either a backend server that runs the estimation model as part of a web-based application and communicates information back to the user's phone, insulin pump or other end device, and/or can also implemented as part of a mobile application that runs on the user's phone or other end device and communicates information back to the insulin pump or other end device. The estimation models can also be implemented/deployed at an insulin pump. Regardless of the implementation, the applications can send push notifications to the patient to alert them in case of an estimated low or high excursion event, can receive requests from user and deliver precise, coherent answers, and can also provide a pathway for entering data which can be used to keep track of the user history as well as improve the prediction estimates.
The estimation model can be trained using a continuous glucose sensor device, and once it is trained and then personalized/tuned for a particular user/patient, the continuous glucose sensor does not need to be used on a regular basis. This can allow users/patients to rely less on a continuous glucose sensor. They only need to use a CGM periodically instead of on a regular basis.
Instead, a blood glucose meter can be used in conjunction with an activity tracker to generate inputs to the estimation model, and the estimation model can then generate estimated glucose values that approximate measured glucose values that would be measured by a continuous glucose sensor. In essence, the disclosed embodiments can allow a patient to use an activity tracker and intermittent measurements from a blood glucose meter without needing a continuous glucose sensor on a regular basis (e.g., the patient can use a continuous glucose sensor periodically for calibration, but then only use a blood glucose meter and an activity tracker in conjunction with an estimation model the remainder of the time).
In some embodiments, additional types of data can be processed by the estimation model, which can improve the accuracy of the estimates. For example, a patient can estimate their glucose values using meal data that describes what the patient has recently eaten, intermittent blood glucose measurements from a blood glucose meter (that are taken by a user, for example, via finger stick measurements), and other contextual activity data that is received on a continuous basis from an activity tracker or other source of activity information about the patient. This other contextual activity data can include data such as physical activity data (e.g., heart rate, number of steps, or metabolic equivalents, etc.), metabolic data (e.g., calories burned), etc. This information can be used via an estimation or predication module to estimate glucose values in a way that approximates, with high accuracy, blood glucose measurements that would be measured by a continuous glucose sensor without having to incur the cost of, or permanently attach, a continuous glucose sensor. When the estimated glucose values are accurate enough, they can be used, for example, to drive an insulin infusion device so that it administers insulin to a patient, or to provide notifications to a patient or health care partner to manually inject a bolus of insulin on an as needed basis depending on their estimated glucose values and other information such as activity data, meal information, etc., or alternatively to provide health care providers with information to adjust basal rate settings of an inulin infusion device so that they are appropriate for a particular patient to meet their needs. It may also be used as a medication titration advisor for individuals on other types of blood sugar lowering medications. When the estimated glucose values are not accurate enough, then this condition can be used to trigger a notification, warning, alarm or alert, for example, at another device such as at the insulin pump, the user's smartphone or other smart device, etc. to inform the user that they should not trust the system and to recommend that the user to wear another sensor (e.g., CGM) to retrain the estimation model.
Thus, the proposed method and system can provide a new digital diabetes management solution that is cheaper and more affordable; more likely to be used by users who are unable to afford digital diabetes management systems; more likely to be approved by insurance, etc. The proposed method and system can also help users start using a digital diabetes management solution until they transition to a CGM-based system for diabetes management.
Turning now to
In the illustrated embodiment of
The sensing arrangement 104 generally represents the components of the infusion system 100 configured to sense, detect, measure or otherwise quantify a condition of the user, and may include a sensor, a monitor, or the like, for providing data indicative of the condition that is sensed, detected, measured or otherwise monitored by the sensing arrangement. In this regard, the sensing arrangement 104 may include electronics and enzymes reactive to a biological condition, such as a blood glucose level, or the like, of the user, and provide data indicative of the blood glucose level to the infusion device 102, the CCD 106 and/or the computer 108. For example, the infusion device 102, the CCD 106 and/or the computer 108 may include a display for presenting information or data to the user based on the sensor data received from the sensing arrangement 104, such as, for example, a current glucose level of the user, a graph or chart of the user's glucose level versus time, device status indicators, alert messages, or the like. In other embodiments, the infusion device 102, the CCD 106 and/or the computer 108 may include electronics and software that are configured to analyze sensor data and operate the infusion device 102 to deliver fluid to the body of the user based on the sensor data and/or preprogrammed delivery routines. Thus, in exemplary embodiments, one or more of the infusion device 102, the sensing arrangement 104, the CCD 106, and/or the computer 108 includes a transmitter, a receiver, and/or other transceiver electronics that allow for communication with other components of the infusion system 100, so that the sensing arrangement 104 may transmit sensor data or monitor data to one or more of the infusion device 102, the CCD 106 and/or the computer 108.
Still referring to
In some embodiments, the CCD 106 and/or the computer 108 may include electronics and other components configured to perform processing, delivery routine storage, and to control the infusion device 102 in a manner that is influenced by sensor data measured by and/or received from the sensing arrangement 104. By including control functions in the CCD 106 and/or the computer 108, the infusion device 102 may be made with more simplified electronics. However, in other embodiments, the infusion device 102 may include all control functions, and may operate without the CCD 106 and/or the computer 108. In various embodiments, the CCD 106 may be a portable electronic device. In addition, in various embodiments, the infusion device 102 and/or the sensing arrangement 104 may be configured to transmit data to the CCD 106 and/or the computer 108 for display or processing of the data by the CCD 106 and/or the computer 108.
In some embodiments, the CCD 106 and/or the computer 108 may provide information to the user that facilitates the user's subsequent use of the infusion device 102. For example, the CCD 106 may provide information to the user to allow the user to determine the rate or dose of medication to be administered into the user's body. In other embodiments, the CCD 106 may provide information to the infusion device 102 to autonomously control the rate or dose of medication administered into the body of the user. In some embodiments, the sensing arrangement 104 may be integrated into the CCD 106. Such embodiments may allow the user to monitor a condition by providing, for example, a sample of his or her blood to the sensing arrangement 104 to assess his or her condition. In some embodiments, the sensing arrangement 104 and the CCD 106 may be used for determining glucose levels in the blood and/or body fluids of the user without the use of, or necessity of, a wire or cable connection between the infusion device 102 and the sensing arrangement 104 and/or the CCD 106.
In some embodiments, the sensing arrangement 104 and/or the infusion device 102 are cooperatively configured to utilize a closed-loop system for delivering fluid to the user. Examples of sensing devices and/or infusion pumps utilizing closed-loop systems may be found at, but are not limited to, the following U.S. Pat. Nos. 6,088,608, 6,119,028, 6,589,229, 6,740,072, 6,827,702, 7,323,142, and 7,402,153 or United States Patent Application Publication No. 2014/0066889, all of which are incorporated herein by reference in their entirety. In such embodiments, the sensing arrangement 104 is configured to sense or measure a condition of the user, such as, blood glucose level or the like. The infusion device 102 is configured to deliver fluid in response to the condition sensed by the sensing arrangement 104. In turn, the sensing arrangement 104 continues to sense or otherwise quantify a current condition of the user, thereby allowing the infusion device 102 to deliver fluid continuously in response to the condition currently (or most recently) sensed by the sensing arrangement 104 indefinitely. In some embodiments, the sensing arrangement 104 and/or the infusion device 102 may be configured to utilize the closed-loop system only for a portion of the day, for example only when the user is asleep or awake.
As best illustrated in
The housing 202 is formed from a substantially rigid material having a hollow interior 214 adapted to allow an electronics assembly 204, a sliding member (or slide) 206, a drive system 208, a sensor assembly 210, and a drive system capping member 212 to be disposed therein in addition to the reservoir 205, with the contents of the housing 202 being enclosed by a housing capping member 216. The opening 220, the slide 206, and the drive system 208 are coaxially aligned in an axial direction (indicated by arrow 218), whereby the drive system 208 facilitates linear displacement of the slide 206 in the axial direction 218 to dispense fluid from the reservoir 205 (after the reservoir 205 has been inserted into opening 220), with the sensor assembly 210 being configured to measure axial forces (e.g., forces aligned with the axial direction 218) exerted on the sensor assembly 210 responsive to operating the drive system 208 to displace the slide 206. In various embodiments, the sensor assembly 210 may be utilized to detect one or more of the following: an occlusion in a fluid path that slows, prevents, or otherwise degrades fluid delivery from the reservoir 205 to a user's body; when the reservoir 205 is empty; when the slide 206 is properly seated with the reservoir 205; when a fluid dose has been delivered; when the infusion device 200 is subjected to shock or vibration; when the infusion device 200 requires maintenance.
Depending on the embodiment, the fluid-containing reservoir 205 may be realized as a syringe, a vial, a cartridge, a bag, or the like. In certain embodiments, the infused fluid is insulin, although many other fluids may be administered through infusion such as, but not limited to, HIV drugs, drugs to treat pulmonary hypertension, iron chelation drugs, pain medications, anti-cancer treatments, medications, vitamins, hormones, or the like. As best illustrated in
In the illustrated embodiment of
As best shown in
As illustrated in
The motor assembly 207 includes one or more electrical leads 236 adapted to be electrically coupled to the electronics assembly 204 to establish communication between the control electronics 224 and the motor assembly 207. In response to command signals from the control electronics 224 that operate a motor driver (e.g., a power converter) to regulate the amount of power supplied to the motor from a power supply, the motor actuates the drive train components of the drive system 208 to displace the slide 206 in the axial direction 218 to force fluid from the reservoir 205 along a fluid path (including tubing 221 and an infusion set), thereby administering doses of the fluid contained in the reservoir 205 into the user's body. Preferably, the power supply is realized one or more batteries contained within the housing 202. Alternatively, the power supply may be a solar panel, capacitor, AC or DC power supplied through a power cord, or the like. In some embodiments, the control electronics 224 may operate the motor of the motor assembly 207 and/or drive system 208 in a stepwise manner, typically on an intermittent basis; to administer discrete precise doses of the fluid to the user according to programmed delivery profiles.
Referring to
Referring to
In exemplary embodiments, the sensing arrangement 504 includes one or more interstitial glucose sensing elements that generate or otherwise output electrical signals (alternatively referred to herein as measurement signals) having a signal characteristic that is correlative to, influenced by, or otherwise indicative of the relative interstitial fluid glucose level in the body 501 of the patient. The output electrical signals are filtered or otherwise processed to obtain a measurement value indicative of the patient's interstitial fluid glucose level. For example, in one implementation, a sensor of the continuous glucose monitor (CGM) measures glucose indirectly via interstitial fluid using, e.g., a subcutaneous sensor. Interstitial fluid is the fluid that surrounds the cells of the patient's tissue below their skin, and usually glucose moves from their blood vessels and capillaries first and then into their interstitial fluid. Thus, sensor glucose readings are taken by the sensing arrangement 504 from the patient's interstitial fluid, and not from their blood, like finger stick measurements. For purposes of explanation, the calibrated blood glucose values that are calculated based on the electrical signals output by the sensing element(s) of the sensing arrangement 504 may alternatively be referred to herein as the sensor glucose (SG) value, the sensed glucose value, or variants thereof. There is a physiological lag between measurements of interstitial SG values versus blood glucose values. As such, in most cases the sensing arrangement 504 needs to be calibrated using a blood glucose meter 530. To explain further, the blood glucose meter 530 measures blood glucose values directly from a blood stream. As such, blood glucose values can refer to glucose values measured directly from a supply of a patient's blood. In this regard, the terms “blood glucose measurement,” “measured blood glucose,” “blood glucose concentration,” “measured blood glucose concentration,” and the like may refer to a glucose level, a blood glucose level, a blood glucose concentration, or the like, that has been obtained via any type of glucose sensor. In exemplary embodiments, a blood glucose meter 530, such as a finger stick device, is utilized to directly sense, detect, measure or otherwise quantify the blood glucose in the body 501 of the patient. In this regard, the blood glucose meter 530 outputs or otherwise provides a measured blood glucose value that may be utilized as a reference measurement for calibrating the sensing arrangement 504 and converting a measurement value indicative of the patient's interstitial fluid glucose level into a corresponding calibrated blood glucose value.
In exemplary embodiments, the infusion system 500 also includes one or more additional sensing arrangements 506, 508 configured to sense, detect, measure or otherwise quantify a characteristic of the body 501 of the patient that is indicative of a condition in the body 501 of the patient. In this regard, in addition to the glucose sensing arrangement 504, one or more auxiliary sensing arrangements 506 (also referred to herein as a source of activity data or “activity tracker”) may be worn, carried, or otherwise associated with the body 501 of the patient to measure characteristics or conditions of the patient (or the patient's activity) that may influence the patient's glucose levels or insulin sensitivity. For example, a heart rate sensing arrangement 506 could be worn on or otherwise associated with the patient's body 501 to sense, detect, measure or otherwise quantify the patient's heart rate, which, in turn, may be indicative of exercise (and the intensity thereof) that is likely to influence the patient's glucose levels or insulin response in the body 501. In yet another embodiment, another invasive, interstitial, or subcutaneous sensing arrangement 506 may be inserted into the body 501 of the patient to obtain measurements of another physiological condition that may be indicative of exercise (and the intensity thereof), such as, for example, a lactate sensor, a ketone sensor, or the like. Depending on the embodiment, the auxiliary sensing arrangement(s) 506 could be realized as a standalone component worn by the patient, or alternatively, the auxiliary sensing arrangement(s) 506 may be integrated with the infusion device 502 or the glucose sensing arrangement 504.
The illustrated infusion system 500 also includes an acceleration sensing arrangement 508 (or accelerometer) that may be worn on or otherwise associated with the patient's body 501 to sense, detect, measure or otherwise quantify an acceleration of the patient's body 501, which, in turn, may be indicative of exercise or some other condition in the body 501 that is likely to influence the patient's insulin response. While the acceleration sensing arrangement 508 is depicted as being integrated into the infusion device 502 in FIG. 5, in alternative embodiments, the acceleration sensing arrangement 508 may be integrated with another sensing arrangement 504, 506 on the body 501 of the patient, or the acceleration sensing arrangement 508 may be realized as a separate standalone component that is worn by the patient.
In the illustrated embodiment, the pump control system 520 generally represents the electronics and other components of the infusion device 502 that control operation of the fluid infusion device 502 according to a desired infusion delivery program in a manner that is influenced by the sensed glucose value indicating the current glucose level in the body 501 of the patient. For example, to support a closed-loop operating mode, the pump control system 520 maintains, receives, or otherwise obtains a target or commanded glucose value, and automatically generates or otherwise determines dosage commands for operating an actuation arrangement, such as a motor 532, to displace the plunger 517 and deliver insulin to the body 501 of the patient based on the difference between the sensed glucose value and the target glucose value. In other operating modes, the pump control system 520 may generate or otherwise determine dosage commands configured to maintain the sensed glucose value below an upper glucose limit, above a lower glucose limit, or otherwise within a desired range of glucose values. In practice, the infusion device 502 may store or otherwise maintain the target value, upper and/or lower glucose limit(s), insulin delivery limit(s), and/or other glucose threshold value(s) in a data storage element accessible to the pump control system 520. As described in greater detail, in one or more exemplary embodiments, the pump control system 520 automatically adjusts or adapts one or more parameters or other control information used to generate commands for operating the motor 532 in a manner that accounts for a likely change in the patient's glucose level or insulin response resulting from a meal, exercise, or other activity.
Still referring to
Still referring to
In exemplary embodiments, the energy source 518 is realized as a battery housed within the infusion device 502 (e.g., within housing 202) that provides direct current (DC) power. In this regard, the motor driver module 514 generally represents the combination of circuitry, hardware and/or other electrical components configured to convert or otherwise transfer DC power provided by the energy source 518 into alternating electrical signals applied to respective phases of the stator windings of the motor 532 that result in current flowing through the stator windings that generates a stator magnetic field and causes the rotor of the motor 532 to rotate. The motor control module 512 is configured to receive or otherwise obtain a commanded dosage from the pump control system 520, convert the commanded dosage to a commanded translational displacement of the plunger 517, and command, signal, or otherwise operate the motor driver module 514 to cause the rotor of the motor 532 to rotate by an amount that produces the commanded translational displacement of the plunger 517. For example, the motor control module 512 may determine an amount of rotation of the rotor required to produce translational displacement of the plunger 517 that achieves the commanded dosage received from the pump control system 520. Based on the current rotational position (or orientation) of the rotor with respect to the stator that is indicated by the output of the rotor sensing arrangement 516, the motor control module 512 determines the appropriate sequence of alternating electrical signals to be applied to the respective phases of the stator windings that should rotate the rotor by the determined amount of rotation from its current position (or orientation). In embodiments where the motor 532 is realized as a BLDC motor, the alternating electrical signals commutate the respective phases of the stator windings at the appropriate orientation of the rotor magnetic poles with respect to the stator and in the appropriate order to provide a rotating stator magnetic field that rotates the rotor in the desired direction. Thereafter, the motor control module 512 operates the motor driver module 514 to apply the determined alternating electrical signals (e.g., the command signals) to the stator windings of the motor 532 to achieve the desired delivery of fluid to the patient.
When the motor control module 512 is operating the motor driver module 514, current flows from the energy source 518 through the stator windings of the motor 532 to produce a stator magnetic field that interacts with the rotor magnetic field. In some embodiments, after the motor control module 512 operates the motor driver module 514 and/or motor 532 to achieve the commanded dosage, the motor control module 512 ceases operating the motor driver module 514 and/or motor 532 until a subsequent dosage command is received. In this regard, the motor driver module 514 and the motor 532 enter an idle state during which the motor driver module 514 effectively disconnects or isolates the stator windings of the motor 532 from the energy source 518. In other words, current does not flow from the energy source 518 through the stator windings of the motor 532 when the motor 532 is idle, and thus, the motor 532 does not consume power from the energy source 518 in the idle state, thereby improving efficiency.
Depending on the embodiment, the motor control module 512 may be implemented or realized with a general purpose processor, a microprocessor, a controller, a microcontroller, a state machine, a content addressable memory, an application specific integrated circuit, a field programmable gate array, any suitable programmable logic device, discrete gate or transistor logic, discrete hardware components, or any combination thereof, designed to perform the functions described herein. In exemplary embodiments, the motor control module 512 includes or otherwise accesses a data storage element or memory, including any sort of random access memory (RAM), read only memory (ROM), flash memory, registers, hard disks, removable disks, magnetic or optical mass storage, or any other short or long term storage media or other non-transitory computer-readable medium, which is capable of storing programming instructions for execution by the motor control module 512. The computer-executable programming instructions, when read and executed by the motor control module 512, cause the motor control module 512 to perform or otherwise support the tasks, operations, functions, and processes described herein.
It should be appreciated that
The communications interface 604 generally represents the hardware, circuitry, logic, firmware and/or other components of the pump control system 600 that are coupled to the pump control module 602 and configured to support communications between the pump control system 600 and the various sensing arrangements 504, 506, 508. In this regard, the communications interface 604 may include or otherwise be coupled to one or more transceiver modules capable of supporting wireless communications between the pump control system 520, 600 and the sensing arrangement 504, 506, 508. For example, the communications interface 604 may be utilized to receive sensor measurement values or other measurement data from each sensing arrangement 504, 506, 508 in an infusion system 500. In other embodiments, the communications interface 604 may be configured to support wired communications to/from the sensing arrangement(s) 504, 506, 508. In various embodiments, the communications interface 604 may also support communications with another electronic device (e.g., CCD 106 and/or computer 108) in an infusion system (e.g., to upload sensor measurement values to a server or other computing device, receive control information from a server or other computing device, and the like).
The pump control module 602 generally represents the hardware, circuitry, logic, firmware and/or other component of the pump control system 600 that is coupled to the communications interface 604 and configured to determine dosage commands for operating the motor 532 to deliver fluid to the body 501 based on measurement data received from the sensing arrangements 504, 506, 508 and perform various additional tasks, operations, functions and/or operations described herein. For example, in exemplary embodiments, pump control module 602 implements or otherwise executes a command generation application 610 that supports one or more autonomous operating modes and calculates or otherwise determines dosage commands for operating the motor 532 of the infusion device 502 in an autonomous operating mode based at least in part on a current measurement value for a condition in the body 501 of the patient. For example, in a closed-loop operating mode, the command generation application 610 may determine a dosage command for operating the motor 532 to deliver insulin to the body 501 of the patient based at least in part on the current glucose measurement value most recently received from the sensing arrangement 504 to regulate the patient's blood glucose level to a target reference glucose value. Additionally, the command generation application 610 may generate dosage commands for boluses that are manually-initiated or otherwise instructed by a patient via a user interface element.
In exemplary embodiments, the pump control module 602 also implements or otherwise executes a personalization application 608 that is cooperatively configured to interact with the command generation application 610 to support adjusting dosage commands or control information dictating the manner in which dosage commands are generated in a personalized, patient-specific manner. In this regard, in some embodiments, based on correlations between current or recent measurement data and the current operational context relative to historical data associated with the patient, the personalization application 608 may adjust or otherwise modify values for one or more parameters utilized by the command generation application 610 when determining dosage commands, for example, by modifying a parameter value at a register or location in memory 606 referenced by the command generation application 610. In yet other embodiments, the personalization application 608 may predict meals or other events or activities that are likely to be engaged in by the patient and output or otherwise provide an indication of the predicted patient behavior for confirmation or modification by the patient, which, in turn, may then be utilized to adjust the manner in which dosage commands are generated to regulate glucose in a manner that accounts for the patient's behavior in a personalized manner.
Still referring to
It should be understood that
In exemplary embodiments, the control system 700 receives or otherwise obtains a target glucose value at input 702. In some embodiments, the target glucose value may be stored or otherwise maintained by the infusion device 502 (e.g., in memory 606), however, in some alternative embodiments, the target value may be received from an external component (e.g., CCD 106 and/or computer 108). In one or more embodiments, the target glucose value may be calculated or otherwise determined prior to entering the closed-loop operating mode based on one or more patient-specific control parameters. For example, the target blood glucose value may be calculated based at least in part on a patient-specific reference basal rate and a patient-specific daily insulin requirement, which are determined based on historical delivery information over a preceding interval of time (e.g., the amount of insulin delivered over the preceding 24 hours). The control system 700 also receives or otherwise obtains a current glucose measurement value (e.g., the most recently obtained sensor glucose value) from the sensing arrangement 504 at input 704. The illustrated control system 700 implements or otherwise provides proportional-integral-derivative (PID) control to determine or otherwise generate delivery commands for operating the motor 532 based at least in part on the difference between the target glucose value and the current glucose measurement value. In this regard, the PID control attempts to minimize the difference between the measured value and the target value, and thereby regulates the measured value to the desired value. PID control parameters are applied to the difference between the target glucose level at input 702 and the measured glucose level at input 704 to generate or otherwise determine a dosage (or delivery) command provided at output 730. Based on that delivery command, the motor control module 512 operates the motor 532 to deliver insulin to the body of the patient to influence the patient's glucose level, and thereby reduce the difference between a subsequently measured glucose level and the target glucose level.
The illustrated control system 700 includes or otherwise implements a summation block 706 configured to determine a difference between the target value obtained at input 702 and the measured value obtained from the sensing arrangement 504 at input 704, for example, by subtracting the target value from the measured value. The output of the summation block 706 represents the difference between the measured and target values, which is then provided to each of a proportional term path, an integral term path, and a derivative term path. The proportional term path includes a gain block 720 that multiplies the difference by a proportional gain coefficient, KP, to obtain the proportional term. The integral term path includes an integration block 708 that integrates the difference and a gain block 722 that multiplies the integrated difference by an integral gain coefficient, KI, to obtain the integral term. The derivative term path includes a derivative block 710 that determines the derivative of the difference and a gain block 724 that multiplies the derivative of the difference by a derivative gain coefficient, KD, to obtain the derivative term. The proportional term, the integral term, and the derivative term are then added or otherwise combined to obtain a delivery command that is utilized to operate the motor at output 730. Various implementation details pertaining to closed-loop PID control and determining gain coefficients are described in greater detail in U.S. Pat. No. 7,402,153, which is incorporated by reference.
In one or more exemplary embodiments, the PID gain coefficients are patient-specific and dynamically calculated or otherwise determined prior to entering the closed-loop operating mode based on historical insulin delivery information (e.g., amounts and/or timings of previous dosages, historical correction bolus information, or the like), historical sensor measurement values, historical reference blood glucose measurement values, user-reported or user-input events (e.g., meals, exercise, and the like), and the like. In this regard, one or more patient-specific control parameters (e.g., an insulin sensitivity factor, a daily insulin requirement, an insulin limit, a reference basal rate, a reference fasting glucose, an active insulin action duration, pharmodynamical time constants, or the like) may be utilized to compensate, correct, or otherwise adjust the PID gain coefficients to account for various operating conditions experienced and/or exhibited by the infusion device 502. The PID gain coefficients may be maintained by the memory 606 accessible to the pump control module 602. In this regard, the memory 606 may include a plurality of registers associated with the control parameters for the PID control. For example, a first parameter register may store the target glucose value and be accessed by or otherwise coupled to the summation block 706 at input 702, and similarly, a second parameter register accessed by the proportional gain block 720 may store the proportional gain coefficient, a third parameter register accessed by the integration gain block 722 may store the integration gain coefficient, and a fourth parameter register accessed by the derivative gain block 724 may store the derivative gain coefficient.
In one or more exemplary embodiments, one or more parameters of the closed-loop control system 700 are automatically adjusted or adapted in a personalized manner to account for potential changes in the patient's glucose level or insulin sensitivity resulting from meals, exercise, or other events or activities. For example, in one or more embodiments, the target glucose value may be decreased in advance of a predicted meal event to achieve an increase in the insulin infusion rate to effectively pre-bolus a meal, and thereby reduce the likelihood of postprandial hyperglycemia. Additionally, or alternatively, the time constant or gain coefficient associated with one or more paths of the closed-loop control system 700 may be adjusted to tune the responsiveness to deviations between the measured glucose value and the target glucose value. For example, based on the particular type of meal being consumed or the particular time of day during which the meal is consumed, the time constant associated with the derivative block 710 or derivative term path may be adjusted to make the closed-loop control more or less aggressive in response to an increase in the patient's glucose level based on the patient's historical glycemic response to the particular type of meal.
In exemplary embodiments, the client device 806 is realized as a mobile phone, a smartphone, a tablet computer, or other similar mobile electronic device; however, in other embodiments, the client device 806 may be realized as any sort of electronic device capable of communicating with the medical device 802 via network 810, such as a laptop or notebook computer, a desktop computer, or the like. In exemplary embodiments, the network 810 is realized as a Bluetooth network, a ZigBee network, or another suitable personal area network. That said, in other embodiments, the network 810 could be realized as a wireless ad hoc network, a wireless local area network (WLAN), or local area network (LAN). The client device 806 includes or is coupled to a display device, such as a monitor, screen, or another conventional electronic display, capable of graphically presenting data and/or information pertaining to the physiological condition of the patient. The client device 806 also includes or is otherwise associated with a user input device, such as a keyboard, a mouse, a touchscreen, or the like, capable of receiving input data and/or other information from the user of the client device 806.
In exemplary embodiments, a user, such as the patient, the patient's doctor or another healthcare provider, or the like, manipulates the client device 806 to execute a client application 808 that supports communicating with the medical device 802 via the network 810. In this regard, the client application 808 supports establishing a communications session with the medical device 802 on the network 810 and receiving data and/or information from the medical device 802 via the communications session. The medical device 802 may similarly execute or otherwise implement a corresponding application or process that supports establishing the communications session with the client application 808. The client application 808 generally represents a software module or another feature that is generated or otherwise implemented by the client device 806 to support the processes described herein. Accordingly, the client device 806 generally includes a processing system and a data storage element (or memory) capable of storing programming instructions for execution by the processing system, that, when read and executed, cause processing system to create, generate, or otherwise facilitate the client application 808 and perform or otherwise support the processes, tasks, operations, and/or functions described herein. Depending on the embodiment, the processing system may be implemented using any suitable processing system and/or device, such as, for example, one or more processor devices, central processing units (CPUs), controllers, microprocessors, microcontrollers, processing cores and/or other hardware computing resources configured to support the operation of the processing system described herein. Similarly, the data storage element or memory may be realized as a random-access memory (RAM), read only memory (ROM), flash memory, magnetic or optical mass storage, or any other suitable non-transitory short or long-term data storage or other computer-readable media, and/or any suitable combination thereof.
In one or more embodiments, the client device 806 and the medical device 802 establish an association (or pairing) with one another over the network 810 to support subsequently establishing a point-to-point or peer-to-peer communications session between the medical device 802 and the client device 806 via the network 810. For example, in accordance with some embodiments, the network 810 is realized as a Bluetooth network, wherein the medical device 802 and the client device 806 are paired with one another (e.g., by obtaining and storing network identification information for one another) by performing a discovery procedure or another suitable pairing procedure. The pairing information obtained during the discovery procedure allows either of the medical device 802 or the client device 806 to initiate the establishment of a secure communications session via the network 810.
In one or more exemplary embodiments, the client application 808 is also configured to store or otherwise maintain an address and/or other identification information for the remote device 814 on the second network 812. In this regard, the second network 812 may be physically and/or logically distinct from the network 810, such as, for example, the Internet, a cellular network, a wide area network (WAN), or the like. The remote device 814 generally represents a server or other computing device configured to receive and analyze or otherwise monitor measurement data, event log data, and potentially other information obtained for the patient associated with the medical device 802. In exemplary embodiments, the remote device 814 is coupled to a database 816 configured to store or otherwise maintain data associated with individual patients. In practice, the remote device 814 may reside at a location that is physically distinct and/or separate from the medical device 802 and the client device 806, such as, for example, at a facility that is owned and/or operated by or otherwise affiliated with a manufacturer of the medical device 802. For purposes of explanation, but without limitation, the remote device 814 may alternatively be referred to herein as a server.
Still referring to
The medical device 802 generally represents the component of the patient monitoring system 800 that is communicatively coupled to the output of the sensing element 804 to receive or otherwise obtain the measurement data samples from the sensing element 804 (e.g., the measured glucose and characteristic impedance values), store or otherwise maintain the measurement data samples, and upload or otherwise transmit the measurement data to the remote device 814 or server via the client device 806. In one or more embodiments, the medical device 802 is realized as an infusion device 102, 200, 502 configured to deliver a fluid, such as insulin, to the body of the patient. That said, in other embodiments, the medical device 802 could be a standalone sensing or monitoring device separate and independent from an infusion device (e.g., sensing arrangement 104, 504). It should be noted that although
In exemplary embodiments, the medical device 802 includes a control module 822, a data storage element 824 (or memory), and a communications interface 826. The control module 822 generally represents the hardware, circuitry, logic, firmware and/or other component(s) of the medical device 802 that is coupled to the sensing element 804 to receive the electrical signals output by the sensing element 804 and perform or otherwise support various additional tasks, operations, functions and/or processes described herein. Depending on the embodiment, the control module 822 may be implemented or realized with a general purpose processor device, a microprocessor device, a controller, a microcontroller, a state machine, a content addressable memory, an application specific integrated circuit, a field programmable gate array, any suitable programmable logic device, discrete gate or transistor logic, discrete hardware components, or any combination thereof, designed to perform the functions described herein. In some embodiments, the control module 822 includes an analog-to-digital converter (ADC) or another similar sampling arrangement that samples or otherwise converts an output electrical signal received from the sensing element 804 into corresponding digital measurement data value. In other embodiments, the sensing element 804 may incorporate an ADC and output a digital measurement value.
The communications interface 826 generally represents the hardware, circuitry, logic, firmware and/or other components of the medical device 802 that are coupled to the control module 822 for outputting data and/or information from/to the medical device 802 to/from the client device 806. For example, the communications interface 826 may include or otherwise be coupled to one or more transceiver modules capable of supporting wireless communications between the medical device 802 and the client device 806. In exemplary embodiments, the communications interface 826 is realized as a Bluetooth transceiver or adapter configured to support Bluetooth Low Energy (BLE) communications.
In exemplary embodiments, the remote device 814 receives, from the client device 806, measurement data values associated with a particular patient (e.g., sensor glucose measurements, acceleration measurements, and the like) that were obtained using the sensing element 804, and the remote device 814 stores or otherwise maintains the historical measurement data in the database 816 in association with the patient (e.g., using one or more unique patient identifiers). Additionally, the remote device 814 may also receive, from or via the client device 806, meal data or other event log data that may be input or otherwise provided by the patient (e.g., via client application 808) and store or otherwise maintain historical meal data and other historical event or activity data associated with the patient in the database 816. In this regard, the meal data include, for example, a time or timestamp associated with a particular meal event, a meal type or other information indicative of the content or nutritional characteristics of the meal, and an indication of the size associated with the meal. In exemplary embodiments, the remote device 814 also receives historical fluid delivery data corresponding to basal or bolus dosages of fluid delivered to the patient by an infusion device 102, 200, 502. For example, the client application 808 may communicate with an infusion device 102, 200, 502 to obtain insulin delivery dosage amounts and corresponding timestamps from the infusion device 102, 200, 502, and then upload the insulin delivery data to the remote device 814 for storage in association with the particular patient. The remote device 814 may also receive geolocation data and potentially other contextual data associated with a device 802, 806 from the client device 806 and/or client application 808, and store or otherwise maintain the historical operational context data in association with the particular patient. In this regard, one or more of the devices 802, 806 may include a global positioning system (GPS) receiver or similar modules, components or circuitry capable of outputting or otherwise providing data characterizing the geographic location of the respective device 802, 806 in real-time.
The historical patient data may be analyzed by one or more of the remote device 814, the client device 806, and/or the medical device 802 to alter or adjust operation of an infusion device 102, 200, 502 to influence fluid delivery in a personalized manner. For example, the patient's historical meal data and corresponding measurement data or other contextual data may be analyzed to predict a future time when the next meal is likely to be consumed by the patient, the likelihood of a future meal event within a specific time period, the likely size or amount of carbohydrates associated with a future meal, the likely type or nutritional content of the future meal, and/or the like. Moreover, the patient's historical measurement data for postprandial periods following historical meal events may be analyzed to model or otherwise characterize the patient's glycemic response to the predicted size and type of meal for the current context (e.g., time of day, day of week, geolocation, etc.). One or more aspects of the infusion device 102, 200, 502 that control or regulate insulin delivery may then be modified or adjusted to proactively account for the patient's likely meal activity and glycemic response.
In one or more exemplary embodiments, the remote device 814 utilizes machine learning to determine which combination of historical sensor glucose measurement data, historical delivery data, historical auxiliary measurement data (e.g., historical acceleration measurement data, historical heart rate measurement data, and/or the like), historical event log data, historical geolocation data, and other historical or contextual data are correlated to or predictive of the occurrence of a particular event, activity, or metric for a particular patient, and then determines a corresponding equation, function, or model for calculating the value of the parameter of interest based on that set of input variables. Thus, the model is capable of characterizing or mapping a particular combination of one or more of the current (or recent) sensor glucose measurement data, auxiliary measurement data, delivery data, geographic location, patient behavior or activities, and the like to a value representative of the current probability or likelihood of a particular event or activity or a current value for a parameter of interest. It should be noted that since each patient's physiological response may vary from the rest of the population, the subset of input variables that are predictive of or correlative for a particular patient may vary from other patients. Additionally, the relative weightings applied to the respective variables of that predictive subset may also vary from other patients who may have common predictive subsets, based on differing correlations between a particular input variable and the historical data for that particular patient. It should be noted that any number of different machine learning techniques may be utilized by the remote device 814 to determine what input variables are predictive for a current patient of interest, such as, for example, artificial neural networks, genetic programming, support vector machines, Bayesian networks, probabilistic machine learning models, or other Bayesian techniques, fuzzy logic, heuristically derived combinations, or the like.
The insulin infusion device may incorporate or leverage the control algorithms, processing schemes, and operating methodologies (or suitably modified, updated, or customized versions thereof) of the type described in U.S. Pat. No. 9,526,834 and International (PCT) patent publication number WO 2014/035570; the content of these published documents is incorporated herein by reference.
Each of the devices or components that are described above with reference to
The illustrated embodiment of the device 900 is intended to be a high-level and generic representation of one suitable platform. In this regard, any of the computer-based or processor-based components of the system 100 can utilize the architecture of the device 900. The illustrated embodiment of the device 900 generally includes, without limitation: at least one processor 902; a suitable amount of memory 904; device-specific hardware, software, firmware, and/or features 906; a user interface 908; a communication module 910; and a display element 912. Of course, an implementation of the device 900 may include additional elements, components, modules, and functionality configured to support various features that are unrelated to the subject matter described here. For example, the device 900 may include certain features and elements to support conventional functions that might be related to the particular implementation and deployment of the device 900. In practice, the elements of the device 900 may be coupled together via a bus or any suitable interconnection architecture 914.
The processor 902 may be implemented or performed with a general purpose processor, a content addressable memory, a digital signal processor, an application specific integrated circuit, a field programmable gate array, any suitable programmable logic device, discrete gate or transistor logic, discrete hardware components, or any combination designed to perform the functions described here. Moreover, the processor 902 may be implemented as a combination of computing devices, e.g., a combination of a digital signal processor and a microprocessor, a plurality of microprocessors, one or more microprocessors in conjunction with a digital signal processor core, or any other such configuration.
The memory 904 may be realized as RAM memory, flash memory, EPROM memory, EEPROM memory, registers, a hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art. In this regard, the memory 904 can be coupled to the processor 902 such that the processor 902 can read information from, and write information to, the memory 904. In the alternative, the memory 904 may be integral to the processor 902. As an example, the processor 902 and the memory 904 may reside in an ASIC. At least a portion of the memory 904 can be realized as a computer storage medium, e.g., a tangible computer-readable medium having computer-executable instructions stored thereon. The computer-executable instructions, when read and executed by the processor 902, cause the device 900 to perform certain tasks, operations, functions, and processes that are specific to the particular embodiment. In this regard, the memory 904 may represent one suitable implementation of such computer-readable media. Alternatively, or additionally, the device 900 could receive and cooperate with computer-readable media (not separately shown) that is realized as a portable or mobile component or platform, e.g., a portable hard drive, a USB flash drive, an optical disc, or the like.
The device-specific hardware, software, firmware, and features 906 may vary from some embodiments of the device 900 to another. For example, the device-specific hardware, software, firmware, and features 906 will support: smartphone functions and features when the device 900 is realized as a mobile telephone; conventional personal computer functions and features if the device 900 is realized as a laptop or tablet computer; insulin pump operations when the device 900 is realized as an insulin infusion device; etc. In practice, certain portions or aspects of the device-specific hardware, software, firmware, and features 906 may be implemented in one or more of the other blocks depicted in
The user interface 908 may include or cooperate with various features to allow a user to interact with the device 900. Accordingly, the user interface 908 may include various human-to-machine interfaces, e.g., a keypad, keys, a keyboard, buttons, switches, knobs, a touchpad, a joystick, a pointing device, a virtual writing tablet, a touch screen, a microphone, or any device, component, or function that enables the user to select options, input information, or otherwise control the operation of the device 900. The user interface 908 may include one or more graphical user interface (GUI) control elements that enable a user to manipulate or otherwise interact with an application via the display element 912.
The communication module 910 facilitates data communication between the device 900 and other components as needed during the operation of the device 900. In the context of this description, the communication module 910 can be employed to transmit or stream device-related control data, patient-related data, device-related status or operational data, glycemic insight messages and notifications, and the like. It should be appreciated that the particular configuration and functionality of the communication module 910 can vary depending on the hardware platform and specific implementation of the device 900. In practice, an embodiment of the device 900 may support wireless data communication and/or wired data communication, using various data communication protocols. For example, the communication module 910 could support one or more wireless data communication protocols, techniques, or methodologies, including, without limitation: RF; IrDA (infrared); Bluetooth; ZigBee (and other variants of the IEEE 802.15 protocol); IEEE 802.11 (any variation); IEEE 802.16 (WiMAX or any other variation); Direct Sequence Spread Spectrum; Frequency Hopping Spread Spectrum; cellular/wireless/cordless telecommunication protocols; wireless home network communication protocols; paging network protocols; magnetic induction; satellite data communication protocols; wireless hospital or health care facility network protocols such as those operating in the WMTS bands; GPRS; and proprietary wireless data communication protocols such as variants of Wireless USB. Moreover, the communication module 910 could support one or more wired/cabled data communication protocols, including, without limitation: Ethernet; powerline; home network communication protocols; USB; IEEE 1394 (Firewire); hospital network communication protocols; and proprietary data communication protocols.
The display element 912 is suitably configured to enable the device 900 to render and display various screens, insight messages, notifications, GUIs, GUI control elements, drop down menus, auto-fill fields, text entry fields, message fields, or the like. Of course, the display element 912 may also be utilized for the display of other information during the operation of the device 900, as is well understood. Notably, the specific configuration, operating characteristics, size, resolution, and functionality of the display element 910 can vary depending upon the practical implementation of the device 900. For example, if the device 900 is a laptop computer, then the display element 912 may be a relatively large monitor. Alternatively, if the device 900 is a cellular telephone device, then the display element 912 may be a relatively small integrated display screen, such as a touch-sensitive screen.
In one nonlimiting embodiment, the inputs include discrete glucose measurements 1010 from a blood glucose meter (BGM), contextual activity data 1020 from an activity tracker or equivalent source of activity data such as a smartphone or other health related device, and optionally contextual data 1030 from other sources. The contextual activity data 1020 collected from the activity tracker or equivalent source of activity data (e.g., smartphone or other health related device) can include, for example, one or more of: metabolic data about the patient (e.g., calories burned); physical activity data about the user (e.g., heart rate data for the patient, exercise data for the patient, activity type, duration and intensity data, data about a number of steps taken by the patient during a time period, metabolic equivalents and other parameters); data from a fitness or exercise advisor device or application that provides fitness/exercise suggestions or recommendations to users; and other raw contextual data such as accelerometer (x,y,z) data, geolocation data, iBeacon location data, skin temperature data, ambient air temperature data, bioimpedance heart rate data, sweat data (e.g., GSR-conductance), blood pressure data, shake detection data, pedometer data, barometer data, gyroscope data, meal log data, orientation (azimuth, pitch, roll (degrees)) data; health kit data (such as data from Apple® Healthkit, Google® Fit, etc.), medication/prescription information data, user gestures data, uv light detector data, magnetometer data, respiration data, muscle activity data, pulse oximeter data, blood+interstitial fluid pH data, metabolic equivalents (METS) data, sleep quality/time data, EMR and lab data, etc. The contextual data 1030 from other sources can include, for example, historical sensor glucose measurement data, historical delivery data, historical auxiliary measurement data such as historical acceleration measurement data, historical heart rate measurement data, skin conductance, and/or the like, historical event log data, historical geolocation data, and other historical or contextual data are correlated to or predictive of the occurrence of a particular event, activity, or metric for a particular patient. The estimation model 1040 processes at least some of the various inputs to continuously generate estimated glucose values 1080.
In one or more exemplary embodiments, the remote device 814 of
Artificial intelligence is an area of computer science emphasizes the creation of intelligent machines that work and react like humans. Some of the activities computers with artificial intelligence are designed for include learning. Examples of artificial intelligence algorithms include, but are not limited to, key learning, actor critic methods, reinforce, deep deterministic policy gradient (DDPG), multi-agent deep deterministic policy gradient (MADDPG), etc. Machine learning refers to an artificial intelligence discipline geared toward the technological development of human knowledge.
Machine learning facilitates a continuous advancement of computing through exposure to new scenarios, testing and adaptation, while employing pattern and trend detection for improved decisions and subsequent, though not identical, situations. Machine learning (ML) algorithms and statistical models can be used by computer systems to effectively perform a specific task without using explicit instructions, relying on patterns and inference instead. Machine learning algorithms build a mathematical model based on sample data, known as “training data,” in order to make predictions or decisions without being explicitly programmed to perform the task. Machine learning algorithms can be used when it is infeasible to develop an algorithm of specific instructions for performing the task.
For example, supervised learning algorithms build a mathematical model of a set of data that contains both the inputs and the desired outputs. The data is known as training data and consists of a set of training examples. Each training example has one or more inputs and a desired output, also known as a supervisory signal. In the case of semi-supervised learning algorithms, some of the training examples are missing the desired output. In the mathematical model, each training example is represented by an array or vector, and the training data by a matrix. Through iterative optimization of an objective function, supervised learning algorithms learn a function that can be used to predict the output associated with new inputs. An optimal function will allow the algorithm to correctly determine the output for inputs that were not a part of the training data. An algorithm that improves the accuracy of its outputs or predictions over time is said to have learned to perform that task. Supervised learning algorithms include classification and regression. Classification algorithms are used when the outputs are restricted to a limited set of values, and regression algorithms are used when the outputs may have any numerical value within a range. Similarity learning is an area of supervised machine learning closely related to regression and classification, but the goal is to learn from examples using a similarity function that measures how similar or related two objects are.
Reinforcement learning is an area of machine learning concerned with how software agents ought to take actions in an environment so as to maximize some notion of cumulative reward. Due to its generality, the field is studied in many other disciplines, such as game theory, control theory, operations research, information theory, simulation-based optimization, multi-agent systems, swarm intelligence, statistics and genetic algorithms. In machine learning, the environment is typically represented as a Markov Decision Process (MDP). Many reinforcement learning algorithms use dynamic programming techniques. Reinforcement learning algorithms do not assume knowledge of an exact mathematical model of the MDP and are used when exact models are infeasible.
In accordance with one non-limiting implementation, the estimation model 1040 can be an ensemble model 1050. An ensemble model 1050 includes two or more related, but different analytical models that are run or executed at the same time, where the results generated by each can then be synthesized into a single score or spread in order to improve the accuracy of predictive analytics. To explain further, in predictive modeling and other types of data analytics, a single model based on one data sample can have biases, high variability or outright inaccuracies that can affect the reliability of its analytical findings. By combining different models or analyzing multiple samples, the effects of those limitations can be reduced to provide better information. As such, ensemble methods can use multiple machine learning algorithms to obtain better predictive performance than could be obtained from any of the constituent learning algorithms alone.
An ensemble is a supervised learning algorithm because it can be trained and then used to make predictions. The trained ensemble, therefore, represents a single hypothesis that is not necessarily contained within the hypothesis space of the models from which it is built. Thus, ensembles can be shown to have more flexibility in the functions they can represent. An ensemble model can include a set of individually trained classifiers (such as neural networks or decision trees) whose predictions are combined.
For instance, one common example of ensemble modeling is a random forest model which is a type of analytical model that leverages multiple decision trees and is designed to predict outcomes based on different variables and rules. A random forest model blends decision trees that may analyze different sample data, evaluate different factors or weight common variables differently. The results of the various decision trees are then either converted into a simple average or aggregated through further weighting. The emergence of Hadoop and other big data technologies has allowed greater volumes of data to be stored and analyzed, which can allow analytical models to be run on different data samples.
In some embodiments, the ensemble model 1050 can include or be built using at least one machine learning model 1060 and a physiological model 1070. The physiological model 1070 can be, for example, a population-based model of one or more equations (e.g., a set of differential equations) that are generally applicable to any patient, but has different variables or parameters that are weighted in accordance with a particular patient's individual parameters to mimic the physiology of that patient. In other words, because each patient's physiological response may vary from the rest of the population, the relative weightings applied to the respective variables of the physiological model 1070 may also vary from other patients. The physiological model 1070 can be used in conjunction with the machine learning model(s) 1060 to help confine estimations or predictions within a realistic range.
Depending on the implementation, any number of machine learning models can be combined to optimize the ensemble model 1050. Examples of machine learning algorithms or models that can be implemented at the machine learning model 1060 can include, but are not limited to: regression models such as linear regression, logistic regression, and K-means clustering; one or more decision tree models (e.g., a random forest model); one or more support vector machines; one or more artificial neural networks; one or more deep learning networks (e.g., at least one recurrent neural network, sequence to sequence mapping using deep learning, sequence encoding using deep learning, etc.); fuzzy logic based models; genetic programming models; Bayesian networks or other Bayesian techniques, probabilistic machine learning models; Gaussian processing models; Hidden Markov models; time series methods such as Autoregressive Moving Average (ARMA) models, Autoregressive Integrated Moving Average (ARIMA) models, Autoregressive conditional heteroskedasticity (ARCH) models; generalized autoregressive conditional heteroskedasticity (GARCH) models; moving-average (MA) models or other models; and heuristically derived combinations of any of the above, etc. The types of machine learning algorithms differ in their approach, the type of data they input and output, and the type of task or problem that they are intended to solve.
A Hidden Markov model (HMM) is a statistical Markov model in which the system being modeled is assumed to be a Markov process with unobserved (hidden) states. An MINI can be considered as the simplest dynamic Bayesian network. A Bayesian network, belief network or directed acyclic graphical model is a probabilistic graphical model that represents a set of random variables and their conditional independence with a directed acyclic graph (DAG). Bayesian networks that model sequences of variables are called dynamic Bayesian networks. Generalizations of Bayesian networks that can represent and solve decision problems under uncertainty are called influence diagrams.
Support vector machines (SVMs), also known as support vector networks, are a set of related supervised learning methods used for classification and regression. Given a set of training examples, each marked as belonging to one of two categories, an SVM training algorithm builds a model that predicts whether a new example falls into one category or the other. An SVM training algorithm is a non-probabilistic, binary, linear classifier. In addition to performing linear classification, SVMs can efficiently perform a non-linear classification using what is called the kernel trick, implicitly mapping their inputs into high-dimensional feature spaces.
Decision tree learning uses a decision tree as a predictive model to go from observations about an item (represented in the branches) to conclusions about the item's target value (represented in the leaves). Tree models where the target variable can take a discrete set of values are called classification trees; in these tree structures, leaves represent class labels and branches represent conjunctions of features that lead to those class labels. Decision trees where the target variable can take continuous values (typically real numbers) are called regression trees. In decision analysis, a decision tree can be used to visually and explicitly represent decisions and decision making.
Deep learning algorithms can refer to a collection of algorithms used in machine learning, that are used to model high-level abstractions and data through the use of model architectures, which are composed of multiple nonlinear transformations. Deep learning is a specific approach used for building and training neural networks. Deep learning consists of multiple hidden layers in an artificial neural network. Examples of deep learning algorithms can include, for example, Siamese networks, transfer learning, recurrent neural networks (RNNs), long short term memory (LSTM) networks, convolutional neural networks (CNNs), transformers, etc. For instance, deep learning approaches can make use of autoregressive Recurrent Neural Networks (RNN), such as the long short-term memory (LSTM) and the Gated Recurrent Unit (GRU). One neural network architecture for time series forecasting using RNNs (and variants) is an autoregressive seq2seq neural network architecture, which acts as an autoencoder.
In some embodiments, the ensemble model 1050 can include one or more deep learning algorithms. It should be noted that any number of different machine learning techniques may also be utilized. Depending on the implementation, the ensemble model 1050 can be implemented as a bootstrap aggregating ensemble algorithm (also referred to as a bagging classifier method), as a boosting ensemble algorithm or classifier algorithm, as a stacking ensemble algorithm or classifier algorithm, as bucket of models ensemble algorithms, as Bayes optimal classifier algorithms, as Bayesian parameter averaging algorithms, as Bayesian model combination algorithms, etc.
Bootstrap aggregating, often abbreviated as bagging, involves having each model in the ensemble vote with equal weight. In order to promote model variance, bagging trains each model in the ensemble using a randomly drawn subset of the training set. As an example, the random forest algorithm combines random decision trees with bagging to achieve very high classification accuracy. A bagging classifier or ensemble method creates individuals for its ensemble by training each classifier on a random redistribution of the training set. Each classifier's training set can be generated by randomly drawing, with replacement, N examples—where N is the size of the original training set; many of the original examples may be repeated in the resulting training set while others may be left out. Each individual classifier in the ensemble is generated with a different random sampling of the training set. Bagging is effective on “unstable” learning algorithms (e.g., neural networks and decision trees), where small changes in the training set result in large changes in predictions.
By contrast, boosting involves incrementally building an ensemble by training each new model instance to emphasize the training instances that previous models mis-classified. In some cases, boosting has been shown to yield better accuracy than bagging, but it also tends to be more likely to over-fit the training data. A boosting classifier can refer to a family of methods that can be used to produce a series of classifiers. The training set used for each member of the series is chosen based on the performance of the earlier classifier(s) in the series. In boosting, examples that are incorrectly predicted by previous classifiers in the series are chosen more often than examples that were correctly predicted. Thus, boosting attempts to produce new classifiers that are better able to predict examples for which the current ensemble's performance is poor. A common implementation of boosting is Adaboost, although some newer algorithms are reported to achieve better results.
Stacking (sometimes called stacked generalization) involves training a learning algorithm to combine the predictions of several other learning algorithms. Stacking works in two phases: multiple base classifiers are used to predict the class, and then a new learner is used to combine their predictions with the aim of reducing the generalization error. First, all of the other algorithms are trained using the available data, then a combiner algorithm is trained to make a final prediction using all the predictions of the other algorithms as additional inputs. If an arbitrary combiner algorithm is used, then stacking can theoretically represent any of the ensemble techniques described in this article, although, in practice, a logistic regression model is often used as the combiner.
A “bucket of models” is an ensemble technique in which a model selection algorithm is used to choose the best model for each problem. When tested with only one problem, a bucket of models can produce no better results than the best model in the set, but when evaluated across many problems, it will typically produce much better results, on average, than any model in the set. One common approach used for model-selection is cross-validation selection (sometimes called a “bake-off contest”). Cross-validation selection can be summed up as try them all with the training set and pick the one that works best. Gating is a generalization of Cross-Validation Selection. It involves training another learning model to decide which of the models in the bucket is best-suited to solve the problem. Often, a perceptron is used for the gating model. It can be used to pick the “best” model, or it can be used to give a linear weight to the predictions from each model in the bucket. When a bucket of models is used with a large set of problems, it may be desirable to avoid training some of the models that take a long time to train. Landmark learning is a meta-learning approach that seeks to solve this problem. It involves training only the fast (but imprecise) algorithms in the bucket, and then using the performance of these algorithms to help determine which slow (but accurate) algorithm is most likely to do best.
The Bayes optimal classifier is a classification technique. It is an ensemble of all the hypotheses in the hypothesis space. On average, no other ensemble can outperform it. The naive Bayes optimal classifier is a version of this that assumes that the data is conditionally independent on the class and makes the computation more feasible. Each hypothesis is given a vote proportional to the likelihood that the training dataset would be sampled from a system if that hypothesis were true. To facilitate training data of finite size, the vote of each hypothesis is also multiplied by the prior probability of that hypothesis. The hypothesis represented by the Bayes optimal classifier, however, is the optimal hypothesis in ensemble space (the space of all possible ensembles.
Bayesian parameter averaging (BPA) is an ensemble technique that seeks to approximate the Bayes optimal classifier by sampling hypotheses from the hypothesis space and combining them using Bayes' law. Unlike the Bayes optimal classifier, Bayesian model averaging (BMA) can be practically implemented. Hypotheses are typically sampled using a Monte Carlo sampling technique such as MCMC. For example, Gibbs sampling may be used to draw hypotheses that are representative of a distribution. It has been shown that under certain circumstances, when hypotheses are drawn in this manner and averaged according to Bayes' law, this technique has an expected error that is bounded to be at most twice the expected error of the Bayes optimal classifier.
Bayesian model combination (BMC) is an algorithmic correction to Bayesian model averaging (BMA). Instead of sampling each model in the ensemble individually, it samples from the space of possible ensembles (with model weightings drawn randomly from a Dirichlet distribution having uniform parameters). This modification overcomes the tendency of BMA to converge toward giving all of the weight to a single model. Although BMC is somewhat more computationally expensive than BMA, it tends to yield dramatically better results. The results from BMC have been shown to be better on average (with statistical significance) than BMA, and bagging. The use of Bayes' law to compute model weights necessitates computing the probability of the data given each model. Typically, none of the models in the ensemble are exactly the distribution from which the training data were generated, so all of them correctly receive a value close to zero for this term. This would work well if the ensemble were big enough to sample the entire model-space, but such is rarely possible. Consequently, each pattern in the training data will cause the ensemble weight to shift toward the model in the ensemble that is closest to the distribution of the training data. It essentially reduces to an unnecessarily complex method for doing model selection. The possible weightings for an ensemble can be visualized as lying on a simplex. At each vertex of the simplex, all of the weight is given to a single model in the ensemble. BMA converges toward the vertex that is closest to the distribution of the training data. By contrast, BMC converges toward the point where this distribution projects onto the simplex. In other words, instead of selecting the one model that is closest to the generating distribution, it seeks the combination of models that is closest to the generating distribution. The results from BMA can often be approximated by using cross-validation to select the best model from a bucket of models. Likewise, the results from BMC may be approximated by using cross-validation to select the best ensemble combination from a random sampling of possible weightings.
Once all the generic estimation models have been validated, the method 1100 proceeds to 1140, where the computing system determines, based on characteristics of a particular user or patient, an applicable generic estimation model that is the best fit for that particular patient based on characteristics of that particular patient. The particular generic estimation model that is applicable for a particular patient varies based on the characteristics of that particular patient.
At 1150, the computing system can then update the particular generic estimation model that was deployed at 1140 based on data that was collected for that particular patient to generate an estimation model for that particular patient. In other words, parameters of the generic estimation model that was selected and deployed 1140 can be updated and fine-tuned at 150 based on data that was collected for the particular patient in order to customize that generic estimation model such that it becomes an estimation model for that particular patient.
After the estimation model for that particular patient has been generated at 1150 it can be executed and updated any time new data is received for that patient. For example, in some embodiments, the new data can come from the blood glucose meter data (e.g., new blood glucose measurements), new contextual activity data that is received for that patient from the activity tracker, and/or new contextual data that is received for that patient from other sources. In one implementation, the estimation model for that particular patient can be updated each time there is a new data from the BGM. In another embodiment, the new data can be from a continuous glucose monitor. For instance, the CGM can be used periodically to collect CGM data for the particular patient for a certain period of time (e.g., for two weeks), and this newly collected CGM data can then be used to assess performance of the estimation model and/or update or re-train the estimation model for that particular patient. In addition, the CGM data can also be used to train and update the generic estimation models that were described above with reference to step 1110. Furthermore, in some implementations, the CGM data can also be used at 1140 to select and deploy a different one of the generic estimation models for the particular patient.
As such, at 1160, the computing system regularly determines whether new data is received, and if so, the method proceeds to 1170 where the estimation model for that particular patient is updated based on the new data to generate an updated estimation model that can be executed at a processing system (as will be described below).
After the estimation model for that particular patient is updated based on the new data at 1170, the method loops to 1160. As the estimation model for the particular patient is updated it becomes more fine-tuned and can be used to generate better estimated glucose values for the patient. In any point the estimation model can be used to estimate glucose values for the patient. One non-limiting embodiment will now be described with respect to
At 1240, a processing system can execute the machine learning-based estimation model for that patient to process the various inputs from steps 1210, 1220, 1230 to generate a sequence or set of estimated glucose values for that particular patient. The estimated glucose values are analogous to glucose measurements that can be taken by a glucose sensor in that the estimated glucose values estimate continuous glucose sensor measurement values. However, the processing system can generate the set of estimated glucose values without using information from a continuous glucose sensor arrangement. This can eliminate the need for a continuous glucose sensor, which can be a relatively expensive component with respect to a blood glucose meter, and thus saves patients a significant cost. It may also support intermittent CGM users that take “sensor vacations” due to various reasons which may include access, discomfort, seasonal needs, etc. In some embodiments, only the first set of inputs and the second set of user inputs are processed via the estimation model to generate the set of estimated glucose values. As used herein, “estimated glucose value,” and “estimated glucose concentration,” and the like may refer to a glucose level that has been estimated via the machine learning-based estimation model. In some embodiments, the estimated glucose values estimate and track interstitial or sensor glucose values (e.g., that are generated by a sensing arrangement such as a CGM device that has been calibrated using measured blood glucose values measured by a blood glucose meter).
Steps 1250-1280 are optional and relate to when and how the estimated glucose values can be used within the context of an insulin management system for diabetes management. For example, in one implementation, at 1250, the computing system can monitor the set or sequence of estimated glucose values (e.g., discrete estimates of blood glucose levels) to determine if the sequence of the estimated glucose values are within an acceptable accuracy range that allows them to be used by an insulin management system. How the computing system can determine if the sequence of the estimated glucose values are within an acceptable accuracy range can vary depending on the implementation. For example, in some embodiments, the computing system can determine if the sequence of the estimated glucose values are within an acceptable accuracy range by measuring a difference between estimated glucose value and intermittent glucose measurements from a blood glucose meter (or equivalent device) and then comparing that difference to an accuracy threshold. In another embodiment, the computing system can determine if the sequence of the estimated glucose values are within an acceptable accuracy range by measuring difference between estimated glucose value and periodic measurements that were taken by a CGM device and then comparing that difference to an accuracy threshold.
When the estimated glucose values are determined (at 1250) to not be within an acceptable accuracy range that allows them to be used by an insulin management system, the method 1200 proceeds to 1260, where a notification or an alert can be generated that can be perceptible to the patient or another user who is monitoring the patient. For example, in some embodiments, when the estimated glucose values are not in acceptable accuracy ranges, a notification or alert can be generated to inform the user that they should not trust the estimated glucose values being generated by the estimation model and to recommend that the user to wear another sensor (e.g., CGM) to retrain the estimation model. Any number of alerting methodologies can be employed at step 1260, including, but not limited to, presenting (via a display device) graphical elements and text associated with notifications and alerts for the glucose control system (see reference 100,
When the estimated glucose values are determined (at 1250) to be within an acceptable accuracy range that allows them to be used by an insulin management system, the method 1200 proceeds to 1280, where the estimated glucose values can be utilized the insulin delivery system, and estimates can continue to be generated in near real-time. The actions that can be taken at the insulin delivery system based on the estimated glucose values vary depending on the implementation. For example, actions that can be taken at the insulin delivery system based on the estimated glucose values can include controlling an insulin infusion device based on the estimated glucose values so that it administers insulin to a patient, or generating a notification or an alert, based on the estimated glucose values, that can be perceptible to the patient or another user who is monitoring the patient. As will be described below, the corrective actions and the notifications or alerts that can be generated vary depending on the implementation.
For example, in one implementation of step 1280, the computing system operates cooperatively with an insulin delivery system and the machine learning-based estimation model to monitor estimated glucose levels of a patient, and to identify abnormalities in the estimated glucose levels. For example, the computing system can monitor the set or sequence of estimated glucose values (e.g., discrete estimates of blood glucose levels) to determine if the sequence of the estimated glucose values are within a low range, a high range, or an acceptable range. For instance, hypoglycemia is a deficiency of glucose in the bloodstream of the patient or user of the insulin delivery pump. Hypoglycemia may be generally associated with glucose values below 70 milligrams per deciliter (mg/dL), depending on certain patient-specific factors. Hyperglycemia is an excess of glucose in the bloodstream of the patient or user of the insulin delivery pump and may be generally associated with glucose values above 180 mg/dL, depending on certain patient-specific factors. Normal, healthy blood glucose levels, over every 24-hour cycle, generally range between 70 and 160 mg/dL. An abnormality in a sequence of blood glucose levels may include blood glucose levels below a predefined hypoglycemia threshold, blood glucose levels above a predefined hyperglycemia threshold, a decreasing blood glucose trend approaching or below a predefined hypoglycemia threshold, an increasing blood glucose trend approaching or above a predefined hyperglycemia threshold, or the like.
The abnormality indicates that a corrective action is required to maintain blood glucose stability at appropriate levels. Abnormalities in blood glucose levels require intervention to bring the patient's blood glucose levels into a safe and healthy range. As such, when the estimated glucose values are not within an acceptable range, the method 1200 can proceed to 1280, where a corrective action can be taken at the insulin infusion device to deliver the patient an amount of insulin required to adequately treat the blood glucose abnormality to bring the blood glucose levels of the patient within normal limits and/or regulate their blood sugar levels. For example, in some embodiments, the estimated glucose values can be used to control and drive an insulin infusion device so that it administers insulin to a patient.
Exemplary embodiments of corrective actions are disclosed in, for example, in U.S. patent application Ser. No. 16/129,552, filed Sep. 12, 2018, entitled “SYSTEMS AND METHODS FOR ALERTING A USER TO DETECTED EFFECTS OF A RECENT THERAPY ADJUSTMENT,” and assigned to the assignee of the present invention, which is incorporated herein by reference in its entirety. Any number of corrective actions or intervention methodologies can be employed at step 1280, such as, administration of a bolus of insulin (to decrease blood glucose levels) or administration of a quantity of carbohydrates, glucose, or other medication (to increase blood glucose levels). In other words, the corrective action may include the administration of a bolus of insulin (to decrease hyperglycemic blood glucose levels) or the administration of a dose of carbohydrates (to increase hypoglycemic blood glucose levels).
In addition, in some embodiments, at step 1280, the computing system operates cooperatively with an insulin delivery system (e.g., insulin pump or manual insulin injection device such as an insulin pen) and the machine learning-based estimation model to monitor estimated glucose levels of a patient, and to generate a notification or an alert, based on the estimated glucose values, that can be perceptible to the patient or another user who is monitoring the patient (e.g., to inform the patient or the user of any abnormalities that are detected when the estimated glucose values are within the low range that is too low or within the high range that is too high). For instance, a notification or alert can be sent to the user to alert them that estimated glucose values are abnormal and help drive decision support for medication administration. Exemplary embodiments of alerts are disclosed, for example, in U.S. patent application Ser. No. 16/129,552, filed Sep. 12, 2018, entitled “SYSTEMS AND METHODS FOR ALERTING A USER TO DETECTED EFFECTS OF A RECENT THERAPY ADJUSTMENT,” and assigned to the assignee of the present invention, which is incorporated herein by reference in its entirety. Any number of alerting methodologies can be employed at step 1280, including, but not limited to, presenting (via a display device) graphical elements and text associated with notifications and alerts for the glucose control system (see reference 100,
For example, a presentation module can identify appropriate graphical elements for presentation via a display device that identify a current estimated blood glucose level (e.g., various icons, text, and/or graphical elements associated with estimated blood glucose levels of a patient), a detected indication of a change of direction in the sequence of blood glucose levels, timing data associated with a detected indication of a change of direction in the sequence of blood glucose levels, an indication that a corrective action should be taken by a user, confirmation of a user-input indication that a corrective action has been completed by a user, and the like.
For example, a display device can display, render, or otherwise convey one or more graphical representations or images associated with blood glucose levels, an indication of a change of direction in a blood glucose trend, and/or notifications or alerts associated with blood glucose levels on the display device. In some embodiments, an alert can notify the patient of an over-corrected or under-corrected abnormality. For instance, in one implementation, an alert can be presented via the insulin delivery pump or a manual insulin injection device (e.g., insulin pen), such that the patient is informed that an administered corrective action has not affected the blood glucose levels of the patient in the desired manner (e.g., that the corrective action has over-corrected or under-corrected the abnormality).
In some embodiments, an alert can be presented to prompt a user to perform a corrective action, in response to a detected abnormality (e.g., blood glucose levels that are too high or too low). For example, when the abnormality includes blood glucose levels that are too high, the user (patient or health care partner) can be prompted to input a command for the insulin delivery pump to administer a bolus of insulin, via the insulin deliver pump, to treat the high blood glucose levels. Alternatively, a notification can provide health care providers with information to adjust basal rate settings of an inulin infusion device so that they are appropriate for a particular patient to meet their needs. In another embodiment, when the abnormality includes blood glucose levels that are too high, the user (patient or health care partner) can be prompted to input a command at a manual insulin injection device (e.g., insulin pen) to administer insulin. By contrast, when the abnormality includes blood glucose levels that are too low, the alert can prompt the user to administer a dose of carbohydrates to treat the low blood glucose levels.
Additionally, the estimated glucose values can be utilized by other downstream systems, such as a fitness or exercise advisor device or application that provides fitness/exercise suggestions or recommendations to users. For instance, the estimated glucose values can be utilized in conjunction with a predicted exercise level in the short future (using similar exercise pattern from the past few minutes) to tell patient if keeping current level of activity will reduce hyperglycemia or introduce hypoglycemia.
After 1280, the method 1280 can loop to the start, and then proceeds to steps 1210, 1220 and 1230 to wait and receive new data regarding blood glucose measurements from a blood glucose meter, contextual activity data from an activity tracker and/or other contextual data from other sources and the model would be retrained.
At 1330, at least one of the population models can be selected, and at 1340, personalized model learning processes can be performed to personalize a selected population model for each particular user. In one non-limiting embodiment, once the population model is developed, the generic, estimation model can be adapted or trained, based on learned characteristics of a particular user (e.g., patient), to generate a personalized estimation model that is optimized for estimating blood glucose of that particular user. The personalized model learning processes 1340 can be performed for any number of users to personalize a population model 1330 selected for each user to generate a personalized model 1350 that is personalized for each user. Nonlimiting examples of the various processes that can be performed as part of 1330 will be described below with reference to
Once the personalized model is generated it can be used for a variety of purposes including to provide insulin therapy to a particular user without the need for a glucose sensor such as a continuous glucose monitor.
Various embodiments of population model learning processes that can be performed will now be described below with reference to
During this part of the workflow process data for multiple users is collected for processing. In this example, the data that is collected can include, but is not limited to, discrete glucose measurement data 1410 from the blood glucose meter or other sensor arrangement that provides discrete glucose measurements, contextual data 1420 from the source of user activity data (e.g., from an activity tracker, electrodermal activity sensor, temperature sensor, oxygen monitor, etc.), and contextual data 1430 from the other sources such as nutritional information about meals consumed by a user, insulin delivered to the user by an insulin infusion device of the user.
Data for each of the inputs can be collected for any number of users for a time period. Stated differently, for each user within a population, data can be collected for different input channels. Each of the different input channels can represent a different variable being measured for each user. In some embodiments, the different input channels can include (1) time of day, (2) a blood glucose level, (3) a blood glucose (BG) measurement from a sensor such as a blood glucose meter, activity data including (4) heart rate (beats per minute), (5) METs and (6) number of steps, (7) active insulin, (8) carbohydrates on board, etc. Active insulin can refer to bolus insulin that has already been delivered to a user's body, but that has not yet been used. An active insulin setting can be considered in determining any active insulin still in a user's body from their prior boluses and that could continue to lower their blood glucose.
In some embodiments, the discrete glucose measurement data 1410, contextual data 1420 from the source of user activity data and contextual data 1430 from the other sources can then be processed at a population learning processor 1440 to generate the optimized population model 1470. In this example, as part of the learning process, the population learning processor 1440 can perform various processing tasks with respect to the discrete glucose measurement data 1410 from the blood glucose meter, contextual data 1420 from the source of user activity data and contextual data 1430 from the other sources to generate the optimized population model 1470. For example, the population learning processor 1440 can apply various machine learning models 1450 including any described herein, such as deep learning models including but not limited to CNNs and RNNs, to the discrete glucose measurement data 1410 from the blood glucose meter, contextual data 1420 from the source of user activity data and contextual data 1430 from the other sources. For instance, in some embodiments, a deep learning network can perform a deep learning process to process a feature matrix to generate a sequence of estimated blood glucose values.
The population learning processor 1440 can also perform various parameter optimizations based on the discrete glucose measurement data 1410 from the blood glucose meter, contextual data 1420 from the source of user activity data and contextual data 1430 from the other sources. For example, one or more machine learning or deep learning model(s) can learn the transfer function or mapping from a series of inputs (e.g., discrete glucose measurement data 1410 from the blood glucose meter, contextual data 1420 from the source of user activity data and contextual data 1430 from the other sources) to a glucose measurement from a glucose sensor. This mapping can be on any time interval such that the model estimates anywhere from a single glucose value to a series of glucose values continuous in time. Each machine learning or deep learning model may have one or more parameters that need to be identified to specify the mathematical transfer function. Parameter optimization generally includes an objective function that is minimized. The objective function measures the mathematical agreement between the estimated output of the model with the actual measured data. Parameters are typically interactively adjusted until the objective function is optimized. The optimization process terminates once the level of agreement reaches a desired threshold or no longer improves.
As in
In this embodiment, each time window can include a discrete segment of the discrete glucose measurement data 1510, contextual data 1520 from the source of user activity data and contextual data 1530. Each time window can then be processed at the population learning processor 1540 to generate the optimized window population model 1570 corresponding to a particular time window. As part of the learning process, the population learning processor 1540 can perform various processing tasks described above with reference to
As in
In this embodiment, a window joining processor 1665 is provided that can join sets of the estimated glucose values corresponding to any number of time windows to generate a joined set of estimated glucose values over any number of time windows. For example, in some embodiments, estimated blood glucose sequences over each time window can be stored, and then be sequentially linked together (e.g., assembled/concatenated) at a later time. For example, in some embodiments, estimated blood glucose sequences over each time window can be stored, and when building the model, the stored outputs corresponding to each of the time windows can then be sequentially linked together (e.g., assembled/concatenated) to create the estimation model for the population of users. For example, in some embodiments, if a two hour window filter is considered, the output of the population model (estimated glucose values) would be limited to two hours, as such to generate a longer period output, multiple outputs from multiple windows may be joined. For instance, three two-hour outputs may be joined to generate a six hour model output. In some embodiments, the window joining step takes output of the previous window's estimated glucose response to initialize the subsequent window. This may be done by replacing or modifying one of the inputs to contain a segment of, or the entire, estimated glucose response output personalized learning step applied to the previous window.
As in
In this non-limiting embodiment, the optimized window population model 1670 (generated in
As in
In this embodiment, the optimized window population model 1670 (generated in
In this embodiment, the other data 1867 can include, for example, one or more past blood glucose meter readings, past glucose sensor data, time and blood glucose value that has the least daily, weekly or monthly variance, past data from any one of the model inputs, etc. Utilizing the other data 1867 in alone or in conjunction with the discrete glucose measurement data 1810 can improve the accuracy and/or calibration interval of the accuracy of the calibration interval of the optimized window population model that is output by the calibration optimization processor 1866. As described above with respect to the calibration optimization processor 1766, other data 1867 from other sources may be used in conjunction with or in place of discrete glucose measurement data 1810. For instance, other data 1867 may provide enough information to provide an accurate calibration method, without the need for invasive blood glucose meter readings. In another instance, to reduce the frequency (lengthen the interval) of blood glucose meter readings, other data 1867 may be used to assist the calibration model to maintain satisfactory performance of the population model accuracy. Processing performed by the calibration optimization processor 1866, and the variety of inputs or data it can use in place of historical and blood glucose meter data, will be described in greater detail below with reference to
As in
In this embodiment, the optimized window population model 1870 (
During this part of the workflow process data for multiple users can be collected for processing. In this example, the data that is collected can include, but is not limited to, discrete glucose measurement data 2010 from the blood glucose meter, contextual data 2020 from the source of user activity data, and contextual data 2030 from the other sources.
The discrete glucose measurement data 2010 from the blood glucose meter, contextual data 2020 from the source of user activity data and contextual data 2030 from the other sources can then be processed at a data modulation processor 2040 to generate a modulated form of the data. For example, in some embodiments, the input data can be modulated (modified) or simulated to generate a change in the content of one or more inputs. The modulated data is executed by a population model to generate the estimated glucose output. In the model response testing step, the output estimated glucose output is compared to a reference or standard that represents the typical directionality or magnitude of a human physiological response to a similar input data modulation. The reference or standard can be, not limited to, documented or general know how of what is typically observed clinically, medically, or physiologically. For example, the typical human physiological response to insulin is a drop in blood glucose levels. Modulation of an insulin input by reduction of the magnitude of the insulin level, in a physiologically accurate model, should result in a lower estimated glucose output.
At 2050 through 2060, the model explainability analysis processor 2000 can then perform processing to evaluate each population model of a number of population models to determine which of those population models should be added to a pool 2080 of candidate population models, and then select (at 2090) an optimized population model from the pool 2080 of candidate population models as an optimized population model (e.g., an optimized window population model, an optimized window population model, etc.). At 2050, the model explainability analysis processor 2000 selects a population model to execute and applies the modulated form of the data to the selected population model to generate a response to the modulated form of the data. At 2055, the model explainability analysis processor 2000 can perform various tests that help assess or evaluate whether the response (that was generated at 2050) is physiologically plausible. At 2060, the model explainability analysis processor 2000 can then analyze or evaluate the test results generated at 2055 to determine whether the selected population model's response is a plausible physiological response. When the model explainability analysis processor 2000 determines (at 2060) that the test results generated at 2055 indicate that the selected population model's response is not a plausible physiological response, the model explainability analysis processor 2000 drops the selected population model from further consideration.
By contrast, when the model explainability analysis processor 2000 determines (at 2060) that the test results generated at 2055 indicate that the selected population model's response is a plausible physiological response, the model explainability analysis processor 2000 adds the selected population model to a pool 2080 of candidate population models at 2070. At 2090, the model explainability analysis processor 2000 can select an optimized population model from the pool 2080 of candidate population models. The selected optimized population model can then be used as the basis for further training and personalized to generate a personalized model for estimating glucose values for a particular user or patient.
Referring again to
During this part of the workflow process data can be collected from different potential calibration sources 2110, 2120, 2130, 2140 for processing and evaluation. In this example, the data that is collected can include, but is not limited to, discrete glucose measurement data 2010 collected from a blood glucose meter, historical or sensor glucose trend data 2120, statistical or heuristic calibration point prediction data 2130, and glucose value data 2140 from blood glucose reading stability analysis. Historical or sensor glucose trends can include, but not limited to, glucose values from the past, several days of glucose sensor data, time of day point(s) where glucose is less or least variables. Statistical or heuristic calibration point prediction can include, but not limited to, a statistical, empirical, or rule-based method that estimates or predicts the glucose value at a given point in time, which may be used to calibrate the population model. Glucose value from blood glucose reading stability analysis, can include, but not limited to, an analysis of variability of glucose on an hourly, daily, weekly, or monthly basis. The glucose value that is most stable, less or least variable compared to other periods of time, may be selected to calibrate the population model. Each source of data 2110, 2120, 2130, 2140, or combination thereof, can be applied to a calibrate population model processor 2150 to generate a calibration model that is applied to the population model generating a corresponding calibrated population model response. In other words, while
An error and interval analysis processor 2160 can receive each calibrated population model response. For each calibrated population model response, the error and interval analysis processor 2160 can (1) perform an error analysis and generate an error result that indicates the performance of the calibrated population model response compared to the measured glucose values, and (2) perform an interval analysis to generate an interval result that indicates how often the population model would need to be calibrated with the given input. For each response, the error and interval analysis processor 2160 can then evaluate, the corresponding error result and the corresponding interval result.
At 2170, based on the evaluation of all of the error result/interval result combinations for each type of calibration model, the calibration optimization processor 2100 can compare performance of any given population model for each type of calibration model to determine an optimized calibration model for that given population model. The optimized calibration model can be selected, for example, to minimize error, maximize time between calibrations, and/or reduce user or patient burden. For example, a blood glucose meter reading every few minutes may minimize error but would be burdensome on the user, and as such would have a calibration interval that is too frequent. A blood glucose meter is considered an invasive measurement and thus burdensome to the patient, thus one of the other calibration models based on the alternate input sources that are less invasive or non-invasive may be preferred. The tradeoff between error, calibration interval, and patient burden, may be based on an objective function, clinical criteria, or product, regulatory or business requirements.
During this part of the workflow process data can be collected from different potential calibration sources 2210, 2220, 2230, 2240, 2245 for processing and evaluation. In this example, the data that is collected can include, but is not limited to, discrete glucose measurement data 2010 collected from a blood glucose meter, historical or sensor glucose trend data 2220, statistical or heuristic calibration point prediction data 2230, glucose value data 2240 from blood glucose reading stability analysis, and contextual data 2245 from the source of user activity data (e.g., an activity tracker), and/or contextual data 2245 from the other sources such as nutritional information about meals consumed by a user, insulin delivered to the user by an insulin infusion device of the user, etc. Historical or sensor glucose trends can include, but not limited to, glucose values from the past, several days of glucose sensor data, time of day point(s) where glucose is less or least variables. Statistical or heuristic calibration point prediction can include, but not limited to, a statistical, empirical, or rule-based method that estimates or predicts the glucose value at a given point in time, which may be used to calibrate the population model. Glucose value from blood glucose reading stability analysis, can include, but not limited to, an analysis of variability of glucose on an hourly, daily, weekly, or monthly basis. The glucose value that is most stable, less or least variable compared to other periods of time, may be selected to calibrate the population model. Each source of data 2210, 2220, 2230, 2240, 2245, or combination thereof, can be applied to a calibrate population model processor 2250 to generate a calibration model that is applied to the population model generating a corresponding calibrated population model response. In other words, while
An error and interval analysis processor 2260 can receive each calibrated population model response. For each calibrated population model response, the error and interval analysis processor 2260 can (1) perform an error analysis and generate an error result that indicates the performance of the calibrated population model response compared to the measured glucose values, and (2) perform an interval analysis to generate an interval result that indicates how often the population model would need to be calibrated with the given input. For each response, the error and interval analysis processor 2260 can then evaluate, the corresponding error result and the corresponding interval result.
At 2270, based on the evaluation of all of the error result/interval result combinations for each type of calibration model, the calibration optimization processor 2200 can compare performance of any given population model for each type of calibration model to determine an optimized calibration model for that given population model. The optimized calibration model can be selected, for example, to minimize error, maximize time between calibrations, and/or reduce user or patient burden. For example, a blood glucose meter reading every few minutes may minimize error but would be burdensome on the user, and as such would have a calibration interval that is too frequent. A blood glucose meter is considered an invasive measurement and thus burdensome to the patient, thus one of the other calibration models based on the alternate input sources that are less invasive or non-invasive may be preferred. The tradeoff between error, calibration interval, and patient burden, may be based on an objective function, clinical criteria, or product, regulatory or business requirements.
In this embodiment, once the optimized calibration model is determined (at 2270), the calibration optimization processor 2200 can determine or select at 2280, based on the evaluation of all of the error result/interval result combinations for each type of input, optimal type(s) of inputs to be used with that optimized calibration model (from 2270) when it is used for calibrating any given population model. As such, once an optimized calibration model has been determined (at 2270), the calibration optimization processor 2200 can select (at 2280) one or more types of the inputs 2210, 2220, 2230, 2240, 2245 to use with that optimized calibration model.
As described above, at least one of the population models can be selected, and personalized model learning processes can be performed to personalize a selected population model for each particular user. Various embodiments of personalized model learning processes that can be performed will now be described below with reference to
Data for each of the inputs can be collected for the particular user over a time period. Stated differently, data for the particular user can be collected for different input channels. Each of the different input channels can represent a different variable being measured for that user. In some embodiments, the different input channels can include (1) time of day, (2) a blood glucose level, (3) a blood glucose (BG) measurement from a sensor such as a CGM device, activity data including (4) heart rate (beats per minute), (5) METs and (6) number of steps, (7) active insulin, (8) carbohydrates on board, etc.
In some embodiments, the discrete glucose measurement data 2310, contextual data 2320 from the source of user activity data and contextual data 2330 from the other sources can then be processed at a personalized learning processor 2340 along with a particular population model 2305 to train or adapt the particular population model 2305 and thus generate the optimized personal model 2370 that is tailored to the particular user or patient. In some embodiments, the particular population model 2305 serves as the machine learning model 2350 and learns based on the inputs 2310, 2320, 2330 such that the particular population model 2305 is transformed (e.g., re-trained and optimized/personalized to the user) to become the optimized personal model 2370 for estimating glucose values that are personalized for a particular user or patient. In other embodiments, the particular population model 2305 is one model of an ensemble of the machine learning models 2350, and the ensemble of machine learning models 2350 learns based on the inputs 2310, 2320, 2330 such that the ensemble is transformed to become the optimized personal model 2370 for estimating glucose values that are personalized for a particular user or patient. In this example, as part of the personalized learning process, the personalized learning processor 2340 can perform various processing tasks with respect to the discrete glucose measurement data 2310 from the blood glucose meter, contextual data 2320 from the source of user activity data and contextual data 2330 from the other sources to train or adapt the particular population model 2305 to generate the optimized personal model 2370. For example, the personalized learning processor 2340 can apply various machine learning models 2350 including the particular population model 2305, to the discrete glucose measurement data 2310, contextual data 2320 from the source of user activity data and the other contextual data 2330 from the other sources to train or adapt the particular population model 2305. In another embodiment, the particular population model 2305 is combined with a separate personalized model to create an ensemble of population and personalized models that constitute a single optimized personal model. In another embodiment, the particular population model 2305 maybe disregarded and skipped over (e.g., given a zero or near zero weight), in which case the personalized learning processor 2340 can apply other various machine learning models 2350 to generate the optimized personal model 2370 that is tailored to the particular user or patient.
The personalized learning processor 2340 can also perform various parameter optimizations based on the discrete glucose measurement data 2310 from the blood glucose meter, contextual data 2320 from the source of user activity data and contextual data 2330 from the other sources to further train the particular population model 2305. For example, one or more machine learning or deep learning model(s) can learn the transfer function or mapping from a series of inputs to a glucose measurement from a glucose sensor. This mapping can be on any time interval such that the model estimates anywhere from a single glucose value to a series of glucose values continuous in time. Each machine learning or deep learning model may have one or more parameters that need to be identified to specify the mathematical transfer function. Parameter optimization generally includes an objective function that must be minimized. The objective function measures the mathematical agreement between the estimated output of the model with the actual measured glucose values. Parameters are typically interactively adjusted until the objective function is optimized. The optimization process terminates once the level of agreement reaches a desired threshold or no longer improves.
As described above with reference to
In this embodiment, each time window can include a discrete segment of the discrete glucose measurement data 2410, contextual data 2420 from the source of user activity data and contextual data 2430 for the particular user. Each time window can then be processed at the personalized learning processor 2440 to generate the optimized personal model 2470 corresponding to a particular time window. As part of the learning process, the personalized learning processor 2440 can perform various processing tasks described above with reference to
As in
In this embodiment, a window joining processor 2565 is provided that can join sets of the estimated glucose values corresponding to any number of time windows to generate a joined set of estimated glucose values over any number of time windows. For example, in some embodiments, estimated blood glucose sequences over each time window can be stored, and then be sequentially linked together (e.g., assembled/concatenated) at a later time. For example, in some embodiments, estimated blood glucose sequences over each time window can be stored, and the stored outputs corresponding to each of the time windows can then be sequentially linked together (e.g., assembled/concatenated). In some embodiments, the window joining step takes output of the previous window's estimated glucose response to initialize the subsequent window. This may be done by replacing or modifying one or more of the inputs to contain a segment of, or the entire, estimated glucose response output personalized learning step applied to the previous window.
As in
In this embodiment, the optimized personal model 2570 (generated in
As in
As explained with reference to
In this embodiment, the optimized personal model that is output by the calibration optimization processor 2766 can then be processed further by the model explainability analysis processor 2768 to generate an optimized personal model 2770 that provides a physiologically consistent response to one or more of the model inputs. For example, the typical human physiological response to food intake is a rise in blood glucose levels. Therefore, each candidate population model may be tested to ensure the response is at least directionally similar, or similar in magnitude, to what is observed clinically, medically, or physiologically. As described with reference to
While at least one exemplary embodiment has been presented in the foregoing detailed description, it should be appreciated that a vast number of variations exist. It should also be appreciated that the exemplary embodiment or embodiments described herein are not intended to limit the scope, applicability, or configuration of the claimed subject matter in any way. Rather, the foregoing detailed description will provide those skilled in the art with a convenient road map for implementing the described embodiment or embodiments. It should be understood that various changes can be made in the function and arrangement of elements without departing from the scope defined by the claims, which includes known equivalents and foreseeable equivalents at the time of filing this patent application.
Claims
1. A method, comprising:
- receiving data for each user within a population of users from a number of different input channels over a time period, the received data for each user comprising: discrete blood glucose measurement data for that user; user activity data for that user; and other contextual data for that user;
- processing the received data for the population of users to generate an input data set;
- processing the input data set, via a supervised machine learning model, to learn a transfer function for a population model that estimates blood glucose values for the population of users by mapping the received data for the population of users over the time period to a sequence of estimated blood glucose values for the population of users over the time period, wherein the supervised machine learning model has parameters that are estimated to generate the transfer function for the population model; and
- performing a parameter optimization process by adjusting one or more parameters of the supervised machine learning model to generate an optimized population model that estimates blood glucose values for the population of users by mapping the received data for the population of users over the time period to the sequence of estimated blood glucose values for the population of users over the time period.
2. The method of claim 1, wherein performing the parameter optimization process comprises:
- iteratively adjusting one or more parameters of the supervised machine learning model;
- continuously evaluating an objective function that measures a level of mathematical agreement between the estimated blood glucose values output by the machine learning model and actual measured blood glucose levels; and
- generating the optimized population model for the population of users when the level of mathematical agreement between the estimated blood glucose values output by the machine learning model and the actual measured blood glucose levels reaches a desired threshold,
- wherein the optimized population model for the population of users has the parameters of the transfer function set to values that were being evaluated when the mathematical agreement reached the desired threshold.
3. The method of claim 2, wherein the optimized population model is an optimized window population model, the method further comprising:
- applying a window filter to the received data, prior to processing the received data for the population of users, to split the received data into a series of different time windows each having a period that is less than the time period that the received data was received over, wherein each time window includes a discrete time segment of the received data; and
- wherein processing the received data for the population of users to generate the input data set, comprises:
- processing each discrete time segment of the received data to generate a corresponding input data set based on the received data received over a particular time window; and
- wherein processing the input data set, comprises:
- processing each corresponding input data set, via the supervised machine learning model, to learn a corresponding transfer function for a corresponding population model that estimates blood glucose values for the population of users by mapping the received data for the population of users over the particular time window to the sequence of estimated blood glucose values for the population of users over the particular time window, wherein the supervised machine learning model has parameters that are estimated to generate each corresponding transfer function for the population model.
4. The method of claim 3, further comprising:
- selecting the corresponding population model having a lowest error as the optimized population model.
5. The method of claim 3, wherein the optimized window population model is an optimized extended window population model, the method further comprising:
- sequentially joining each of the sequences of estimated blood glucose values for the population of users for each particular time window to generate a joined set of estimated blood glucose values over a number of the particular time windows to generate the optimized extended window population model that estimates the blood glucose values for the population of users.
6. The method of claim 1, wherein the discrete blood glucose measurement data for each user is measured using a sensor arrangement that provides discrete blood glucose measurements for that user, wherein the user activity data for each user is provided from a source of user activity data that correlates to activity of that user, and wherein the other contextual data for each user is provided from at least one other source associated with that user and provides other information about that user.
7. The method of claim 1, wherein each input channel comprises a different data type being measured for that user over the time period.
8. A system, comprising:
- one or more hardware-based processors configured by machine-readable instructions to: receive data for each user within a population of users from a number of different input channels over a time period, the received data for each user comprising: discrete blood glucose measurement data for that user; user activity data for that user; and other contextual data for that user; process the received data for the population of users to generate an input data set; process the input data set, via a supervised machine learning model, to learn a transfer function for a population model that estimates blood glucose values for the population of users by mapping the received data for the population of users over the time period to a sequence of estimated blood glucose values for the population of users over the time period, wherein the supervised machine learning model has parameters that are estimated to generate the transfer function for the population model; and perform a parameter optimization process by adjusting one or more parameters of the supervised machine learning model to generate an optimized population model that estimates blood glucose values for the population of users by mapping the received data for the population of users over the time period to the sequence of estimated blood glucose values for the population of users over the time period.
9. The system according to claim 8, wherein the one or more hardware-based processors are further configured by machine-readable instructions to:
- iteratively adjust one or more parameters of the supervised machine learning model;
- continuously evaluate an objective function that measures a level of mathematical agreement between the estimated blood glucose values output by the machine learning model and actual measured blood glucose levels; and
- generate the optimized population model for the population of users when the level of mathematical agreement between the estimated blood glucose values output by the machine learning model and the actual measured blood glucose levels reaches a desired threshold, wherein the optimized population model for the population of users has the parameters of the transfer function set to values that were being evaluated when the mathematical agreement reached the desired threshold.
10. The system according to claim 9, wherein the optimized population model is an optimized window population model, and wherein the one or more hardware-based processors are further configured by machine-readable instructions to:
- apply a window filter to the received data, prior to processing the received data for the population of users, to split the received data into a series of different time windows each having a period that is less than the time period that the received data was received over, wherein each time window includes a discrete time segment of the received data; and
- process each discrete time segment of the received data to generate a corresponding input data set based on the received data received over a particular time window; and
- process each corresponding input data set, via the supervised machine learning model, to learn a corresponding transfer function for a corresponding population model that estimates blood glucose values for the population of users by mapping the received data for the population of users over the particular time window to the sequence of estimated blood glucose values for the population of users over the particular time window, wherein the supervised machine learning model has parameters that are estimated to generate each corresponding transfer function for the population model.
11. The system according to claim 10, wherein the one or more hardware-based processors are further configured by machine-readable instructions to
- select the corresponding population model having a lowest error as the optimized population model.
12. The system according to claim 10, wherein the optimized window population model is an optimized extended window population model, and wherein the one or more hardware-based processors are further configured by machine-readable instructions to:
- sequentially join each of the sequences of estimated blood glucose values for the population of users for each particular time window to generate a joined set of estimated blood glucose values over a number of the particular time windows to generate the optimized extended window population model that estimates the blood glucose values for the population of users.
13. The system according to claim 14, the discrete blood glucose measurement data for each user is measured using a sensor arrangement that provides discrete blood glucose measurements for that user, wherein the user activity data for each user is provided from a source of user activity data that correlates to activity of that user, and wherein the other contextual data for each user is provided from at least one other source associated with that user and provides other information about that user.
14. At least one non-transient computer-readable medium having instructions stored thereon that are configurable to cause at least one processor to perform a method, the method comprising:
- receiving data for each user within a population of users from a number of different input channels over a time period, the received data for each user comprising: discrete blood glucose measurement data for that user; user activity data for that user; and other contextual data for that user;
- processing the received data for the population of users to generate an input data set;
- processing the input data set, via a supervised machine learning model, to learn a transfer function for a population model that estimates blood glucose values for the population of users by mapping the received data for the population of users over the time period to a sequence of estimated blood glucose values for the population of users over the time period, wherein the supervised machine learning model has parameters that are estimated to generate the transfer function for the population model; and
- performing a parameter optimization process by adjusting one or more parameters of the supervised machine learning model to generate an optimized population model that estimates blood glucose values for the population of users by mapping the received data for the population of users over the time period to the sequence of estimated blood glucose values for the population of users over the time period.
15. The computer-readable medium according to claim 14, the method further comprising:
- iteratively adjusting one or more parameters of the supervised machine learning model;
- continuously evaluating an objective function that measures a level of mathematical agreement between the estimated blood glucose values output by the machine learning model and actual measured blood glucose levels; and
- generating the optimized population model for the population of users when the level of mathematical agreement between the estimated blood glucose values output by the machine learning model and the actual measured blood glucose levels reaches a desired threshold,
- wherein the optimized population model for the population of users has the parameters of the transfer function set to values that were being evaluated when the mathematical agreement reached the desired threshold.
16. The computer-readable medium according to claim 15, wherein the optimized population model is an optimized window population model, the method further comprising:
- applying a window filter to the received data, prior to processing the received data for the population of users, to split the received data into a series of different time windows each having a period that is less than the time period that the received data was received over, wherein each time window includes a discrete time segment of the received data; and
- wherein processing the received data for the population of users to generate the input data set, comprises:
- processing each discrete time segment of the received data to generate a corresponding input data set based on the received data received over a particular time window; and
- wherein processing the input data set, comprises:
- processing each corresponding input data set, via the supervised machine learning model, to learn a corresponding transfer function for a corresponding population model that estimates blood glucose values for the population of users by mapping the received data for the population of users over the particular time window to the sequence of estimated blood glucose values for the population of users over the particular time window, wherein the supervised machine learning model has parameters that are estimated to generate each corresponding transfer function for the population model.
17. The computer-readable medium according to claim 16, the method further comprising:
- selecting the corresponding population model having a lowest error as the optimized population model.
18. The computer-readable medium according to claim 16, wherein the optimized window population model is an optimized extended window population model, the method further comprising:
- sequentially joining each of the sequences of estimated blood glucose values for the population of users for each particular time window to generate a joined set of estimated blood glucose values over a number of the particular time windows to generate the optimized extended window population model that estimates the blood glucose values for the population of users.
19. The computer-readable medium according to claim 14, the discrete blood glucose measurement data for each user is measured using a sensor arrangement that provides discrete blood glucose measurements for that user, wherein the user activity data for each user is provided from a source of user activity data that correlates to activity of that user, and wherein the other contextual data for each user is provided from at least one other source associated with that user and provides other information about that user.
20. The computer-readable medium according to claim 14, wherein each input channel comprises a different data type being measured for that user over the time period.
Type: Application
Filed: Aug 6, 2020
Publication Date: Feb 10, 2022
Inventors: Arthur Mikhno (Princeton, NJ), Yuxiang Zhong (Arcadia, CA), Pratik Agrawal (Porter Ranch, CA)
Application Number: 16/987,330
