Abstract: Techniques for classifying heartbeats using patient electrocardiogram (ECG) data are described. ECG data is received, including waveform data and time interval data relating to a plurality of heartbeats for the patient. A convolutional neural network in a first path of a machine learning architecture generates a first plurality of output values by analyzing the waveform data. A fully-connected neural network in a second path of the machine learning architecture generates a second plurality of output values by analyzing the time interval data. The plurality of heartbeats in the ECG data are classified by concatenating the first plurality of output values and the second plurality of output values using the machine learning architecture.

Abstract: Techniques for securing user data in a healthcare data management system are described. A client system receives a request to authenticate a user. The user is associated with applications and roles. The user is authenticated, at the client system, for all applications and all roles. A login token relating to the authenticated user is maintained at the client system. A role is selected for the user, and an authorization token relating to the selected role is maintained at the client system. A session for the user is initiated. This includes generating an encrypted session cookie relating to the user and the session, storing the encrypted session cookie at the client system, and periodically updating a timestamp for the session cookie.

Abstract: Techniques for administering a care plan for a patient. A monitoring device is selected, at a care plan management system based on a care plan, to use in administering the care plan. The care plan specifies biometric data to monitor for a patient. The monitoring device is configured to collect the biometric data, based on the care plan. The biometric data collected using the monitoring device is received at the care plan management system. The biometric data includes a first event initially classified as a first type of event using the monitoring device. The first event is selected for reclassification, based on the initial classification, and in response the first event is reclassified as a second type of event using the care plan management system. The patient is treated based on the care plan and the reclassified first event.

Abstract: Techniques for determining reliability of electrocardiogram (ECG) data are disclosed. ECG data, including waveform data for a patient, is received. An amplitude associated with an R-wave in the waveform data is determined. First and second baseline regions in the waveform data, each relating to the R-Wave, are identified. A signal-to-noise ratio (SNR) is determined, based on the amplitude, the first baseline region, and the second baseline region. The SNR facilitates medical treatment for the patient.

Abstract: Techniques for administering a care plan. Biometric data collected by a monitoring device is received at a care plan management system using a communication network. The biometric data includes a first event classified by a remote device as a first type of event. A relative processing priority for the first event is determined, at the care plan management system, based on the first type. The first event is processed, at the care plan management system, based on the relative processing priority. The processing includes re-classifying the first event as a second type of event. The re-classification is more computationally intensive than the classification by the remote device. At least one treatment task specified in the care plan is initiated, using a computer processor, and based on the re-classified first event.

Abstract: Embodiments presented herein provide a dynamically adaptive care plan. More specifically, embodiments presented herein describe techniques for modifying a care plan assigned to a patient based on data observed from the patient. According to one embodiment of the present disclosure, data corresponding to one or more types of biometric data for the patient is received. One or more reported symptoms for the patient is received. The care plan includes one or more conditions corresponding to the one or more reported symptoms. Upon determining that at least one of the received data satisfy at least one of the biometric thresholds and the one or more reported symptoms satisfy at least one of the conditions, an event rule for modifying the care plan is identified. The care plan is modified based on the event rule.

Abstract: Embodiments of the present disclosure generally relate wearable patches having rigid inserts. The rigid insert is positioned adjacent to one or more connectors, such as studs for receiving sockets to distribute the force transferred to a patient when reconnecting a socket to the stud. The rigid insert may be sized to maintain flexibility in areas of the wearable patch. An optional adhesive layer may be applied to the wearable patch adjacent the rigid insert to reduce the likelihood of delamination of the wearable patch.

Abstract: Embodiments presented herein describe techniques for generating a care plan tailored for an individual. According to one embodiment of the present disclosure, a care platform receives a selection of one or more care protocol templates to assign to a patient. Each care protocol template is associated with a medical condition to be treated. The care platform generates tasks to be performed by the patient, observation metrics specifying types of patient biometrics to monitor, and thresholds associated with each selection. The care platform receives one or more customizations to the tasks, observation metrics, and thresholds and determines whether any of the tasks or the customizations to the tasks conflict with one another. Upon identifying one or more conflicts, the conflicts are resolved based on specified rules associated with each conflicting task. The care plan is generated based on the observation metrics, thresholds, customizations, and resolved tasks.

Abstract: Embodiments presented herein describe techniques for providing feedback for a care plan for a patient. Embodiments receive a care plan specifying a plurality of assigned tasks for the patient to perform, timing information specifying when each of the plurality of assigned tasks should be performed, and a plurality of observation metrics that each indicate a type of biometric data to monitor. Biometric data corresponding to the plurality of observation metrics is collected using at least one monitoring device. Embodiments receive symptom information specifying one or more reported symptoms experienced by the patient and monitor adherence of the patient to completing the plurality of assigned tasks according to the timing information specified in the care plan. Embodiments further include providing feedback to the patient, based on the collected biometric data, received symptom information and monitored adherence.

Abstract: Collecting biometric data from a patient provides numerous opportunities for a care provider to monitor the health of the patient. In one embodiment, the biometric data is used to identify health events that are processed in a workflow that includes a plurality of interconnected processing nodes and queues. In one embodiment, the workflow processes the health events using a priority assigned to the health events. The priority may be assigned based on a severity of the event, the type of the health event, a characteristic of the patient whose biometric data generated the health event, and the like. Each processing node in the event engine may execute multiple consumers for processing health events with different priorities. The health event can further be re-classified based on processing by a consumer.

Abstract: Techniques for administering a care plan. Embodiments receive the care plan specifying observation metrics to monitor biometric data collected from a patient. At least one monitoring device available is identified and embodiments receive biometric data collected using the at least one monitoring device, where the biometric data is initially classified as a first type of event by the at least one monitoring device. Additionally, embodiments analyze the received biometric data to reclassify the first event as an occurrence of a second type of even, and, upon determining that the occurrence of the second type of event satisfies at least one threshold condition specified in the care plan, initiate at least one treatment plan specified in the care plan and corresponding to the satisfied at least one threshold value.

Abstract: Techniques for determining reliability of electrocardiogram (ECG) data are disclosed. ECG data, including waveform data relating to a detected heartbeat for a patient, is received. A peak amplitude associated with an R-wave in the waveform data is determined. First and second baseline regions in the waveform data are identified. The first baseline region precedes the R-wave and the second baseline region follows the R-wave. A signal-to-noise ratio (SNR) is determined, based on the peak amplitude, the first baseline region, and the second baseline region. A confidence metric relating to the waveform data is determined, based on the determined SNR. The confidence metric is used in medical treatment related to the patient.

Abstract: Techniques presented herein disclose a method for generating a derived observation for a care plan assigned to a patient. According to one embodiment, an application executing on a care platform server receives an event log having a plurality of events. Each event includes one or more types of biometric data for the patient. The care plan includes one or more definitions for metrics that are derived from the one or more types of biometric data. A temporal slice of the event log is selected based on one or more of the definitions. The derived metrics are generated by evaluating events within the selected temporal slice of the event log against one or more of the definitions. A condition of the patient is evaluated based on the derived metrics.

Abstract: Embodiments presented herein describe techniques for administering a care plan that describes an overall treatment regimen for a patient. Embodiments include receiving a care plan specifying a plurality of assigned tasks for the patient to perform, timing information specifying when each of the plurality of assigned tasks should be performed, and a plurality of observation metrics that each indicate a type of biometric data to monitor. At least one monitoring device available to collect biometric data to compare against observation metrics specified in the care plan is identified. Biometric data is collected using the at least one monitoring device. Embodiments further include, upon determining that the collected biometric data satisfies at least one threshold condition specified in the care plan, initiating at least one treatment plan specified in the care plan and corresponding to the satisfied at least one threshold value.

Abstract: Collecting biometric data from a patient provides numerous opportunities for a care provider to monitor the health of the patient. In one embodiment, the biometric data is used to identify health events that are processed in a workflow that includes a plurality of interconnected processing nodes and queues. In one embodiment, the workflow processes the health events using a priority assigned to the health events. The priority may be assigned based on a severity of the event, the type of the health event, a characteristic of the patient whose biometric data generated the health event, and the like. Each processing node in the event engine may include multiple consumers or threads for processing health events with different priorities. By having more consumers assigned to high priority events, the latency for processing these events in the workflow may be reduced.