Abstract: The present description relates generally to methods and systems for power management of a digital (e.g., electronic) stethoscope. In one example, an electronic stethoscope includes a chestpiece configured to be positioned on a patient, the chestpiece including a sensor, wherein the chestpiece further includes a computer processing unit (CPU) operatively coupled to a memory storing instructions that, when executed by the CPU, cause the CPU to automatically activate the electronic stethoscope in response to detecting a touch input via the sensor.

Abstract: An automated tuning service is used to automatically tune, or modify, the operational parameters of a large-scale cloud infrastructure. The tuning service performs automated and fully data/model-driven configuration based from learning various real-time performance of the cloud infrastructure. Such performance is identified through monitoring various telemetric data of the cloud infrastructure. The tuning service leverages a mix of domain knowledge and principled data-science to capture the essence of our cluster dynamic behavior in a collection of descriptive machine learning (ML) models. The ML models power automated optimization procedures for parameter tuning, and inform administrators in most tactical and strategical engineering/capacity decisions (such as hardware and datacenter design, software investments, etc.).

Abstract: Approaches described herein can capture an audio signal using at least one microphone while a user of an electronic device is determined to be asleep. At least one audio frame can be determined from the audio signal. The at least one audio frame represents a spectrum of frequencies detected by the at least one microphone over some period of time. One or more sounds associated with the at least one audio frame can be determined. Sleep-related information can be generated. The information identifies the one or more sounds as potential sources of sleep disruption.

Abstract: An automated tuning service is used to automatically tune, or modify, the operational parameters of a large-scale cloud infrastructure. The tuning service performs automated and fully data/model-driven configuration based from learning various real-time performance of the cloud infrastructure. Such performance is identified through monitoring various telemetric data of the cloud infrastructure. The tuning service leverages a mix of domain knowledge and principled data-science to capture the essence of our cluster dynamic behavior in a collection of descriptive machine learning (ML) models. The ML models power automated optimization procedures for parameter tuning, and inform administrators in most tactical and strategical engineering/capacity decisions (such as hardware and datacenter design, software investments, etc.).

Abstract: Embodiments described herein enable tracking machine learning (“ML”) model data provenance. In particular, a computing device is configured to accept ML model code that, when executed, instantiates and trains an ML model, to parse the ML model code into a workflow intermediate representation (WIR), to semantically annotate the WIR to provide an annotated WIR, and to identify, based on the annotated WIR and ML API corresponding to the ML model code, data from at least one data source that is relied upon by the ML model code when training the ML model. A WIR may be generated from an abstract syntax tree (AST) based on the ML model code, generating provenance relationships (PRs) based at least in part on relationships between nodes of the AST, wherein a PR comprises one or more input variables, an operation, a caller, and one or more output variables.

Abstract: Linguistic schema mapping via semi-supervised learning is used to map a customer schema to a particular industry-specific schema (ISS). The customer schema is received and a corresponding ISS is identified. An attribute in the customer schema is selected for labeling. Candidate pairs are generated that include the first attribute and one or more second attributes which may describe the first attribute. A featurizer determines similarities between the first attribute and second attribute in each generated pair, one or more suggested labels are generated by a machine learning (ML) model, and one of the suggested labels is applied to the first attribute.

Abstract: Approaches described herein can capture an audio signal using at least one microphone while a user of an electronic device is determined to be asleep. At least one audio frame can be determined from the audio signal. The at least one audio frame represents a spectrum of frequencies detected by the at least one microphone over some period of time. One or more sounds associated with the at least one audio frame can be determined. Sleep-related information can be generated. The information identifies the one or more sounds as potential sources of sleep disruption.

Abstract: A wearable computing device includes an electronic display with a configurable brightness level setting, a physiological metric sensor system including a light source configured to direct light into tissue of a user wearing the wearable computing device and a light detector configured to detect light from the light source that reflects back from the user. The device may further include control circuitry configured to activate the light source during a first period, generate a first light detector signal indicating a first amount of light detected by the light detector during the first period, deactivate the light source during a second period, generate a second light detector signal indicating a second amount of light detected by the light detector during the second period, generate a physiological metric based at least in part on the first light detector signal and the second light detector signal, and modify the configurable brightness level setting based on the second light detector signal.

Abstract: A system enables tracking machine learning (“ML”) model data provenance. In particular, a computing device is configured to accept ML model code that, when executed, instantiates and trains an ML model, to parse the ML model code into a workflow intermediate representation (WIR), to semantically annotate the WIR to provide an annotated WIR, and to identify, based on the annotated WIR and ML API corresponding to the ML model code, data from at least one data source that is relied upon by the ML model code when training the ML model. A WIR may be generated from an abstract syntax tree (AST) based on the ML model code, generating provenance relationships (PRs) based at least in part on relationships between nodes of the AST, wherein a PR comprises one or more input variables, an operation, a caller, and one or more output variables.

Abstract: An electronic stethoscope device can be integrated into a conventional stethoscope to digitize auscultated sounds from the body of a patient. The device can be switched off so that the conventional stethoscope can be used as a standard stethoscope. When the device is switched on, the digitized auscultated sounds can be modified to remove the noise. Such modified sounds can be sent wirelessly from the electronic stethoscope device to a peripheral device that can receive such wireless signals, such as computer, cell phone, or cloud application, where the data can be viewed and manipulated further as desired.

Abstract: Aspects of automatically detecting periods of sleep of a user of a wearable electronic device are discussed herein. For example, in one aspect, an embodiment may obtain a set of features for periods of time from motion data obtained from a set of one or more motion sensors in the wearable electronic device or data derived therefrom. The wearable electronic device may then classify the periods of time into one of a plurality of statuses of the user based on the set of features determined for the periods of time, where the statuses are indicative of relative degree of movement of the user. The wearable electronic device may also derive blocks of time each covering one or more of the periods of time during which the user is in one of a plurality of states, wherein the states include an awake state and an asleep state.

Abstract: Various methods and systems are provided for monitoring a pulmonary artery pressure and cardiac synchronization of a subject. In one example, a method includes acquiring at least one of electrocardiogram (ECG) data, phonocardiogram (PCG) data, and seismocardiogram (SCG) data from a subject via a digital stethoscope, inputting one or more of the ECG data, the PCG data, and the SCG data into a machine learning algorithm, and estimating at least one of a pulmonary artery pressure and a cardiac synchronization of the subject using the machine learning algorithm. In this way, the pulmonary artery pressure and the cardiac synchronization may be estimated using artificial intelligence and inputs that are non-invasively measured by the digital stethoscope, allowing conditions like heart failure and pulmonary hypertension to be more simply and reliably detected and monitored.

Abstract: In one embodiment, a data processing method comprises obtaining one or more first photoplethysmography (PPG) signals based on one or more first light sources that are configured to emit light having a first light wavelength corresponding to a green light wavelength; obtaining one or more second PPG signals based on one or more second light sources that are configured to emit light having a second light wavelength corresponding to a red light wavelength, one or more of the first light sources and one or more of the second light sources being co-located; generating an estimated heart rate value based on one or more of the first PPG signals and the second PPG signals; and causing the estimated heart rate value to be displayed via a user interface on a client device.

Abstract: Techniques for music selection based on exercise detection are disclosed. In one aspect, a method of operating a wearable device may involve determining, based on output of one or more biometric sensors, that a user of the wearable device has started an exercise and playing music for the user of the wearable device in response to determining the start of the exercise. For example, playing the music may involve turning on a music player based on the start of the exercise. In another example, the wearable device includes the music player.

Abstract: Various methods and systems are provided for an automated clinical exam workflow. In one example, method comprises performing a signal quality check of an electronic stethoscope at a first recording location on a subject, recording physiological data for an exam at the first recording location via the electronic stethoscope in response to the signal quality check satisfying a quality threshold, and outputting a signal quality alert in response to the signal quality check not satisfying the quality threshold. In this way, clinically relevant data may be obtained with reduced user effort and fewer manual inputs.

Abstract: Methods, systems, apparatuses, and computer-readable storage mediums described herein are directed to determining and recommending an optimal compute resource configuration for a cloud-based resource (e.g., a server, a virtual machine, etc.) for migrating a customer to the cloud. The embodiments described herein utilize a statistically robust approach that makes recommendations that are more flexible (elastic) and account for the full distribution of the amount of resource usage. Such an approach is utilized to develop a personalized rank of relevant recommendations to a customer. To determine which compute resource configuration to recommend to the customer, the customer’s usage profile is matched to a set of customers that have already migrated to the cloud. The compute resource configuration that reaches the performance most similar to the performance of the configurations utilized by customers in the matched set is recommended to the user.

Abstract: A wearable device for tracking swim activities of a user is provided. The wearable device may include one or more sensors configured to generate sensor data, and based on the sensor data, the wearable device may determine swim metrics such as swim stroke count, swim stroke type, swim lap count, and swim speed. The determined swim metrics may be filtered based on one or more swim periods during which the user is likely to have been swimming. The wearable device may determine such swim periods based on the sensor data and/or the determined swim metrics.

Abstract: The present description relates generally to methods and systems for wireless communication between a digital (e.g., electronic) stethoscope and other electronic devices (e.g., computing and listening devices). In one example, a method comprises operating an electronic stethoscope in one of an internal digital mode and a wireless digital mode based on a detected user action. In this way, the electronic stethoscope may be quickly adjusted between projecting sound via integrated speakers and transmitting sound to an external electronic device, enabling efficient remote monitoring of a patient.

Abstract: An electronic stethoscope device can be integrated into a conventional stethoscope to digitize auscultated sounds from the body of a patient. The device can be switched off so that the conventional stethoscope can be used as a standard stethoscope. When the device is switched on, the digitized auscultated sounds can be modified to remove the noise. Such modified sounds can be sent wirelessly from the electronic stethoscope device to a peripheral device that can receive such wireless signals, such as computer, cell phone, or cloud application, where the data can be viewed and manipulated further as desired.

Abstract: In an embodiment, a data processing method comprises obtaining one or more photoplethysmography (PPG) signals from one or more PPG sensors of a monitoring apparatus, the PPG signals being generated based upon optically detecting pulsed variations in blood flow; obtaining a motion sensor signal from a motion sensor in the monitoring apparatus; identifying, based upon the motion sensor signal, one or more periods of motion (e.g., low motion) of the monitoring apparatus; and selectively obtaining and storing segments of the PPG signals based on a temporal relationship between the segments of the PPG signals and the identified periods of motion.