Cardiac monitoring system with sleep and activity correlation

Abstract

A cardiac monitoring system with sleep and activity correlation is described. In one or more embodiments, measurements of a user generated by a wearable monitoring device during an observation period are acquired, including accelerometer data and potential measurements. Sleep periods and wake periods are detected based on the accelerometer data. Sleep stage classification and activity level classification are generated based on the accelerometer data during the detected sleep periods. Potential measurements are processed using a machine learning model trained to correlate patterns in the potential measurements with heart rhythm classifications. Arrhythmia correlations can be generated and output based on the sleep stage classification, activity level classification, and contemporaneous heart rhythm classification. Arrhythmia correlations can describe the temporal relationship between detected arrhythmias and specific sleep stages in the sleep stage classification and can be output in a health report.

Description

[0001] Cross-references to related applications

[0002] This application claims priority to U.S. Provisional Application No. 63 / 703,761, filed October 4, 2024, entitled “Analysis-triggered arrhythmia detection and correlation”; U.S. Provisional Application No. 63 / 703,806, filed October 4, 2024, entitled “Dynamic cardiac stress testing”; U.S. Provisional Application No. 63 / 703,833, filed October 4, 2024, entitled “Constrained sleep and activity prediction”; and U.S. Non-Provisional Application No. 19 / 347,508, filed October 1, 2025, entitled “Sleep and activity-related cardiac monitoring system”, the entire contents of which are incorporated herein by reference. Background Technology

[0003] Cardiac monitoring devices are typically used to detect and analyze cardiac arrhythmias over extended periods. Traditional electrocardiogram (ECG) monitoring systems acquire continuous cardiac data but often lack contextual information about patient activity and physiological status during data acquisition. Existing wearable cardiac monitors typically perform high-frequency data acquisition across multiple sensors, which can rapidly deplete battery resources and consume significant memory capacity. Furthermore, conventional cardiac monitoring methods may process and analyze cardiac rhythm data in isolation, without correlating detected arrhythmias with concurrent patient behaviors such as sleep patterns, physical activity levels, or body positions. Additionally, the continuous high-power operation of sensors in wearable devices can impair long-term patient adherence due to the need for frequent device charging or battery replacements and reduced wearing comfort, potentially affecting the quality and completeness of the acquired cardiac data. Attached Figure Description

[0004] Figure 1 This is a block diagram of a non-limiting example of an environment operable to employ the sleep and activity-related cardiac monitoring described herein.

[0005] Figure 2 Non-limiting examples of monitoring equipment are described.

[0006] Figure 3 A non-limiting system with sleep and activity-related cardiac monitoring is described in an example implementation, which is shown in more detail. Figure 1 The operation of the prediction system.

[0007] Figure 4 A non-limiting example of a monitoring system with sleep stages and activity-related cardiac monitoring is depicted, wherein the monitoring system generates a comprehensive health report for output.

[0008] Figure 5A and Figure 5B A non-limiting example of a user interface for displaying patient cardiac monitoring data with sleep and activity correlations is shown.

[0009] Figure 6 An example of an activity-based comprehensive cardiac assessment interface is shown.

[0010] Figure 7 An example of a multi-level activity classification display is shown, which provides a comprehensive analysis of cardiac activity at different exercise intensities.

[0011] Figure 8 An example of a dynamic cardiac stress testing interface is shown, which helps monitor and analyze cardiac activity during extended periods of wear.

[0012] Figure 9 A flowchart is provided that depicts the algorithm as a step-by-step process in an example implementation, which can be executed by a processing device to correlate cardiac monitoring with sleep and activity data based on measurements from a wearable monitoring device.

[0013] Figure 10 A flowchart is provided that depicts the algorithm as a step-by-step process in an example implementation, which can be executed by a processing device for cardiac monitoring with activity level classification and arrhythmia correlation.

[0014] Figure 11 A flowchart was drawn that depicts the algorithm as a step-by-step process in which dynamic cardiac stress testing is performed using wearable monitoring devices.

[0015] Figure 12 A flowchart was drawn that depicts the algorithm as a step-by-step process, in which analysis and data collection are triggered by sleep / activity. Detailed Implementation

[0016] Conventional cardiac monitoring systems typically use high-frequency sampling rates (e.g., 25–100 Hz) and multiple sensors to acquire continuous electrocardiogram (ECG) data to detect arrhythmias over extended periods. These conventional methods often operate in isolation from contextual information about patient activity, sleep patterns, and physiological status during data acquisition. Traditional ECG monitoring methods cannot correlate detected arrhythmias with contemporaneous patient behaviors such as sleep stage, physical activity level, or body position, which can hinder clinical assessment due to a lack of information about the surrounding circumstances of the detected cardiac event. Furthermore, conventional wearable cardiac monitors often rely on continuous high-power operation of sensors across multiple channels, which can rapidly deplete battery resources and consume significant memory capacity. This high power consumption can impair long-term patient adherence due to frequent device charging and / or reduced wearing comfort, potentially affecting the quality and integrity of the acquired cardiac data. Moreover, conventional systems may struggle to provide comprehensive physiological insights that integrate cardiac monitoring with sleep and activity analysis, thus limiting their ability to identify relationships between cardiac events and daily life patterns.

[0017] Therefore, this paper describes techniques and systems for cardiac monitoring with sleep and activity correlations that address these limitations by providing comprehensive cardiac analysis that correlates detected arrhythmias with contemporaneous sleep stages and / or activity levels, while operating efficiently with low-power sensor configurations. In one or more embodiments, a wearable monitoring device acquires measurements from the user during an observation period, including accelerometer data and potential measurements. Accelerometer data is acquired at a sampling rate of less than 5 Hz to extend battery life and reduce power consumption while maintaining accurate detection of sleep and activity patterns. Sleep and wake periods are detected based on patterns in the accelerometer data. For example, the analysis system can analyze body angle calculations and movement patterns to differentiate between different physiological states. Sleep stage classifications are generated based on accelerometer data during detected sleep periods, categorizing sleep stages into light sleep, deep sleep, rapid eye movement (REM) sleep, and / or non-REM sleep to provide a detailed temporal mapping of the sleep architecture. The system also generates activity level classifications based on the number of steps detected during waking periods, categorizing physical activity as sedentary, mild, moderate, or vigorous based on movement determined from accelerometer data that is at least equivalent to a threshold velocity.

[0018] The system uses at least one machine learning model to process potential measurements. The machine learning model is trained to associate patterns in the potential measurements with heart rhythm classifications, thereby enabling the identification of various arrhythmia types and cardiac events throughout the observation period. Arrhythmia correlations are generated based on sleep stage classifications and / or activity level classifications, as well as contemporaneous heart rhythm classifications. These correlations describe the temporal relationship between detected arrhythmias and specific sleep stages and / or activity levels. In one or more embodiments, odds ratio analysis is generated based on arrhythmia correlations to quantify the probability of a specific arrhythmia occurring during sleep and wakefulness periods, and to calculate the statistical association between specific arrhythmias and sleep stages within a user population.

[0019] In one or more embodiments, the system identifies the user's exercise periods during the observation period based on accelerometer data and generates arrhythmia correlations for periods before, during, and after the exercise periods based on concurrent heart rate classification. The system detects chronotropic incompetence by comparing the expected heart rate response during the activity event with the actual heart rate response, thereby identifying situations where cardiac output does not adequately increase during physical exertion.

[0020] In another implementation, the system performs dynamic cardiac stress testing by correlating exercise-related activity patterns with concurrent heart rate changes. The system can detect exercise periods based on accelerometer data and analyze concurrent cardiac responses based on potential data to identify exercise-induced arrhythmias and cardiac recovery patterns that may indicate underlying coronary artery disease or other cardiovascular conditions. Dynamic stress testing capabilities may include detecting ST-segment depression, ST-segment elevation, T-wave inversion, T-wave flattening, and T-wave pseudonormalization by comparing exercise and recovery ECG data with resting baseline measurements. Body position can be determined based on accelerometer data during the post-exercise recovery period, where the system identifies when the patient transitions to a supine position, which may induce ischemic abnormalities not visible during exercise. User instructions can be provided to facilitate the acquisition of data that will contribute to dynamic cardiac stress testing.

[0021] The system enables sleep / activity-triggered analysis and data acquisition, where different sleep stages and / or activity levels trigger adaptive monitoring strategies to reduce battery consumption while ensuring comprehensive data capture during physiologically relevant periods. For example, when sleep is detected via accelerometer analysis, the system can activate LED-based photoplethysmography (PPG) for blood oxygen saturation measurement, which can be used for sleep apnea screening and diagnosis but consumes more power. PPG can be deactivated when the user is no longer asleep to conserve battery resources. When activity is detected, the system can activate additional sensors such as gyroscopes to provide supplemental data that can be used to differentiate activity type or intensity; these sensors can remain deactivated during inactive periods to save power. This selective sensor activation approach allows the system to provide comprehensive physiological monitoring while extending battery life and improving long-term patient compliance during extended observation periods.

[0022] The described technique offers significant advantages over conventional methods, enabling comprehensive cardiac monitoring that correlates detected arrhythmias with concurrent sleep and activity states, while operating with a low-power sensor configuration that extends battery life and improves patient compliance. Combining accelerometer-based sleep and activity detection with concurrent cardiac monitoring provides comprehensive physiological analysis results that cannot be obtained by analyzing individual sensor modalities separately, allowing healthcare providers to understand cardiac events within the context of daily life patterns and sleep structures. Maintaining high accuracy in both sleep stage detection and activity classification while operating sensors at low sampling frequencies reduces hardware complexity and power consumption compared to conventional multi-sensor systems. Furthermore, the correlation between cardiac events and sleep stages and / or activity levels allows healthcare providers to make informed clinical decisions without relying entirely on patient recollection, particularly where specific sleep stages or activity intensities may trigger arrhythmia episodes. The following sections include further discussion of these and other examples and advantages, illustrated with corresponding figures.

[0023] In some aspects, the technology described herein relates to a method implemented by a processing device, the method comprising: acquiring measurements of a user generated by a wearable monitoring device during an observation period, the measurements including accelerometer data and potential measurements; detecting sleep periods and wake periods based on the accelerometer data; generating a sleep stage classification based on the accelerometer data during the detected sleep periods; processing the potential measurements using a machine learning model trained to correlate patterns in the potential measurements with a heart rhythm classification; generating arrhythmia correlations based on the sleep stage classifications and contemporaneous heart rhythm classifications, the arrhythmia correlations describing the temporal relationship between detected arrhythmias and specific sleep stages in the sleep stage classifications; and outputting a health report including the arrhythmia correlations.

[0024] In some respects, the techniques described herein relate to a method in which accelerometer data is acquired at a sampling rate of less than 5 Hz.

[0025] In some respects, the techniques described herein relate to a method in which detecting sleep and wake periods based on accelerometer data includes: detecting peak values ​​above a threshold in the accelerometer data as steps; calculating body angle measurements based on the accelerometer data to determine body position; and distinguishing between sleep and wake periods based on the detected steps and body angle measurements.

[0026] In some respects, the techniques described herein relate to a method in which generating a sleep stage classification includes categorizing sleep periods into light sleep, deep sleep, rapid eye movement (REM) sleep, and non-REM sleep.

[0027] In some respects, the technology described herein relates to a method that further includes: indicating activity events during the waking period based on accelerometer data acquired during the waking period; generating an activity level classification of the activity events based on the intensity of the physical activity performed; and generating an additional arrhythmia correlation based on the activity level classification and the corresponding contemporaneous cardiac rhythm classification, the additional arrhythmia correlation describing the temporal relationship between detected arrhythmias and a specific activity level in the activity level classification.

[0028] In some respects, the techniques described herein relate to a method in which activity level classification is based on movement at least equivalent to a threshold velocity determined from accelerometer data, categorizing physical activity as mild, moderate, or vigorous.

[0029] In some respects, the techniques described herein relate to a method that further includes detecting chronotropic dysfunction based on a comparison of expected and actual heart rate responses during physical activity.

[0030] In some respects, the techniques described herein relate to a method that further includes generating a ratio analysis based on arrhythmia correlations, the ratio analysis quantifying the probability of a particular arrhythmia occurring during sleep and wakefulness periods.

[0031] In some respects, the techniques described herein relate to a method in which odds ratio analysis calculates the statistical association between a specific arrhythmia and sleep periods within a user population.

[0032] In some respects, the techniques described herein relate to a method in which health reports include visualizations of the correlation between cardiac arrhythmias and sleep and wake periods.

[0033] In some aspects, the technology described herein relates to a processing device comprising: one or more processors; and a memory storing computer-readable instructions executable by the one or more processors to perform operations including: acquiring user measurements generated by a wearable monitoring device during an observation period, the measurements including accelerometer data and potential measurements; detecting sleep periods and wake periods based on the accelerometer data; generating a sleep stage classification based on the accelerometer data during the sleep period; generating an activity level classification based on the number of steps detected during the wake period; processing the potential measurements using a machine learning model trained to associate patterns in the potential measurements with a heart rhythm classification; associating the sleep stage classification and activity level classification with the heart rhythm classification during the same period to generate arrhythmia correlations; generating a heart wellness prediction based on the arrhythmia correlations; and outputting a health report including the heart wellness prediction.

[0034] In some respects, the technology described herein relates to a processing device in which accelerometer data is acquired at a sampling rate of less than 5 Hz.

[0035] In some respects, the technology described herein relates to a processing device in which generating an activity level classification includes categorizing physical activity as sedentary, mild, moderate, or vigorous based on the speed of motion determined from accelerometer data.

[0036] In some respects, the technology described herein relates to a processing device in which operation further includes detecting chronotropic dysfunction by comparing a user’s expected heart rate response during an active event within a waking period with the actual heart rate response.

[0037] In some respects, the technology described herein relates to a processing device in which the operation further includes: determining a user’s heart rate over time based on potential measurements; generating a time series plot of the heart rate during detected activity events; stratifying the time series plot based on activity level classification; and outputting the stratified time series plot as part of a health report.

[0038] In some aspects, the technology described herein relates to a system comprising: a wearable monitoring device capable of being worn by a user to detect one or more measurements of the user during an observation period, the one or more measurements including accelerometer measurements and potential measurements of the user's heart; and a computing device configured to: receive the one or more measurements from the wearable monitoring device; identify periods of exercise performed by the user during the observation period based on the accelerometer measurements; process the potential measurements using a machine learning model trained to associate patterns in the potential measurements with heart rhythm classifications; generate arrhythmia correlations for periods before, during, and after the exercise period based on the contemporaneous heart rhythm classifications; and generate a prediction of cardiac health status based on the arrhythmia correlations.

[0039] In some respects, the technology described herein relates to a system in which accelerometer measurements are acquired at a sampling rate of less than 5 Hz, and in which the observation period is between 7 and 14 days.

[0040] In some respects, the technology described herein relates to a system in which a computing device is further configured to: generate an expected heart rate response for a given exercise period based on user group data; generate an actual heart rate response of a user for a given exercise period based on potential measurements acquired during the given exercise period; and indicate chronotropic insufficiency based on the actual heart rate response deviating from the expected heart rate response by at least a threshold amount.

[0041] In some respects, the technology described herein relates to a system in which, in order to identify motion periods based on accelerometer measurements, a computing device is further configured to: detect peak values ​​in the accelerometer measurements that are above a measurement threshold as steps; and identify motion periods based on the detected steps exceeding a step threshold during a predetermined time period.

[0042] In some respects, the technology described herein relates to a system in which a computing device is further configured to: detect a user’s sleep periods based on accelerometer measurements; generate sleep stage classifications based on accelerometer measurements during the detected sleep periods; and generate additional arrhythmia correlations based on the sleep stage classifications and contemporaneous heart rhythm classifications.

[0043] Figure 1This is a block diagram of a non-limiting example 100 of an environment operable to employ the sleep and activity-related cardiac monitoring described herein. Example 100 includes a person 102 depicted wearing a monitoring device 104. The environment also includes an analytics platform 106. The analytics platform 106 may be directly connected to the monitoring device 104 via one or more wireless connections, or via one or more wired and / or wireless connections and one or more intermediate devices, such as computing devices, network routing devices and apparatuses, server devices, and / or the Internet associated with the person 102, to name just a few.

[0044] Monitoring device 104 can be used to monitor one or more aspects of person 102, such as generating measurements 108. For example, in some cases, monitoring device 104 can be provided to record the electrical activity of person 102's heart over an observation period, such as several seconds or minutes, or several days. For example, person 102 may have the amplitude of his or her heart potential monitored over time to generate one or more electrocardiograms (ECGs), which can be used to predict various events. In one or more embodiments, alternatively or additionally, monitoring device 104 can be provided to record accelerometer data over an observation period, such as acquiring motion and activity data for sleep / wake detection, activity level classification, and correlation with cardiac events. In one or more embodiments, accelerometer data can be sampled at a low sampling rate (e.g., in the range between 1-5 Hz, such as 1.6 Hz) to extend battery life while maintaining accurate detection of sleep and activity patterns. The monitoring device 104 can output measurement values ​​108 (e.g., time series of measurement values, such as potential measurements, acceleration measurements, and / or other types of physical and / or physiological measurements), which can indicate observations or be used to generate predictions for one or more events.

[0045] In conjunction with the monitoring device, instructions can be provided to person 102 instructing them on how to operate the monitoring device 104 and / or how to act while wearing the monitoring device 104 (e.g., sleep, engage in activities), examples of which will be described in detail herein. In one or more embodiments, the instructions may be provided as part of a kit, such as as written instructions. Alternatively or additionally, the analysis platform 106 may transmit and output the instructions via a computing device associated with person 102 (e.g., for display and / or audio output). In one or more embodiments, the analysis platform 106 may provide the instructions for output only after the person 102 has worn the monitoring device 104 for a predetermined amount of time during the observation period (e.g., two days) and / or based on patterns in various aspects of the person 102.

[0046] Monitoring device 104 can be configured in various ways to monitor one or more aspects of personnel 102. Furthermore, Figure 1 and Figure 2 The depicted external dimensions are merely an example, and the external dimensions of monitoring device 104 may vary in variations. It should be understood that monitoring device 104 may be configured with one or more sensors, examples of which include one or more of the following: multiple electrodes (e.g., capable of being placed on the skin of person 102), accelerometers, and pulse oximeters (e.g., to measure and record blood oxygen saturation (SpO2) and / or generate photoplethysmography of person 102), to name just a few. Of course, monitoring device 104 may be configured with various types of sensors without departing from the described technology.

[0047] While the monitoring device 104 may be configured in a manner similar to that used for clinical patient monitoring, in one or more embodiments, the monitoring device 104 may be configured differently from devices used for clinical patient monitoring and / or diagnosis. For example, and not limitingly, the monitoring device 104 may be configured as a ring, watch, patch, and / or band, to name just a few form factors. Alternatively or additionally, the monitoring device 104 may have a similar form factor to that in a clinical setting but may have different functions, such as preventing the wearer from viewing the measurement value 108.

[0048] As used herein, the term “continuous” in conjunction with any signal associated with monitoring and person 102 (e.g., acceleration data and / or the electrical activity of person 102’s heart) can refer to the ability of a device to generate measurements substantially continuously, such that the device can be configured to generate measurements 108 at time intervals (e.g., every hour, every 30 minutes, every 5 minutes, every minute, every 30 seconds, every second, every half second, etc.), in response to events (e.g., an inflection point in an electrical signal, such as a peak or trough). Without departing from the spirit or scope of the art described, the functionality of the monitoring device 104 in generating measurements 108 along with other measurements and / or recording any of a variety of signals can vary.

[0049] In one or more embodiments, the monitoring device 104 may be configured to unload the measurement value 108 during the observation period. For example, the monitoring device 104 may unload the measurement value 108 by transmitting it to an external computing device via a wired or wireless connection, for example, at predetermined time intervals and / or in response to establishing or re-establishing a connection with the computing device. In one or more embodiments, the measurement value 108 and / or other data from the monitoring device 104 may be compressed by the monitoring device 104 for wireless transmission, for example, using one or more of a variety of data compression techniques. Compressing sensor data in this way can reduce the battery consumption of the monitoring device 104 during the observation period and facilitate wearing the monitoring device while assessing sleep patterns, activity levels, arrhythmias, and / or cardiac health. For example, the monitoring device 104 may lightly process the measurement value 108 during wear, for example, by averaging or encoding at least a portion of the measurement value 108, to reduce the amount of data stored and / or transmitted while retaining information for post-wear analysis. In some of these implementations, more comprehensive data processing can be performed after the wearing period to save battery power and reduce memory storage usage.

[0050] In one or more embodiments, the monitoring device 104 may also implement other power management strategies, wherein data from one type of sensor is used to selectively trigger measurement captures by other sensors. As an illustrative example, additionally or alternatively, accelerometer data indicating sleep / wake transitions or specific activity levels may trigger enhanced cardiac monitoring data during periods of interest, enabling analysis of whether arrhythmias such as atrial fibrillation, cardiac conduction block, or ventricular arrhythmias are associated with sleep stages, activity levels, and / or exercise intensity. For example, this selective triggering method can achieve dynamic cardiac stress testing by associating exercise-related activity patterns with contemporaneous cardiac rhythm changes. In one or more embodiments, the monitoring device 104 may implement sleep / activity-triggered analysis and / or data acquisition, wherein detected sleep stages may trigger different cardiac monitoring modes, such as reducing the sampling rate during deep sleep periods or enhancing arrhythmia detection during rapid eye movement (REM) sleep stages. Similarly, detected activity levels may trigger adaptive data acquisition strategies, such as increasing cardiac monitoring resolution during exercise periods or modifying the analysis algorithm during sedentary periods. This sleep / activity triggering method optimizes battery usage while ensuring comprehensive data capture during physiologically relevant periods. Additional details about the sleep / activity triggering method are described below.

[0051] In cases where monitoring device 104 can be configured to store measurements 108 for the entire observation period, in one or more embodiments, monitoring device 104 can be configured without wireless transmission means, such as any antenna for wirelessly transmitting measurements 108, and without hardware or firmware for generating data packets for such wireless transmission. Instead, monitoring device 104 can be configured with hardware for transmitting measurements 108 via physical wired coupling. In such cases, monitoring device 104 can be "plugged in" to retrieve measurements 108 from the device's storage.

[0052] Therefore, monitoring device 104 may be configured with one or more ports to transmit measurement values ​​108 via a wired connection to an external computing device. Examples of such physical coupling may include a micro universal serial bus (USB) connection, a mini USB connection, and a USB-C connection, to name just a few. Although monitoring device 104 may be configured to retrieve measurement values ​​108 via a wired connection as discussed above, alternatively or additionally, in different situations, monitoring device 104 may be configured to offload measurement values ​​108 via one or more wireless connections.

[0053] Once the monitoring device 104 generates a measurement value 108, it provides the measurement value 108 to the analysis platform 106. As described above, the measurement value 108 can be transmitted to the analysis platform 106 via a wired and / or wireless connection.

[0054] For example, where the analysis platform 106 is partially or fully implemented on the monitoring device 104, the measured value 108 can be transferred from the device's local storage to the device's processing system via a bus. Where the monitoring device 104 is configured to generate one or more predictions 110 by processing the measured value 108, the monitoring device 104 can also be configured to provide the generated one or more predictions 110 as output, for example, by transmitting the one or more predictions 110 to an external computing device. In other cases, the measured value 108 can be processed by an external computing device configured to generate one or more predictions 110. For example, the measured value 108 can be processed by a smartphone associated with a user, a smartphone or other dedicated device associated with the monitoring device 104, and / or one or more server computers in a data center or other location, which can be used by an entity associated with the monitoring device 104, etc. In other words, these other devices can implement at least a portion of the analysis platform 106 and / or the prediction system 114.

[0055] In one or more embodiments, the monitoring device 104 is configured to transmit measurement values ​​108 to an external device via a wired connection (e.g., via USB-C or some other physical communication coupling). Here, a connector can be inserted into the monitoring device 104, or the monitoring device 104 can be inserted into a device having a socket having corresponding contacts that engage with the monitoring device 104. Measurement values ​​108 can be retrieved from the storage device of the monitoring device 104 via this wired connection, for example, by transmitting the measurement values ​​to the external device via the wired connection. Such a connection can be used when personnel 102 mail the monitoring device 104 to, for example, a healthcare provider, a telemedicine provider, the provider of the monitoring device 104, or a medical testing laboratory after the observation period.

[0056] Alternatively or additionally, monitoring device 104 may provide measurement value 108 to analysis platform 106 by transmitting measurement value 108 via one or more wireless connections. For example, monitoring device 104 may wirelessly transmit measurement value 108 to external computing devices, such as mobile phones, tablets, laptops, smartwatches, other wearable health trackers, etc. Therefore, monitoring device 104 may be configured to communicate with external devices using one or more wireless communication protocols or technologies. For example, monitoring device 104 may communicate with external devices using one or more of Bluetooth (e.g., Bluetooth Low Energy link), Near Field Communication (NFC), Long Term Evolution (LTE) standards such as 5G, etc. In the case of transmitting measurement value 108 to external devices for processing, monitoring device 104 may be configured with corresponding antennas and other wireless transmitting means. In these cases, measurement value 108 may be transmitted to analysis platform 106 in various ways, such as at predetermined time intervals (e.g., daily, hourly, or every five minutes), in response to the occurrence of an event (e.g., filling the storage buffer of monitoring device 104), or in response to the end of an observation period, to name a few.

[0057] Therefore, regardless of where the analysis platform 106 is implemented (e.g., at monitoring device 104, at a smartphone associated with person 102, or at a server device), the analysis platform 106 acquires the measurements 108 generated by monitoring device 104. In one or more embodiments, the analysis platform 106 also acquires additional measurements generated by monitoring device 104 and / or any other device used during the observation period (e.g., a smartwatch, chest strap, etc.). As mentioned above, examples of such additional measurements include, but are not limited to, blood oxygen saturation (SpO2) measurements. In one or more embodiments, the combination of accelerometer data and cardiac measurements can enable enhanced arrhythmia correlation capabilities, including the ability to correlate arrhythmias such as atrial fibrillation, ventricular arrhythmias, or bradycardia episodes with contemporaneous sleep stages and / or activity levels. In one or more embodiments, the analysis performed by the analysis platform 106 may include correlation analysis between sleep / wake patterns and arrhythmias to identify patterns in which specific sleep stages may make a user more susceptible to arrhythmias or activity levels may trigger arrhythmias. Additionally or alternatively, the analysis platform 106 can correlate exercise intensity with cardiac events, thereby identifying which activity intensities may induce arrhythmias or other physiological responses and supporting dynamic cardiac stress testing capabilities. Sleep and activity background can provide analytical results that are not available through separate analyses, such as determining the optimal exercise intensity for a specific user or patient group while monitoring exercise-related arrhythmias.

[0058] In one or more embodiments, the analysis platform 106 may be implemented wholly or partially at the monitoring device 104. Alternatively or additionally, the analysis platform 106 may be implemented wholly or partially using one or more computing devices external to the monitoring device 104, such as one or more computing devices associated with person 102 (e.g., mobile phone, tablet, laptop, desktop computer, or smartwatch), or one or more computing devices associated with a service provider (e.g., healthcare provider, telemedicine provider, service provider corresponding to the provider of monitoring device 104, medical testing laboratory service provider, etc.). In the latter case, the analysis platform 106 may be implemented at least partially on one or more server devices.

[0059] In the illustrated example 100, the analytics platform includes a storage device 112. According to the technology, the storage device 112 can be configured to store measurement values ​​108 and / or other measurement values ​​or information processed by the prediction system 114 to generate one or more predictions 110. The storage device 112 can represent one or more databases and / or other types of storage devices capable of storing measurement values ​​108 and / or other types of measurement values. The storage device 112 can also store various other data, such as personal information, demographic information describing the person 102, information about healthcare providers, information about insurance providers, payment information, prescription information, determined health indicators, account information (e.g., username and password), etc. The storage device 112 can also store data of other users within the user group.

[0060] In one or more embodiments, storage device 112 may also store sleep / wake detection parameters, activity thresholds, arrhythmia classification algorithms, sleep and activity correlation algorithms, historical sleep and activity patterns, baseline accelerometer data for comparison purposes, body angle calculation parameters, and multimodal data fusion algorithms that determine how to combine accelerometer data with other physiological measurements such as ECG or SpO2 data to improve prediction accuracy. One or more algorithms may be, or may include, machine learning algorithms. For example, at least one machine learning algorithm may be trained to identify the temporal relationship between sleep stages and / or activity levels and cardiac events. In one or more embodiments, storage device 112 may also store arrhythmia classification models, a sleep-arrhythmia correlation database, cardiac event time points relative to sleep and activity patterns, exercise-related arrhythmia detection algorithms, and / or algorithms for identifying causal relationships between specific sleep stages and / or activity intensity and arrhythmia types. Storage device 112 can also store sleep / activity trigger parameters that define how different sleep stages and activity levels should trigger specific analysis modes, data acquisition strategies, and processing algorithms, thereby enabling adaptive monitoring based on physiological background.

[0061] In the illustrated example 100, the analysis platform 106 also includes a prediction system 114. The prediction system 114 represents the functionality of processing measurements 108 to generate one or more predictions 110. Alternatively or additionally, the prediction system 114 may output one or more time series indicating observations or predictions of one or more events over time. It should also be understood that, in variations, the prediction system 114 may output different combinations of multiple predictions.

[0062] In at least one embodiment, prediction system 114 uses machine learning to generate at least a portion of one or more predictions 110. For example, and not limitingly, prediction system 114 may include one or more neural networks trained on historical measurements and historical outcome data of a user group. Prediction system 114 may include one or more machine learning models (e.g., an ensemble of models). Alternatively or additionally, prediction system 114 may include logic for preprocessing measurements 108 (machine learning models and / or other types of logic), such as extracting various cardiovascular features, motion features, and / or other features from the measurement sequence. One or more predictions 110 may, for example, be the output of prediction system 114.

[0063] In various examples, prediction system 114 may represent and / or may include activity detection system 116 and cardiac monitoring system 118, and one or more predictions 110 may include and / or may represent sleep stage classification 120, activity level classification 122, heart rhythm classification 124, arrhythmia correlation 126, and / or cardiac health status prediction 128. For example, as described further in more detail below, activity detection system 116 may operate to implement accelerometer-based techniques to generate sleep stage classification 120 and activity level classification 122. Activity detection system 116 may receive accelerometer data and detect peaks above a threshold to identify steps, analyze movement patterns, analyze body angles, and / or analyze activity intensity to classify sleep stages and wakefulness as part of sleep stage classification 120, and classify activity levels within a predetermined time period as part of activity level classification 122. For example, activity level classification 122 may categorize activity levels as sedentary, mild, moderate, or vigorous based on steps within a predetermined time period and / or accelerometer features and movement patterns extracted from accelerometer data. For example, the activity detection system 116 can incorporate time-domain and frequency-domain features of accelerometer data to enhance activity classification.

[0064] The cardiac monitoring system 118 can operate to analyze cardiac data patterns and correlate these patterns with sleep and activity data to generate a heart rhythm classification 124 and an arrhythmia correlation 126, which can indicate the time point and context of cardiac events, including the correlation between cardiac events and sleep stages and / or activity levels. In one or more embodiments, the cardiac monitoring system 118 can utilize cardiac data (e.g., ECG waveform data) from measurement 108 to identify heart rhythm types and correlate these types with contemporaneous accelerometer data in measurement 108, thereby providing context for arrhythmia diagnosis and patient care. The cardiac monitoring system 118 can employ one or more algorithms (e.g., machine learning algorithms and / or conventionally programmed algorithms) to analyze cardiac data, classifying heart rhythms and identifying arrhythmias by considering factors such as heart rate variability, heart rhythm patterns, and morphological characteristics. The cardiac monitoring system 118 may also employ one or more algorithms (e.g., machine learning algorithms and / or conventionally programmed algorithms) to analyze sleep and activity data (e.g., accelerometer data) of measurement 108 to provide contextual information about cardiac events, thereby obtaining arrhythmia correlation 126. Alternatively or additionally, the prediction system 114 may combine the analyses performed by the activity detection system 116 and the cardiac monitoring system 118 to output arrhythmia correlation 126 and / or cardiac health status prediction 128.

[0065] In one or more embodiments, the prediction system 114 can enable sleep / activity-triggered analysis and / or data acquisition, where different sleep stages and activity levels trigger adaptive monitoring strategies. For example, when activity is detected by the activity detection system 116, the monitoring device 104 can activate additional data acquisition channels, such as gyroscopes, which are more costly in terms of battery consumption but can provide additional analytical results about the type or intensity of the activity performed. When sleep is detected by the activity detection system 116, more power-intensive data acquisition can be triggered, such as LED-based photoplethysmography (PPG) for SpO2, which can be used for sleep apnea screening and diagnosis. When the person 102 is no longer asleep, the LED-based PPG can be turned off. As another example, the cardiac monitoring system 118 can also utilize sleep and activity detection to select baseline events or planned events that provide relatively clean data measurements (e.g., ECG bands) for analysis. Similarly, detecting noisy activity allows the prediction system 114 to label noisy ECG data for exclusion from heart rhythm analysis, thereby improving the accuracy of heart rhythm classification 124 and arrhythmia correlation 126. Detecting sleep and activity while wearing the monitoring device 104 also enables the reporting of additional background information, such as nocturnal heart rate, nocturnal heart rate variability, and other health conditions, during the observation period.

[0066] In one or more embodiments, the cardiac monitoring system 118 may be configured to identify specific arrhythmia types and correlate these specific arrhythmia types with contemporaneous sleep or activity states, which may be output as arrhythmia correlation 126. For example, arrhythmia correlation 126 may indicate bradycardia episodes occurring during sleep, atrial fibrillation episodes associated with specific activity levels, and / or ventricular arrhythmias that may be triggered by exercise intensity or sleep transitions, to name just a few non-limiting examples. The cardiac monitoring system 118 may also analyze whether sleep stage transitions themselves trigger arrhythmias, whether specific activity intensities induce arrhythmias, and / or the correlation between sleep quality and heart rhythm stability. The cardiac monitoring system 118 may also correlate exercise intensity with arrhythmias to enable dynamic cardiac stress testing, thereby providing analytical results regarding exercise-related cardiac risk patterns that cannot be obtained by separately analyzing cardiac monitoring and activity tracking. By combining the activity detection system 116 with the cardiac monitoring system 118, the prediction system 114 can reveal the relationship between cardiac events and physiological states, thereby providing healthcare providers with a comprehensive tool to assess patients’ cardiac health in the context of daily activities and sleep patterns.

[0067] For example, Heart Health Prediction 128 can provide comprehensive physiological analysis by combining sleep and activity analysis with a cardiac context, enabling the detection of relationships between sleep patterns, activity levels, heart health measures, and overall health. Heart Health Prediction 128 can include correlations between sleep quality and heart health indicators, relationships between daily activity levels and cardiac function, and patterns that indicate cardiovascular fitness, exercise tolerance, or potential health risks. For instance, Heart Health Prediction 128 can indicate the relationship between sleep duration and heart rate variability, the correlation between activity intensity and arrhythmia frequency, or the association between sleep / wake patterns and overall cardiovascular health. Heart Health Prediction 128 can also provide analysis of how different sleep stages and activity levels affect cardiac function over time, including information on exercise capacity, recovery patterns, and the impact of sleep quality on heart health. By combining sleep and activity data from an accelerometer with contemporaneous cardiac measurements, Heart Health Prediction 128 can identify optimal sleep and activity patterns for individual users, indicate early signs of declining cardiovascular function, and / or provide personalized recommendations for maintaining or improving overall heart health.

[0068] Alternatively or additionally, cardiac health prediction 128 may include dynamic cardiac stress test results, wherein the activity detection system 116 detects exercise periods based on accelerometer data, and the cardiac monitoring system 118 analyzes the cardiac response during the same period based on potential data to identify exercise-induced arrhythmias and cardiac recovery patterns that may indicate underlying coronary artery disease or other cardiovascular conditions. Cardiac health prediction 128 may include assessing chronotropic dysfunction by comparing the expected and actual heart rate responses during detected activity events, thereby providing analytical results regarding cardiac function during physical exertion. Cardiac health prediction 128 may also generate exercise prescription recommendations and stress test analysis results based on observed relationships between specific activity intensities and cardiac health measures, thereby providing personalized exercise guidance that promotes cardiovascular benefits while minimizing the risk of arrhythmias for individual users or patient groups. This multimodal approach can provide clinical benefits by supporting preventative cardiac care strategies; enabling early intervention when certain patterns are detected in the combined sleep, activity, and cardiac data; and facilitating dynamic cardiac stress testing by correlating exercise intensity with cardiac response patterns without requiring a visit to the hospital.

[0069] Example 100 also illustrates accessory device 130 and healthcare provider 132. For example, accessory device 130 may include one or more devices associated with person 102 and / or monitoring device 104, such as those described above. For example, accessory device 130 may include display devices (e.g., smartphones and / or personal display devices) to display one or more predictions 110 and / or control functions of monitoring device 104. Accessory device 130 may also display notifications 134 and / or instructions 136, as will be further detailed below.

[0070] For example, healthcare provider 132 may represent one or more additional processing devices associated with an authorized healthcare system, such as healthcare worker devices, electronic health record systems, diagnostic imaging equipment, laboratory information systems, telemedicine platforms, clinical decision support tools, etc. In various embodiments, notification 134 may be generated based on one or more predictions 110, such as alerts for detected arrhythmias, changes in sleep patterns, changes in activity levels, changes in cardiac health status prediction 128, exercise-related cardiac events, and / or other detected events, and notification 134 may be displayed by accessory device 130 to deliver information to personnel 102, healthcare providers, caregivers, and / or emergency contacts. As a non-limiting example, prediction system 114 may generate reports and / or alerts based on arrhythmia correlation 126, which can lead to more targeted cardiac interventions and improved cardiac care strategies.

[0071] In conjunction with monitoring device 104, instructions 136 can be provided to person 102 to instruct them on how to operate monitoring device 104 and / or tasks to perform while wearing monitoring device 104 (e.g., sleeping, resting, or engaging in activity). Instructions 136 can help ensure sufficient data is collected for use in cardiac stress testing by specifying a particular duration and / or intensity of physical activity and / or recovery (e.g., supine recovery). In one or more embodiments, clinical workflows can be modified to collect motion data while person 102 wears monitoring device 104, rather than (or except) having person 102 perform cardiac stress testing alone in a clinic using 12-lead ECG cardiac monitoring. For example, instructions 136 could provide specific exercise protocols and recovery methods during wearing monitoring device 104 to collect specific data for cardiac stress testing.

[0072] In various examples, one or more operations of the analysis platform 106, the prediction system 114, the activity detection system 116, and / or the cardiac monitoring system 118 can be performed by one or more of the monitoring device 104, the accessory device 130, the device and system of the healthcare provider 132, and / or one or more additional devices not shown.

[0073] Figure 2 A non-limiting example 200 of a monitoring device is depicted. The example 200 depicts a monitoring device 104.

[0074] According to the described technology, the monitoring device 104 includes one or more sensors 202, examples of which include, but are not limited to, a pair or more pairs of electrodes, an accelerometer, a pulse oximeter, and a sweat sensor, to name just a few. The monitoring device 104 may also include a transmitter 204. In this example 200, the monitoring device 104 also includes one or more adhesive portions 206. In operation, the monitoring device 104 is configured to be attached to the skin via one or more adhesive portions 206, such that, for example, one or more sensors 202 are positioned to detect and record the cardiac electrical activity of the person 102, for example, generating an electrocardiogram (ECG and / or EKG). In at least one embodiment, the monitoring device 104 can be removed by peeling one or more adhesive portions 206 from the skin.

[0075] It should be understood that the monitoring device 104 and its components are only one form factor, and the monitoring device 104 and its components may have different form factors without departing from the spirit or scope of the technology.

[0076] In one or more embodiments, monitoring device 104 may include a processor and / or memory (not shown). By utilizing the processor, monitoring device 104 can generate measurements 108 based on communication with one or more sensors 202 that indicate aspects of person 102, such as the person's cardiac electrical activity. In one or more embodiments, the processor also generates one or more communicable data packets including measurements 108 and / or one or more other measurements such as blood oxygen saturation (SpO2) measurements. Alternatively or additionally, the processor generates and / or stores other data that can be used to predict the classification of physiological conditions, such as arrhythmias, activity levels, etc.

[0077] In embodiments where monitoring device 104 is configured for wireless transmission, transmitter 204 may wirelessly transmit measurement value 108 as a data stream to a computing device. For example, in one or more embodiments, monitoring device 104 is configured to transmit (e.g., transmit and / or receive) information (e.g., potential measurement value) via Bluetooth Low Energy (BLE) connection. Alternatively or additionally, monitoring device 104 may buffer measurement value 108 (e.g., buffer in memory) and have transmitter 204 later transmit the buffered measurement value at various intervals (e.g., time intervals (per second, every thirty seconds, per minute, every five minutes, per hour, etc.), storage intervals (when the buffered measurement value reaches a threshold data amount), etc.).

[0078] Sleep and activity-related cardiac monitoring systems

[0079] Figure 3 A non-limiting system in an example implementation 300 of cardiac monitoring with sleep and activity correlation is depicted, which is shown in more detail. Figure 1 The operation of the prediction system 114.

[0080] In this example, the prediction system 114 first receives sensor data 302, which may include electrical data 304 (e.g., potential measurements and / or ECG data), accelerometer data 306, SpO2 data 308 (e.g., blood oxygen saturation data), and / or various supplementary data 310. In various examples, sensor data 302 is acquired by one or more devices and / or sensors (e.g., wearable monitoring device 104). Sensor data 302 may include time-ordered data instances, such as continuous data, data acquired at predetermined intervals (e.g., every half-second, every minute, every five-minute, etc.) over the duration of an observation period, such as a day, multiple days of a week, a month, etc. For example, sensor data 302 may be a combination of... Figure 1The measurement value 108 is described. In one or more embodiments, accelerometer data 306 is acquired at a sampling rate of less than 5 Hz (e.g., 1.6 or 1.56 Hz). Typically, sensor data 302 is processed by prediction system 114 to generate one or more predictions 110 according to techniques described in more detail below, which may include one or more of sleep stage classification 120, activity level classification 122, heart rhythm classification 124, arrhythmia correlation 126, and heart health status prediction 128.

[0081] For example, prediction system 114 includes a training module 312 operable to train a machine learning model 316 using training data 314 to perform one or more sleep stage classification tasks, activity classification tasks, and / or cardiac monitoring tasks. Sleep stage classification tasks, activity classification tasks, and cardiac monitoring tasks involve, for example, generating one or more predictions 110, such as heart rate classification 124 based on patterns in electrical data 304, associating sleep stages and activity patterns with cardiac data from electrical data 304 to obtain arrhythmia correlation 126 and cardiac health status prediction 128, and monitoring body position and sleep stages based on accelerometer data 306 to obtain sleep stage classification 120. These tasks may include: classifying sleep stages (such as light sleep, deep sleep, REM sleep, and wakefulness); classifying activity levels (such as sedentary, mild, moderate, or vigorous); detecting arrhythmias; and associating arrhythmias with sleep stage and activity data (e.g., based on accelerometer data 306). For example, by identifying arrhythmias that may be associated with a specific sleep stage or activity level based on electrical data 304, it is possible to analyze the relationship between sleep stage patterns, exercise intensity, and the occurrence of arrhythmias. Therefore, machine learning model 316 is trained to associate patterns in sensor data 302 (e.g., various potential measurements from electrical data 304 and / or accelerometer measurements and body angle calculations from accelerometer data 306) with one or more predictions 110. It should be understood that more than one machine learning model 316 can be trained separately, for example, separately for one or more different predictions 110.

[0082] The foregoing examples describe various instances of artificial intelligence (“AI”) models and / or machine learning models. In one or more examples, an AI model (e.g., a machine learning model) refers to a computer representation capable of being adjusted (e.g., through training and retraining) based on inputs without being actively programmed by the user to automatically and without user intervention approximate an unknown function. For example, the term “machine learning model” includes models that utilize algorithms to learn from and make predictions about known data by analyzing training data for learning and relearning, thereby generating outputs that reflect patterns and attributes of the training data. In the context of cardiac monitoring with sleep stages and activity correlations, a machine learning model can be implemented (e.g., implemented by one or more processing devices of prediction system 114) to analyze accelerometer data patterns and correlate these patterns with cardiac data to identify sleep stages, activity levels, detect arrhythmias, and generate comprehensive physiological analysis results.

[0083] The training module 312 is also capable of operating to initialize various parameters for the machine learning model 316, which can use these parameters as internal variables to represent and process information during training. These parameters can also be used to represent inferences acquired through training. In one or more embodiments, the training data 314 is divided into multiple batches to improve the efficiency of parameter processing and optimization of the machine learning model 316 during training, which is particularly beneficial for model accuracy when processing large amounts of accelerometer time-series data acquired at low sampling rates and correlating sleep stages and activity patterns with contemporaneous cardiac measurements.

[0084] In this example, training data 314 may include historical electrical potential measurements (e.g., such as ECG) and accelerometer measurements from a user group, along with corresponding historical outcome data, such as arrhythmia classification, activity classification, sleep stage classification, and / or overall cardiac health status classification. This data may be collected from clinical studies, sleep monitoring studies, cardiac monitoring studies, clinical cardiac stress tests, or other sources that simultaneously record electrical data 304 and accelerometer data 306. In some cases, training data 314 may also include additional physiological measurements such as blood oxygen saturation levels (e.g., SpO2 data 308) and / or additional relevant biomarkers. Training data 314 can be labeled with various sleep stages, activity, and cardiac-related information, such as the presence of sleep periods, activity types (e.g., walking, running, sleeping), activity levels, sleep stages (e.g., light sleep, deep sleep, REM sleep, and wakefulness), arrhythmias, body positions, etc.

[0085] Training data 314 can also be labeled with various heart rate-related information. That is, to train a machine learning model for cardiac monitoring with sleep stages and activity correlations, training data 314 can provide examples of what the machine learning model "will learn," such as a basis for learning patterns from data. For cardiac monitoring applications, training data 314 can include labeled datasets of accelerometer measurements and cardiac data from users known to have abnormal sleep stage patterns, activity levels, and heart conditions, as well as measurements from individuals with normal sleep stage patterns, activity levels, and no cardiac abnormalities. For example, training module 312 can acquire and preprocess training data 314, which includes input features (e.g., accelerometer waveforms, body angle calculations, sleep stage patterns, activity patterns, ECG data, heart rate patterns) and corresponding target labels (e.g., "deep sleep stage," "vigorous activity," "atrial fibrillation detected," "exercise-related arrhythmia," or specific sleep stage classification, activity classification, and cardiac classification). Training data can also be labeled with physiological features such as normal sinus rhythm, atrial fibrillation episodes, ventricular arrhythmias, bradycardia events, cardiac conduction block, premature ventricular contractions, supraventricular tachycardia, and / or other arrhythmias that may be temporally associated with sleep stages or changes in activity levels.314

[0086] Training data 314 can be received as input by a machine learning model 316 and used as the basis for generating predictions based on the current state of the model's layers and corresponding nodes, with the results output as output data. For example, the machine learning model can be configured using multiple layers, each with multiple nodes. These multiple layers can be configured to include an input layer, an output layer, and one or more hidden layers. In the context of cardiac monitoring with sleep stages and activity correlations, the input layer can receive sensor data 302 or its features, including amplitude calculations, body angle measurements, step count peaks, movement patterns, activity level indices, and cardiac features. The hidden layer processes these inputs, for example, through weighted connections, to identify complex patterns indicating sleep stages, activity levels, arrhythmias, and correlations between sleep stages / activity and cardiac status, patterns that cannot be detected by conventional analysis modalities relying solely on basic cardiac monitoring or activity tracking. The output layer can generate one or more predictions 110, including sleep stage classification 120, activity level classification 122, heart rate classification 124, arrhythmia correlation 126, and / or cardiac health status prediction 128. Nodes within a layer perform computations via a weighted connection system “learned” during the training of the machine learning model 316 through hidden states, enabling various sleep stage detection tasks, activity detection tasks, and heart monitoring tasks.

[0087] In some implementations, different training schemes and / or model architectures are employed based on what one or more predictions 110 are to include. For example, the composition and structure of training data 314 can vary depending on the specific type of prediction to be generated. In an example where one or more predictions 110 are to indicate a binary classification of sleep / wake state (e.g., whether the user is asleep or awake), training data 314 can be labeled with a yes / no index. In an additional or alternative example where one or more predictions 110 are to include fine-grained predictions such as specific sleep stages (light sleep, deep sleep, REM sleep), activity levels, and / or arrhythmia types, training data 314 includes detailed annotations related to the fine-grained predictions. In an example where one or more predictions 110 are to indicate the correlation between sleep stages / activity patterns from accelerometer data 306 and cardiac health measurements from electrical data 304, training data 314 can include health status analysis results associated with various sleep stages, activity levels, and cardiac measurements. In one or more additional or alternative examples where prediction 110 is to include a comprehensive analysis such as dynamic cardiac stress testing using motion detection and cardiac response monitoring, sleep stage-arrhythmia correlation, and / or motion-related arrhythmia detection combined with physiological background from electrical data 304, training data 314 may include multimodal input data to capture the complex relationships between sleep stages, activity, cardiac status, and physiological outcomes.

[0088] In some examples, training data 314 is constructed to support multi-task learning, wherein machine learning model 316 can simultaneously predict multiple aspects of sleep stage classification, activity classification, and / or cardiac monitoring, such as predicting combinations of sleep stages and activity levels, as well as sleep stage-cardiac correlation and arrhythmia detection with physiological context, such as cardiac health status analysis results based on sleep stage-heart rate relationships and / or exercise-related arrhythmia patterns based on combinations of activity-cardiac data. In additional or alternative examples, training module 312 trains machine learning model 316 task-by-task, for example, implementing a first round of training to train machine learning model 316 to perform sleep stage-cardiac correlation analysis, and implementing a second round of training to train machine learning model 316 to perform exercise-related arrhythmia detection with activity contextualization. In this way, the technology described herein enables targeted training of the machine learning model 316 for specific tasks, which improves the performance and efficiency of the model in performing individual aspects such as sleep stage detection, activity classification, and cardiac monitoring, as well as the performance and efficiency of the model in performing complex multimodal physiological analysis results that combine accelerometer data 306 with electrical data 304, rather than just basic sleep stages or cardiac monitoring.

[0089] In one or more implementations, the training module 312 trains the machine learning model 316 using an iterative process that adjusts weights and learning parameters to minimize the loss function. For example, the training module 312 may use backpropagation and / or gradient descent algorithms to update the model's parameters based on the differences between predicted sleep / wake states, activity classifications, and arrhythmia classifications in the training data 314 and the actual sleep / wake states, activity classifications, and arrhythmia classifications. The learning rate, batch size, and / or number of epochs may be adjusted to optimize the performance of the machine learning model 316.

[0090] Training the machine learning model 316 may include computing a loss function to quantify the loss associated with the operations performed by the nodes of the machine learning model 316. The loss function can be configured in various ways to control the operation and / or functionality of the machine learning model 316. For example, the loss function may be designed to prioritize the accuracy of arrhythmia detection while minimizing false positives and optimizing the accuracy of the correlation between sleep stage / activity patterns and cardiac health metrics. Computing the loss function may include, for example, comparing the differences between predictions specified in the output data (e.g., predicted sleep stages, activity levels, arrhythmias, or sleep stage-cardiac correlations) and target labels specified in the training data 314 (e.g., clinically confirmed sleep stage classifications, activity classifications, and recorded arrhythmias). Examples of the various configuration options for the loss function include quadratic loss functions as part of least-squares techniques for continuous cardiac parameters, cross-entropy loss for classification tasks such as sleep stage classification, and custom loss functions incorporating clinical risk factors specific to arrhythmia detection and sleep stage-cardiac correlations.

[0091] Furthermore, various architectures / types of the machine learning model 316 are considered. In one or more implementations, the machine learning model 316 may include neural networks, such as convolutional neural networks (CNNs), recurrent neural networks (RNNs), or combinations thereof. In some instances, the machine learning model 316 incorporates one or more U-Net and / or ResNet architectures, features, or components. The model can also be implemented as an ensemble of different algorithms that combine one or more decision trees, random forests, and / or gradient boosters with neural network methods. For example, CNNs can be used to analyze accelerometer waveform patterns and detect sleep / wake features. As another example, long short-term memory (LSTM) neural networks can be used to analyze sleep and activity patterns over time and correlate activity sequences with cardiac events. In other examples, generative adversarial networks (GANs), decision trees (e.g., for sleep / wake classification and arrhythmia assessment), support vector machines, linear regression, logistic regression for binary sleep / wake detection, Bayesian networks, random forest learning for feature importance in the correlation between accelerometer and cardiac data, dimensionality reduction algorithms, boosting algorithms, deep learning neural networks, etc., can be employed. It should be understood that the above examples are given by way of illustration, and other configurations may be used without departing from the spirit or scope of the technology described.

[0092] In some examples, training module 312 may employ techniques such as dropout or regularization to prevent overfitting. The training process can continue until the model reaches the desired accuracy level on the validation dataset and / or until a predetermined number of iterations are completed. This method enables machine learning model 316 to learn complex patterns in sensor data 302, indicating relationships between sleep / activity patterns from accelerometer data 306 and cardiac health measures from electrical data 304, correlations between sleep stage changes during sleep transitions and heart rate responses, temporal associations between exercise intensity and arrhythmias, and multimodal analysis results that combine sleep, activity, and cardiac data to identify subtle features and form physiological understandings that cannot be obtained using conventional analysis methods. For example, manually analyzing sensor data 302 to obtain such analysis results may be impractical or impossible. This is merely an example and not a limitation, and various suitable training techniques are considered.

[0093] For example, multiple attention heads can allow the machine learning model 316 to allocate resources to focus on different aspects of the input data, thereby making different predictions. Continuing with the example above where prediction 110 indicates a sleep-cardiac correlation, each analysis head can be trained to detect a specific relationship between sleep / activity patterns from accelerometer data 306 and physiological responses from electrical data 304. This enables the prediction system 114 to provide a comprehensive multimodal analysis of sensor data 302 for use in cardiac health status analysis and dynamic cardiac stress testing. In this way, the techniques described herein support the adaptability of the prediction system 114 to efficiently provide targeted sleep-cardiac related information and physiologically contextualized arrhythmia detection.

[0094] Once the machine learning model 316 is trained, the sensor data 302 is processed by the feature extraction module 318, which generates data features 320. For example, the feature extraction module 318 preprocesses the sensor data 302 to generate usable (e.g., processable) input for the trained machine learning model 326. The feature extraction module 318 may generate at least a portion of the data features 320 based on various properties of the accelerometer signal present in the accelerometer data 306, such as amplitude calculations, body angle measurements, sleep / wake indices, movement patterns, and / or activity level indices. These extracted features may include time-domain, frequency-domain, and statistical measures that capture relevant information about sleep stages, movement, and body position. It should be understood that, at least in some embodiments, the training data 314 may include accelerometer data acquired at the same sampling frequency as the accelerometer data 306.

[0095] The feature extraction module 318 is also operable to perform body angle calculations as part of the output data features 320. In some implementations, body angle techniques can be used to extract positional information from accelerometer data 306 to determine changes in body position and sleep / wake states. The feature extraction module 318 can analyze changes in accelerometer orientation, such as changes in the gravity vector component, which can be associated with and / or influenced by body position, sleep stage, and movement activity. These positional properties of the accelerometer data 306 can be used to derive body angle measurements, sleep / wake patterns, and activity patterns. Features derived from body angles can also be combined with other data features 320 to provide a comprehensive set of inputs for a trained machine learning model 326. In this way, the prediction system 114 can simultaneously capture motion and positional information from a single accelerometer signal, which improves the detection of correlations between body position, sleep stage, and activity events—something that is not possible using conventional scale.

[0096] In one or more embodiments, the feature extraction module 318 may also extract cardiac features from the electrical data 304, such as heart rate variability, QRS complex morphology, RR interval patterns, and arrhythmia features that may be associated with changes in sleep stage or activity level.

[0097] The feature extraction module 318 can also implement various additional techniques, such as wavelet decomposition, principal component analysis, peak detection, statistical analysis, and / or other signal processing methods, to isolate and quantify relevant aspects of the electrical data 304 and / or accelerometer data 306. For example, a step count detection algorithm can be applied to identify peaks above a predetermined threshold, while amplitude calculation can provide information about the overall activity level. The feature extraction module 318 can incorporate both time-domain features (e.g., mean, variance, and peak characteristics) and frequency-domain features (e.g., spectral energy and dominant frequency) to enhance sleep / wake classification and activity classification. The feature extraction module 318 can also optimize and / or refine the data features 320, for example, based on their discriminative ability to detect specific sleep stages, activity levels, or cardiac events.

[0098] Once data features 320 are extracted, analysis module 322 configures them for input into a trained machine learning model 326 (e.g., machine learning model 316 output by training module 312). In one or more embodiments, analysis module 322 uses encoder 324. Encoder 324 can be configured to process input data and compress it into a compact representation, and may include one or more of convolutional encoders, recurrent encoders, transformer encoders, one or more autoencoder variants, etc. For example, encoder 324 generates a compressed representation based on data features 320 that can be efficiently processed by trained machine learning model 326. In the example, encoder 324 reduces the dimensionality of data features 320 while preserving relevant information, thereby creating a compact representation suitable for use as input to trained machine learning model 326.

[0099] The analysis module 322 processes the encoded data features 320 using a trained machine learning model 326 to generate one or more predictions 110 for output. Prediction 110 may include various information. For example, sleep stage classification 120 may indicate sleep stages, activity level classification 122 may indicate different levels of physical activity during wakefulness, heart rhythm classification 124 may indicate the detection of arrhythmias, arrhythmia correlation 126 may indicate the correlation between cardiac events and sleep / activity states, and cardiac health status prediction 128 may provide a comprehensive physiological analysis combining sleep and activity analysis with the cardiac context and / or additional predictions providing supplementary analysis results. In one or more examples, one or more predictions 110 may include confidence intervals, such as confidence levels for associated values ​​or conditions.

[0100] In at least one embodiment, sleep stage classification 120 may include classifying sleep and wakefulness states by analyzing patterns in accelerometer data 306 to identify sleep and wakefulness periods, and may provide accurate sleep / wake detection while operating at a lower sampling frequency compared to conventional systems. Activity level classification 122 may distinguish activity levels including, but not limited to, sedentary, light activity, moderate activity, vigorous activity, or combinations thereof. Activity level classification 122 may incorporate features such as time-domain and frequency-domain characteristics extracted from accelerometer data 306, classifying activity intensity based on movement patterns and / or step counts over a predetermined time period. As a non-limiting example, activity equivalent to a walking speed of 2 miles per hour (mph) or higher may indicate light activity, while speeds below 2 mph may indicate inactivity / rest. Heart rhythm classification 124 may, for example, indicate whether the user associated with sensor data 302 has experienced a normal heart rhythm, a specific type of arrhythmia (e.g., atrial fibrillation, bradycardia, ventricular arrhythmia), and the time point of a cardiac event, to name just a few.

[0101] Arrhythmia correlation 126 can include indications of the correlation between arrhythmias and additional physiological events such as sleep stages, activity events, etc., for example, when a trained machine learning model 326 is capable of identifying and analyzing the relationship between arrhythmia occurrence and various sleep stages or levels of physical activity. For example, arrhythmia correlation 126 could indicate that, for a particular user, atrial fibrillation events are more likely to occur during a specific sleep stage or activity intensity. Combining arrhythmia correlation 126 with heart rhythm classification 124 can indicate the temporal association between specific arrhythmias and sleep / activity states, such as atrial fibrillation episodes during REM sleep, bradycardia events during deep sleep, or ventricular arrhythmias that may be triggered by exercise intensity. Such analysis results can provide a comprehensive view of physiological responses and triggers of cardiac events, enabling healthcare providers to better understand whether arrhythmias are related to sleep patterns, activity levels, or other physiological contexts. Furthermore, arrhythmia correlation 126 and heart rhythm classification 124 can provide clinicians with detailed information to make informed decisions regarding cardiac care and lifestyle modifications without relying solely on patient self-reporting.

[0102] In various examples, sensor data 302 also includes SpO2 data 308 and / or various additional data 310. These additional measurements can be fed together with electrical data 304 and / or accelerometer data 306 into a machine learning model 316 to predict one or more predictions 110. Thus, the techniques described herein support multimodal prediction, which provides analytical results that conventional techniques relying solely on accelerometer data 306 or solely on electrical data 304 cannot obtain.

[0103] In some implementations, SpO2 data 308 can be used to improve the accuracy of sleep prediction, activity prediction, and cardiac prediction and / or to validate these predictions. For example, analysis module 322 can process SpO2 data 308 in conjunction with electrical data 304 and / or accelerometer data 306 to identify potential sleep or activity periods. A trained machine learning model 326 can be configured to detect changes in blood oxygen saturation levels during sleep periods (which may coincide with sleep apnea events) and correlate these changes with sleep patterns in accelerometer data 306. Additionally or alternatively, trained machine learning model 326 can be configured to detect changes in blood oxygen saturation levels during activity periods and correlate these changes with arrhythmias identified based on electrical data 304. By combining these data sources, prediction system 114 can operate to distinguish different types of sleep stages, activity events, and arrhythmias in a more accurate manner. This multimodal approach enables detailed and accurate sleep, activity, and cardiac predictions, which reduces false positives and provides additional background information on detected sleep quality, activity levels, and arrhythmias.

[0104] The following discussion describes techniques that can be implemented using the systems and devices described above. References will be made to these techniques in the following sections of the discussion. Figures 1 to 3 .

[0105] Figure 4 A non-limiting example of cardiac monitoring with sleep stages and activity-related factors is depicted, wherein the monitoring system 400 generates a comprehensive health report for output. For example, the monitoring system 400 is one implementation of the prediction system 114 and can utilize [a combination of...]. Figure 3 Described machine learning.

[0106] In this example, the monitoring system 400 first receives electrical data 304 and accelerometer data 306 as input. The accelerometer data 306 is first processed at the sleep-wake analysis 402 to generate activity data 404 and sleep data 406. The activity data 404 and sleep data 406 together provide comprehensive information about the patient's sleep-wake cycle and activity level throughout the observation period. Activity data 404 is categorized into active states 408 and inactive states 410. For example, active states 408 and inactive states 410 can be data labels or bins to distinguish data collected during periods when the person 102 is awake and active from periods when they are awake but inactive. Although not shown, sleep data 406 can be categorized into different sleep stages (e.g., light sleep, deep sleep, REM sleep, non-REM sleep, etc.).

[0107] In parallel, electrical data 304 is processed by a feature extraction process 412 that generates heart rhythm / heartbeat data 414. Feature extraction process 412, for example, extracts relevant cardiac features and patterns from electrical data 304. Heart rhythm / heartbeat data 414 may include information about heart rate variability, RR interval, QRS complex characteristics, and other temporal and morphological features extracted from electrical data 304. The cardiac monitoring system 118 may, for example, generate a heart rhythm classification 124 using heart rhythm / heartbeat data 414 by using a trained machine learning model 326. For example, heart rhythm / heartbeat data 414 may be part of data features 320.

[0108] Aggregator 416 combines the outputs from cardiac monitoring system 118 and activity detection system 116 to generate health report 418. For example, aggregator 416 can combine heart rhythm analysis with sleep stage and activity status information. Aggregator 416 can integrate heart rhythm / beat data 414, activity data 404, and sleep data 406 to generate correlations between cardiac events and patient activity status. The health report 418 generated by aggregator 416 may include one or more analytical outputs for comprehensive patient assessment.

[0109] In the example shown, health report 418 includes arrhythmia correlation 126, sleep stage classification 120, activity level classification 122, and cardiac health status prediction 128 as integrated outputs. However, it should be understood that one or a subset of these outputs may be included in health report 418 in various situations. Because monitoring system 400 processes various types of sensor data, health report 418 may include various multimodal information, such as correlations between cardiac events and sleep stage patterns, and assessments of cardiac health status measures related to daily activity patterns and sleep behavior. It should be understood that although monitoring system 400 is shown as processing electrical data 304 and accelerometer data 306, it may also process other data, including data 308 and various supplementary data 310 (e.g., gyroscope data).

[0110] Odds ratio analysis 420 can be output as part of arrhythmia correlation 126. Odds ratio analysis 420 can highlight trends in arrhythmia occurrence related to sleep-wake cycles and / or activity levels in a patient population. Odds ratio analysis 420 indicates the probability of arrhythmia occurring during different physiological states by comparing the likelihood of cardiac events occurring during active and inactive periods and / or sleep and wake periods. Prediction system 114 can process or retrieve data from a large patient population to generate statistically significant analyses of arrhythmia patterns, which can provide analysis of whether a particular sleep stage or activity level increases or decreases the probability of arrhythmia occurrence. For example, odds ratio analysis 420 can determine that patients in a specific demographic group or the general population are 2.5 times more likely to experience atrial fibrillation during deep sleep than during wakefulness, or that the frequency of ventricular arrhythmias during vigorous activity is 1.8 times higher than during sedentary periods.

[0111] Odds ratio analysis 420 can analyze fourteen heart rhythm types divided into two groups. The first group can include second-degree type II atrioventricular block, complete atrioventricular block, atrial fibrillation, bigeminy, trigeminy, ectopic atrial rhythm, ventricular voluntary rhythm, junctional rhythm, arrest, supraventricular tachycardia, ventricular tachycardia, and Wenckebach rhythm. The second group can include sinus rhythm and artifacts. Odds ratio analysis 420 can also classify atrial fibrillation devices into high-atrial fibrillation devices (devices with an atrial fibrillation rate of 85% or higher and no sinus rhythm) and low-atrial fibrillation devices (all other atrial fibrillation devices).

[0112] In one or more embodiments, the ratio analysis 420 may be in the form of a two-squared contingency table, with “heart rhythm” versus “non-heart rhythm” and “sleep” state versus “wake” state (and / or a specific sleep stage or activity level), and apply the following formula:

[0113]

[0114] Where OR stands for odds ratio. One or more odds ratio analyses can be generated using the above formula, which can be applied to each individual heart rhythm to determine the odds ratio for that rhythm. In the first example, a heart rhythm duration analysis is performed, where... This indicates the time (in seconds) spent at the assessed specific heart rate during sleep. This represents the time spent in all heart rhythms during sleep minus , This indicates the time during which the heart is at that specific rhythm while the person is awake, and This represents the time spent in all heart rhythms during wakefulness minus In the second example, an analysis of the onset of cardiac rhythm episodes was performed, where... This indicates a specific heart rhythm that begins during sleep (e.g., the specific heart rhythm begins during the detected sleep period). This represents the total number of first-group heart rhythm episodes that begin during sleep, minus... , This indicates a specific heart rhythm that begins during wakefulness (e.g., the specific heart rhythm begins during the detected wakefulness period), and This represents the total number of first-group heart rhythm episodes that begin during the period of wakefulness minus For each heart rhythm type, the ratio analysis 420 can select data that includes that specific heart rhythm (e.g., analyzed electrical data 304 corresponding to a specific heart rhythm category 124).

[0115] In general, in at least one embodiment, the odds ratio analysis 420 can indicate the proportion of a specific heart rhythm that occurs during sleep relative to wakefulness, wherein a higher value corresponds to a higher probability that the heart rhythm occurs during sleep. The odds ratio analysis 420 can provide clinical analysis results regarding how person 102 compares to the general population or another user group segmented by age, sex, ethnicity, and / or other demographic indicators. Therefore, the health report 418 can provide population-level analysis results as well as analysis results regarding person 102, or provide population-level analysis results as an alternative to analysis results regarding person 102.

[0116] Figure 5A and Figure 5B A non-limiting example 500 of a user interface for displaying patient cardiac monitoring data with sleep and activity correlations is shown. As depicted in example 500, the user interface 502 presents comprehensive cardiac monitoring results, including various analyses derived from arrhythmia correlation 126, sleep stage classification 120, and activity level classification 122 generated according to the techniques described herein. The user interface 502 represents the results of an observation period in which a person 102 wears monitoring device 104 to capture sensor data 302, including electrical data 304 and accelerometer data 306. In some examples, the user interface 502 displays data generated during the observation period. Additionally or alternatively, the user interface 502 displays data generated after the end of the observation period.

[0117] In this example, the user interface 502 includes a patient event reporting section 504 that displays detailed cardiac monitoring information for a specific time period (e.g., January 14-15, 2025). The patient event reporting section 504 provides a comprehensive view of cardiac events associated with sleep and activity states, detected by the cardiac monitoring system 118 and the activity detection system 116, respectively. A graph 506 within the patient event reporting section 504 displays heart rate measurements in beats per minute (BPM) and detected arrhythmias, plotted relative to time within the specific time period. Graph 506 visualizes heart rhythm patterns and their temporal relationship to patient activity and sleep cycles.

[0118] An information panel 508 located adjacent to chart 506 provides background details about the displayed data, such as indicating "Day 11 and Day 12" of the monitoring period. Information panel 508 can display statistical summaries, measurement parameters, or other relevant metadata associated with the cardiac monitoring session. In this example, information panel 508 also marks specific events, such as patient diary events, patient-triggered events, and specific cardiac events and / or abnormalities. Event area 510 indicates the portion of the chart of interest that has been indicated, for example, via heart rhythm classification 124, for cardiac events or abnormalities. Within event area 510, the elevated heart rate area 512 indicates periods when heart rate measurements exceed normal baseline levels, which may correspond to an episode of arrhythmia or increased cardiac activity.

[0119] Chart 506 also includes a sleep region 514 and an activity region 516 displayed below the heart rate trajectory to provide an active context for cardiac measurements. The sleep region 514 indicates time periods identified as sleep by the person 102 based on the analysis of accelerometer data 306 by the activity detection system 116. The activity region 516 indicates time periods identified as active by the person 102, such as periods involving exercise detected by the activity detection system 116 equivalent to a walking speed of 2 mph or higher. Event markers 518 are positioned throughout chart 506 to indicate specific patient events identified during the monitoring period, such as arrhythmia episodes, symptomatic events, or other clinically relevant events. An event selector 520, located at the bottom of the patient event reporting section 504, enables navigation between different detected events and time periods within the monitoring data. The event selector 520 allows the user to select a specific event marked by event marker 518 for detailed examination, thus responding to the selection to... Figure 5B The extended analysis view shown.

[0120] Figure 5B Additional details of the user interface 502 are shown, including those for checking pass / fail. Figure 5AThe event selector 520 provides access to extended analytical capabilities for a specific cardiac event. For example, in this example, event selector 520 indicates that "Event 1" has been selected. Voltage graph 522 displays electrocardiogram waveform data corresponding to the selected cardiac event, thus showing the potential measurements captured by monitoring device 104 during the time period corresponding to that specific event. In this example, "Event 1" is a supraventricular tachycardia (SVT) event, where the heart rate is abnormally fast and originates above the ventricles (e.g., in the atrium or atrioventricular node). Voltage graph 522 indicates the cardiac signal morphology (e.g., as part of rhythm classification 124) analyzed to detect and classify SVT.

[0121] The magnified voltage graph 524 provides a magnified view of a selected portion of the voltage graph 522, enabling detailed examination of waveform characteristics such as ST segment variations caused by SVT, T-wave morphology, or other anomalies. The event region 526 on the voltage graph 522 corresponds to the portion included in the magnified voltage graph 524. In at least one embodiment, the event region 526 can be adjusted in response to user input to display another portion of the voltage graph 522 in the magnified voltage graph 524.

[0122] Information panel 528 displays detailed metrics and background information about the selected cardiac event, including heart rate measurements, event duration, and time point information. Information panel 528 can present quantitative analysis results generated by cardiac monitoring system 118, such as rhythm classification, beat interval, or other cardiac parameters derived from electrical data 304. Activity level index 530 indicates the activity intensity level associated with the selected cardiac event, such as "Active (Intense)" in this example. The activity intensity level can correspond to the activity level classification 122 determined by activity detection system 116. Activity level index 530 provides background information about the patient's activity status at the time of the cardiac event, thereby enabling the association of physical activity with the arrhythmia episode.

[0123] The return button 532 allows navigation from the detailed event analysis interface back to the main view of the patient event report section 504 (e.g., Figure 5A The return button 532 facilitates the user workflow by allowing switching between overview analysis mode and detailed analysis mode within the user interface 502. The comprehensive interface design enables healthcare providers to examine cardiac monitoring data at multiple levels of detail while maintaining an understanding of the sleep and activity context throughout the analysis.

[0124] In some implementations, the user interface 502 may incorporate the results of odds ratio analyses and / or regression analyses performed on a large patient population to identify trends and patterns of arrhythmias. Such population-level analyses enable the prediction system 114 to generate more accurate arrhythmia correlations 126 by leveraging statistical analysis results derived from a broad patient dataset. For example, an odds ratio analysis 420 may be provided within the user interface 502 to demonstrate statistically significant associations with specific sleep or activity states based on population-level trends.

[0125] Figure 6 Example 600 of an activity-based comprehensive cardiac assessment interface is shown. Example 600 is... Figure 5A Another example of the user interface 502 described herein. In this example 600, the user interface 502 displays an activity report section 602 that provides a detailed analysis of cardiac performance during various exercise periods. The activity report section 602 includes multiple graphical representations that enable healthcare providers to assess cardiac response patterns and identify exercise-related arrhythmias during prolonged monitoring periods.

[0126] The activity report section 602 includes a heart rate response curve 604, which displays the actual heart rate measurements recorded during the detected activity event. The heart rate response curve 604 presents time-series data showing heart rate data collected from the start of physical activity until the end of the recovery period. An expected heart rate curve 606 is superimposed on the heart rate response curve 604 to provide a comparative analysis between the predicted and observed cardiac responses. For example, the expected heart rate curve 606 can be generated using an algorithmic model that considers patient demographics, baseline fitness levels, and established cardiac response parameters. The comparison between the heart rate response curve 604 and the expected heart rate curve 606 can detect chronovariant dysfunction, where cardiac output does not increase appropriately during exertion. For instance, the prediction system 114 can indicate chronovariant dysfunction in response to the heart rate response curve 604 deviating from the expected heart rate curve 606 by at least a threshold amount.

[0127] Activity background graph 608 displays a quantitative analysis of arrhythmic events across different activity states. Activity background graph 608 categorizes cardiac events into different phases, including rest periods (e.g., "rest"), active exercise periods (e.g., "during exercise"), and post-exercise recovery periods immediately following the active exercise periods (e.g., "post-exercise"). Activity background graph 608 shows event counts for each activity background, allowing healthcare providers to identify patterns of arrhythmias that occur more frequently during specific activity levels. In at least one variant, activity background graph 608 may separate data by activity intensity level, including mild, moderate, and vigorous exercise categories based on exercise patterns comparable to walking speed and activity classifications derived from accelerometers (e.g., activity level classification 122).

[0128] The activity report section 602 also includes an arrhythmia stress graph 610, which quantifies the percentage of time spent in various arrhythmic states during different activity backgrounds. The arrhythmia stress graph 610 presents data as percentage values, allowing for standardized comparisons across different monitoring periods and patient groups. Therefore, the activity report section 602 can display a comprehensive assessment of exercise-related cardiac evaluations.

[0129] Figure 7 Example 700 of a multi-level activity classification display is shown, which provides a comprehensive analysis of cardiac activity at different exercise intensities. Example 700 also shows another example of an activity reporting section 602. In this example, the activity reporting section 602 includes multiple graphical representations for analyzing heart rate responses during various activity levels over a long monitoring period. For example, the activity reporting section 602 may include heart rate responses stratified by activity level.

[0130] The activity report section 602 includes a mild activity graph 702, a moderate activity graph 704, a vigorous activity graph 706, and an overlay graph 708. The mild activity graph 702 displays heart rate measurements taken during periods of mild physical activity, such as walking or gentle exercise. The moderate activity graph 704 presents heart rate data taken during periods of moderate-intensity activity, which may include activities comparable to brisk walking or light jogging. Compared to periods of mild activity, the moderate activity graph 704 may show an elevated heart rate response. The vigorous activity graph 706 displays heart rate measurements during periods of high-intensity physical activity, such as running, climbing stairs, or other vigorous exercise. Each of the mild activity graph 702, moderate activity graph 704, and vigorous activity graph 706 presents heart rate data plotted relative to elapsed time, allowing for detailed analysis of cardiac response within specific activity intensity categories. Multiple graphs may be displayed on each graph, where different line patterns (e.g., solid lines, short dashed lines, long dashed lines) indicate data acquired during different activity events. Furthermore, in this example, different line widths are used to distinguish different levels of activity, with the light activity graph 702 using the thinnest line and the intense activity graph 706 using the thickest line. In variations, other visual indicators (such as colors, labels, or other line styles) can be used to differentiate the different graphs from one another.

[0131] Overlay Graph 708 combines heart rate data from all activity intensity levels to provide a comprehensive view of cardiac performance across the range of physical activity. This combined visualization enables healthcare providers to compare heart rate responses at different exercise intensities and identify patterns or abnormalities that may occur during specific activity levels. Overlay Graph 708 can facilitate the detection of chronotropic dysfunctions that may not be apparent during rest periods or single-intensity monitoring.

[0132] Figure 8 An example 800 of a dynamic cardiac stress testing interface is shown, which facilitates the monitoring and analysis of cardiac activity during extended periods of wear. Example 800 includes a user interface 802 that provides navigation and control functions to a person 102 to perform a dynamic cardiac stress testing procedure.

[0133] Activity graph 804 displays data from vigorous activity detection, thus illustrating periods of high-intensity physical activity that may trigger cardiac stress responses during extended monitoring periods. Activity graph 804 presents time data indicating the detection of vigorous activity.

[0134] User interface 802 displays instruction 136, which provides specific guidance for post-exercise recovery positioning to optimize cardiac data acquisition for dynamic cardiac stress testing in this example. Instruction 136 instructs the person 102 to recover in a supine position for 10 minutes immediately after detected strenuous activity, which enables the acquisition of electrical data 304 in a supine position, thereby collecting specific data that can be used for cardiac stress testing. Instruction 136 can be automatically triggered when monitoring device 104 detects strenuous activity by analyzing accelerometer data 306, thereby ensuring that a recovery position is taken at an appropriate time during the extended wear period.

[0135] User interface 802 supports the detection of cardiac abnormalities that may manifest during or after exercise, including ST-segment depression, ST-segment elevation, T-wave inversion, T-wave flattening, and T-wave pseudonormalization, by issuing instructions 136 to facilitate data acquisition under appropriate activity and rest conditions. Therefore, monitoring device 104 can be used to acquire diagnostic information comparable to conventional exercise stress testing, while allowing for prolonged monitoring in dynamic environments.

[0136] The following sections describe example programs of a sleep and activity-related cardiac monitoring system according to one or more embodiments. Aspects of each program can be implemented in hardware, firmware, software, or a combination thereof. These programs are shown as a set of blocks specifying operations that can be performed by one or more devices, and are not necessarily limited to the order in which these operations are performed by the corresponding blocks as shown. For example, one or more blocks of the program specify operations that can be programmed into instructions by hardware (e.g., a processor, microprocessor, controller, firmware) to create a dedicated machine for executing the algorithm illustrated in the flowchart. Therefore, these instructions can be stored on a computer-readable storage medium that enables the hardware to execute the algorithm. References will be made in the following discussion sections. Figures 1 to 8 .

[0137] Figure 9 A flowchart is provided, depicting the algorithm as a step-by-step process 900 in an example implementation. This algorithm can be executed by a processing device to correlate cardiac monitoring with sleep and activity data based on measurements from a wearable monitoring device. The process 900 processes both accelerometer and potential measurements to produce an analysis that correlates arrhythmia episodes with sleep and / or activity periods.

[0138] In this example, firstly, the wearable monitoring device acquires measurements of the user during the observation period, including accelerometer measurements and electrical potential measurements (box 902). For example, measurement 108 is generated by monitoring device 104 during continuous wear by person 102. In various examples, monitoring device 104 detects motion patterns based on accelerometer data 306 while simultaneously capturing cardiac electrical activity using one or more sensors 202. Accelerometer data 306 can be sampled at a lower frequency than conventional activity trackers typically operate at 25-100 Hz, thereby extending the battery life of monitoring device 104 and conserving its memory resources. Electrical data 304 (e.g., electrical potential measurements) can be acquired substantially continuously to provide heart rate information that can be correlated with detected motion patterns.

[0139] Monitoring device 104 may process the measurement value 108 locally, at least partially, before sending it to analysis platform 106. Alternatively, analysis platform 106 may be included as part of monitoring device 104. In yet another example, monitoring device 104 may send the measurement value 108 to analysis platform 106 for processing. Additionally or alternatively, monitoring device 104 may use various data compression techniques to compress sensor data to reduce battery consumption during transmission (e.g., wireless or wired transmission) to an external computing device after the observation period ends.

[0140] Sleep and wake periods are detected based on patterns and body angle calculations from accelerometer data (box 904). For example, activity detection system 116 processes accelerometer measurements (e.g., accelerometer data 306) to identify resting and moving periods corresponding to sleep and wake cycles. Body angle calculations are performed using accelerometer measurements to determine when the user is in a reclined position indicating a sleep period or an upright position indicating a wake period. Activity detection system 116 can analyze movement patterns, positions, and activity levels (e.g., based on the number of steps within a predetermined time period) to distinguish different physiological periods throughout the observation period. Activity detection system 116 can analyze movement patterns, positions, and activity levels (e.g., based on the number of steps within a predetermined time period) to distinguish different physiological periods throughout the observation period. Activity detection system 116 can classify sleep based on resting states detected through body movement and body angle analysis, while identifying wake periods through movement patterns and upright positions.

[0141] A sleep stage classification is generated based on the detected sleep periods (box 906). For example, prediction system 114 processes the detected sleep periods to generate a sleep stage classification 120, which can categorize different sleep stages into, for example, light sleep, deep sleep, REM sleep, and / or non-REM sleep. Sleep stage classification 120 can provide a temporal mapping of when a user experiences different sleep stages.

[0142] A machine learning model is used to process electrical potential measurements. This model is trained to associate patterns in the electrical potential measurements with heart rhythm classifications (box 908). For example, a trained machine learning model 326 receives electrical potential measurements as input and analyzes features such as heart rate variability, heart rhythm patterns, and electrical signal morphology to classify different types of arrhythmias and cardiac events (e.g., heart rhythm classification 124). The trained machine learning model 326 is trained using historical electrical potential measurements and historical outcome data from a user population to perform the heart rhythm classification task. Heart rhythm classification 124 identifies specific types of arrhythmias, heart rate variability, and other cardiac events that occur at specific times during the observation period.

[0143] Arrhythmia correlations are generated by associating sleep stage classifications with contemporaneous heart rhythm classifications (box 910). For example, prediction system 114 temporally aligns sleep stage classification 120 with heart rhythm classification 124, for example via aggregator 416, to identify relationships between sleep periods and the occurrence of arrhythmias. Arrhythmia correlation 126 indicates the occurrence of a specific type of cardiac event relative to different sleep stages, thereby providing analytical results regarding whether arrhythmias are more likely to occur during a particular sleep stage.

[0144] Odds ratio analysis (box 912) is generated based on the correlation between arrhythmias. For example, prediction system 114 performs statistical analysis on associated data to calculate odds ratios that quantify the probability of a specific arrhythmia occurring during different sleep stages. Odds ratio analysis 420 provides a quantitative measure of the association between sleep stages and arrhythmia occurrence, enabling healthcare providers to understand the statistical significance of observed patterns. For example, this analysis may include calculations comparing the incidence of arrhythmias during sleep periods and wake periods.

[0145] The output includes a health report (box 914) comprising at least one of sleep stage classification, arrhythmia correlation, or odds ratio analysis. For example, health report 418 presents a comprehensive analysis in a format suitable for clinical review and decision-making. This report may include visualizations overlaying sleep and activity information with cardiac rhythm events, providing healthcare providers with an intuitive representation of arrhythmia correlation 126. In various examples, health report 418 includes percentage details of arrhythmia types during different sleep stages, statistical analysis results, and time patterns, all of which aid in diagnosis and treatment planning.

[0146] Figure 10 A flowchart is provided, depicting the algorithm as a step-by-step process 1000 in an example implementation. This algorithm can be executed by a processing device to perform cardiac monitoring with activity level classification and arrhythmia correlation. Program 1000 performs a comprehensive analysis of cardiac events related to physical activity status, thereby providing healthcare providers with detailed analytical results on exercise-related cardiac patterns and health status predictions.

[0147] Measurements of the user, including accelerometer measurements and electrical potential measurements, are acquired by a wearable monitoring device during the observation period (box 1002). For example, monitoring device 104 continuously acquires physiological data 302 of person 102 over a long monitoring period. Accelerometer data 306 captures movement patterns and physical activity levels, while electrical data 304 records cardiac electrical activity via one or more sensors 202. During the observation period, monitoring device 104 can operate at a low sampling rate of less than 5 Hz (e.g., 1.6 or 1.56 Hz) for accelerometer data 306 to conserve battery power and memory resources. The long wear period can range from 7 to 14 days to capture comprehensive activity patterns and cardiac events across multiple daily cycles. This document, for example, combines... Figure 9 Box 902 describes additional example details.

[0148] Sleep and wake periods are detected based on patterns and body angle calculations from accelerometer data (box 1004). For example, activity detection system 116 processes accelerometer measurements (e.g., accelerometer data 306) to identify resting and moving periods corresponding to sleep and wake cycles. Body angle calculations are performed using accelerometer measurements to determine when the user is in a reclined position indicating a sleep period or an upright position indicating a wake period. Activity detection system 116 can analyze movement patterns, positions, and activity levels (e.g., based on the number of steps taken within a predetermined time period) to distinguish different physiological periods throughout the observation period. Activity detection system 116 can classify sleep based on resting states detected through body movement and body angle analysis, while identifying wake periods through movement patterns and upright positions.

[0149] Activity level classifications are generated based on the number of steps detected during waking periods (box 1006). For example, activity detection system 116 counts steps to determine whether person 102 is classified as having an activity level classification 122 during waking periods. Activity detection system 116 can classify activity (as opposed to inactivity) based on movement equivalent to a walking speed of 2 mph or higher. Activity can also be classified into mild, moderate, and vigorous intensity levels based on the number of steps taken within a predetermined time period, according to multiple step count thresholds. For example, activity detection system 116 can process accelerometer data 306 to detect peaks above a measurement threshold as steps. In one or more embodiments, monitoring device 104 may include a gyroscope capable of being triggered when activity is detected to provide additional analysis of the activity type or intensity. In some such examples, the gyroscope may remain deactivated during inactive periods to conserve power.

[0150] A machine learning model is used to process electrical potential measurements. This model is trained to associate patterns in the electrical potential measurements with heart rhythm classifications (box 1008). For example, a trained machine learning model 326 receives electrical potential measurements as input and analyzes features such as heart rate variability, heart rhythm patterns, and electrical signal morphology to classify different types of arrhythmias and cardiac events (e.g., heart rhythm classification 124). The trained machine learning model 326 is trained using historical electrical potential measurements and historical outcome data from a user population to perform the heart rhythm classification task. Heart rhythm classification 124 identifies specific types of arrhythmias, heart rate variability, and other cardiac events that occur at specific times during the observation period.

[0151] Arrhythmia correlations are generated by associating activity level classifications with contemporaneous heart rhythm classifications (box 1010). For example, prediction system 114 temporally aligns activity level classification 122 with heart rhythm classification 124, for instance, via aggregator 416, to identify relationships between activity periods and arrhythmia occurrence. Arrhythmia correlation 126 indicates the occurrence of a specific type of cardiac event relative to different activity levels, thus providing analytical results regarding whether arrhythmias are more likely to occur before, during, or after activity at a specific intensity level.

[0152] Cardiac health status predictions are generated based on arrhythmia correlations (box 1012). For example, prediction system 114 processes the results of arrhythmia correlation 126 to generate a cardiac health status prediction 128 that can provide analytical results regarding overall cardiac health status. Prediction system 114 can detect chronotropic dysfunction by comparing the expected heart rate response during a detected activity event with the actual heart rate response, thereby identifying situations where the heart rate does not increase as expected during exercise. Cardiac health status prediction 128 may incorporate analysis of heart rate variability patterns, recovery rates after exercise periods, and / or the frequency and timing of arrhythmia episodes relative to activity levels. In one or more embodiments, cardiac health status prediction 128 may compare cardiac rate load and counts during exercise periods with non-exercise wakefulness periods. Cardiac health status prediction 128 may also take into account patterns observed in multiple activity sessions captured during long monitoring periods.

[0153] The output includes a health report (box 1014) comprising at least one of activity level classification, arrhythmia correlation, or cardiac health status prediction. For example, prediction system 114 generates health report 418, which can present comprehensive analysis results through visualization and quantitative summarization. Health report 418 may include activity level classification 122 data, showing the distribution of daily activity patterns, steps, and / or intensity levels during the monitoring period. Results of arrhythmia correlation 126 can be presented as an overlay of cardiac events with activity periods, providing healthcare providers with context regarding the timing of a particular arrhythmia relative to physical activity. Cardiac health status prediction 128 can be incorporated into health report 418 as a recommendation or assessment of cardiac fitness and exercise tolerance based on the observed correlation between activity and cardiac response patterns.

[0154] Figure 11 A flowchart is drawn that depicts the algorithm as a step-by-step process 1100, in which dynamic cardiac stress testing is performed using wearable monitoring devices.

[0155] Measurements of the user are acquired from the wearable monitoring device during the observation period, including accelerometer measurements and electrical potential measurements (box 1102). For example, monitoring device 104 continuously acquires measurements 108 over a long wear period that can span 7 to 14 days. Accelerometer data 306 captures motion patterns and postural information at a low sampling rate (e.g., less than 5 Hz) to conserve power and memory resources of monitoring device 104. Electrical data 304 can represent a continuous electrocardiogram signal recording the electrical activity of the heart throughout the observation period. The long monitoring duration enables the capture of multiple motion events and recovery periods occurring during normal daily activities, thus providing a comprehensive dataset for dynamic stress test analysis. This paper, for example, in... Figure 9 box 902 and Figure 10 Additional details are described in box 1002.

[0156] Activity level classifications are generated based on accelerometer data to identify periods of activity (box 1104). For example, activity detection system 116 processes accelerometer data 306 to detect movement patterns corresponding to the activity state indicating a period of activity. As a non-limiting example, a period of activity could correspond to a walking speed of 2 mph or higher. Prediction system 114 can classify activity levels into mild, moderate, and intense categories based on the amplitude and frequency of detected movement and / or based on the number of steps taken within a predetermined time period. Activity level classification 122 can identify discrete movement events that can be used for stress test analysis, including the start and duration of each activity period. The system can also detect when a gyroscope should be activated to provide additional analysis results on the type or intensity of the detected activity.

[0157] Body position is determined based on accelerometer data during the post-exercise recovery period (box 1106). For example, activity detection system 116 uses accelerometer data 306 to calculate body angles within a 0-90 degree range from an upright to a supine position. Activity detection system 116 can calculate a reference vector based on a high-activity area where the patient can be in an upright position, and then determine the body angle at each time point as the polar angle relative to the reference upright vector. During the post-exercise recovery phase, activity detection system 116 can identify when the person 102 transitions to a supine position, which may induce ischemic abnormalities not visible during exercise. Body position determination allows prediction system 114 to categorize potential measurements based on whether the patient was in an upright or supine position during data acquisition. Therefore, accelerometer data 306 can provide continuous monitoring of positional changes throughout the recovery period without relying on patient input or manual positioning procedures.

[0158] Identify potential measurements acquired during rest, exercise, and post-exercise recovery periods in upright and supine positions (box 1108). For example, cardiac monitoring system 118 processes electrical data 304 to extract electrocardiogram (ECG) segments corresponding to different physiological states and positions. Cardiac monitoring system 118 can identify resting ECG data acquired before an exercise event when the patient is in both standing and supine positions. During exercise periods identified by activity level classification 122, cardiac monitoring system 118 can identify ECG data acquired when the patient is in an upright position performing physical activity. Cardiac monitoring system 118 can also identify ECG data acquired during recovery periods, such as potential measurements acquired per minute for up to 10 minutes when the person 102 is supine after exercise. Feature extraction process 412 organizes potential measurements according to concurrent activity status and position to enable systematic comparisons across different physiological conditions.

[0159] Potential measurements are compared across different activity levels and body positions to detect cardiac abnormalities (box 1110). For example, cardiac monitoring system 118 can analyze electrocardiogram (ECG) patterns to identify abnormalities indicating myocardial ischemia, which represents a mismatch between myocardial oxygen delivery and myocardial oxygen demand. Cardiac monitoring system 118 can detect ST-segment depression, ST-segment elevation, T-wave inversion, T-wave flattening, and T-wave pseudonormalization by comparing exercise and recovery ECG data with resting baseline measurements acquired in the same body position. In one or more embodiments, a trained machine learning model 326 processes potential measurements to identify patterns associated with coronary artery disease that may be present during exercise-induced stress or the recovery phase after exercise. The comparison process takes into account different body positions during data acquisition, as ECG morphology may differ between upright and supine positions and is unrelated to cardiac lesions.

[0160] The system outputs a cardiac health status prediction, which includes an assessment of potential coronary artery disease (CAD) (box 1112). For example, prediction system 114 generates cardiac health status prediction 128 based on analysis of electrocardiogram (ECG) abnormalities detected during a dynamic cardiac stress test procedure. Cardiac health status prediction 128, or the health report 418 generated therefrom, may include a set of sample ECGs from various physiological states, including supine resting ECGs, upright resting ECGs, upright ECGs for up to 10 minutes per minute post-exercise, and supine ECGs for up to 10 minutes per minute post-exercise. Cardiac health status prediction 128 may provide an assessment of CAD risk based on a comprehensive analysis of cardiac function during real-world exercise conditions, rather than a controlled laboratory protocol. User interface 502 may display cardiac health status prediction 128 along with supporting ECG data and activity context for clinical interpretation of the dynamic stress test results.

[0161] Figure 12 A flowchart was drawn that depicts the algorithm as a step-by-step procedure 1200, in which analysis and data acquisition are performed based on sleep / activity.

[0162] Accelerometer measurements of the user are acquired via a wearable monitoring device during the observation period (box 1202). For example, monitoring device 104 continuously acquires accelerometer data 306 to detect movement patterns and body position information throughout the observation period. Accelerometer data 306 can be sampled at a low sampling rate (e.g., less than 5 Hz) to conserve battery power while maintaining accurate detection capabilities for sleep and activity patterns.

[0163] Determine whether sleep was detected based on accelerometer measurements (box 1204). For example, activity detection system 116 processes accelerometer data 306 to identify rest periods and body positions corresponding to sleep states. For instance, sleep-wake analysis 402 may include analyzing body angle measurements and movement patterns to distinguish between sleep and wake periods.

[0164] If sleep is detected (e.g., "Yes" at box 1204), an LED-based photoplethysmography measurement is activated at the wearable monitoring device (box 1206). For example, monitoring device 104 can trigger more power-intensive photoplethysmography data acquisition to collect SpO2 data 308, which can be used for sleep apnea screening and diagnosis. When the person 102 is no longer asleep, the LED-based photoplethysmography can be turned off to conserve battery resources during the observation period.

[0165] If sleep is not detected (e.g., "No" at box 1204), it is determined whether activity was detected based on accelerometer measurements (box 1208). For example, activity detection system 116 analyzes accelerometer data 306 to identify movement patterns equivalent to walking speeds greater than a threshold (e.g., 2 mph), thereby distinguishing between active and inactive awake states.

[0166] If activity is detected (e.g., "Yes" at box 1208), the gyroscope of the wearable monitoring device is activated to obtain additional activity measurements (box 1210). For example, when activity is detected by activity detection system 116, monitoring device 104 can activate an additional data acquisition channel, such as the gyroscope, which is more expensive in terms of battery consumption but can provide additional analysis results on the type or intensity of the activity. For example, gyroscope activation enables enhanced activity classification and correlation with contemporaneous cardiac measurements from electrical data 304.

[0167] If no activity is detected (e.g., "No" at box 1208), the gyroscope remains disabled (box 1212). For example, when person 102 is in an inactive waking state, monitoring device 104 can maintain standard monitoring parameters without activating high-power sensors. For example, selective sensor activation methods extend battery life while ensuring comprehensive data acquisition during physiologically relevant periods for use in cardiac health prediction 128 and arrhythmia correlation analysis 126.

[0168] The various functional units shown in the accompanying drawings and / or described herein are implemented in any of a variety of different ways, such as hardware circuitry, software or firmware executing on a programmable processor, or any combination of two or more of hardware, software, and firmware. The provided methods are implemented in any of a variety of devices, such as a general-purpose computer, a processor, or a processor core. Suitable processors include, for example, general-purpose processors, special-purpose processors, conventional processors, digital signal processors (DSPs), graphics processing units (GPUs), parallel accelerator processors, multiple microprocessors, one or more microprocessors associated with a DSP core, controllers, microcontrollers, application-specific integrated circuits (ASICs), field-programmable gate array (FPGA) circuitry, any other type of integrated circuit (IC), and / or state machines.

[0169] In one or more embodiments, the methods and programs provided herein are implemented in computer programs, software, or firmware, and are included in a non-transitory computer-readable storage medium for execution by a general-purpose computer or processor. Examples of non-transitory computer-readable storage media include read-only memory (ROM), random access memory (RAM), registers, cache memory, semiconductor storage devices, magnetic media such as internal hard disks and removable disks, magneto-optical media, and optical media such as CD-ROM discs and digital versatile optical discs (DVDs).

[0170] It should be understood that many variations are possible based on the disclosure herein. Although features and elements have been described above in specific combinations, each feature or element can be used alone without other features and elements, or in various combinations with or without other features and elements.

[0171] Clause 1. A method implemented by a processing device, the method comprising: acquiring measurements of a user generated by a wearable monitoring device during an observation period, the measurements including accelerometer data and potential measurements; detecting sleep periods and wake periods based on the accelerometer data; generating a sleep stage classification based on the accelerometer data during the detected sleep periods; processing the potential measurements using a machine learning model trained to associate patterns in the potential measurements with a heart rhythm classification; generating a heart rhythm correlation based on the sleep stage classification and the contemporaneous heart rhythm classification, the heart rhythm correlation describing the temporal relationship between detected heart rhythm irregularities and specific sleep stages in the sleep stage classification; and outputting a health report including the heart rhythm correlation.

[0172] Clause 2. The method according to Clause 1, wherein the accelerometer data is acquired at a sampling rate of less than 5 Hz.

[0173] Clause 3. The method according to Clause 1 or Clause 2, wherein detecting the sleep period and the wake period based on the accelerometer data comprises: detecting peak values ​​above a threshold in the accelerometer data as steps; calculating body angle measurements based on the accelerometer data to determine body position; and distinguishing the sleep period and the wake period based on the detected steps and the body angle measurements.

[0174] Clause 4. The method according to any one of Clauses 1 to 3, wherein generating the sleep stage classification includes classifying the sleep periods into light sleep, deep sleep, rapid eye movement (REM) sleep, and non-REM sleep.

[0175] Clause 5. The method according to any one of Clauses 1 to 4 further comprises: indicating activity events during the awake period based on the accelerometer data acquired during the awake period; generating an activity level classification of the activity events based on the intensity of the physical activity performed; and generating an additional arrhythmia correlation based on the activity level classification and a corresponding concurrent heart rhythm classification, the additional arrhythmia correlation describing the temporal relationship between a detected arrhythmia and a specific activity level in the activity level classification.

[0176] Clause 6. The method according to Clause 5, wherein the activity level classification is based on motion determined from the accelerometer data that is equivalent to at least a threshold velocity, classifying the physical activity as mild, moderate, or vigorous.

[0177] Clause 7. The method described in Clause 5 or Clause 6 further comprises: detecting chronotropic dysfunction based on a comparison of the expected heart rate response and the actual heart rate response during the said physical activity.

[0178] Clause 8. The method according to any one of Clauses 1 to 7 further comprises: generating an odds ratio analysis based on the correlation of the arrhythmia, the odds ratio analysis quantifying the probability of a particular arrhythmia occurring during the sleep period and the wake period.

[0179] Clause 9. The method according to Clause 8, wherein the ratio analysis calculates the statistical association between the specific arrhythmia and the sleep period in the user group.

[0180] Clause 10. The method according to any one of Clauses 1 to 9, wherein the health report includes visualization of the correlation between the arrhythmia and the sleep and wake periods.

[0181] Clause 11. A processing device comprising: one or more processors; and a memory storing computer-readable instructions executable by the one or more processors to perform operations including: acquiring measurements of a user generated by a wearable monitoring device during an observation period, the measurements including accelerometer data and potential measurements; detecting sleep periods and wake periods based on the accelerometer data; generating a sleep stage classification based on the accelerometer data during the sleep period; generating an activity level classification based on steps detected during the wake period; processing the potential measurements using a machine learning model trained to associate patterns in the potential measurements with a heart rhythm classification; associating the sleep stage classification and the activity level classification with a contemporaneous heart rhythm classification to generate a heart rhythm correlation; generating a heart health status prediction based on the heart rhythm correlation; and outputting a health report including the heart health status prediction.

[0182] Clause 12. The processing apparatus according to Clause 11, wherein the accelerometer data is acquired at a sampling rate of less than 5 Hz.

[0183] Clause 13. The processing device according to Clause 11 or Clause 12, wherein generating the activity level classification includes: classifying physical activity as sedentary, mild, moderate or vigorous based on the speed of motion determined according to the accelerometer data.

[0184] Clause 14. The processing apparatus according to Clause 13, wherein the operation further comprises: detecting chronotropic dysfunction by comparing the user’s expected heart rate response during an activity event within the wakefulness period with the actual heart rate response.

[0185] Clause 15. The processing apparatus according to Clause 14, wherein the operation further comprises: determining the user's heart rate over time based on the potential measurement; generating a time series plot of the heart rate during a detected activity event; stratifying the time series plot based on the activity level classification; and outputting the stratified time series plot as part of the health report.

Claims

1. A method implemented by a processing device, the method comprising: Acquire user measurements generated by wearable monitoring devices during the observation period, including accelerometer data and potential measurements; Sleep and wake periods are detected based on the accelerometer data; A sleep stage classification is generated based on the accelerometer data during the detected sleep periods; The potential measurements are processed using a machine learning model, which is trained to associate patterns in the potential measurements with heart rhythm classification. Arrhythmia correlations are generated based on the sleep stage classification and the concurrent heart rate classification, wherein the arrhythmia correlations describe the temporal relationship between detected arrhythmias and specific sleep stages in the sleep stage classification; and The output includes a health report related to the aforementioned arrhythmia.

2. The method according to claim 1, wherein, The accelerometer data is acquired at a sampling rate of less than 5 Hz.

3. The method according to claim 1 or 2, wherein, Detecting the sleep period and the wake period based on the accelerometer data includes: Peak values ​​above a threshold in the accelerometer data are detected as steps; Body angle measurements are calculated based on the accelerometer data to determine body position; and The sleep period and the wake period are distinguished based on the detected number of steps and the body angle measurement.

4. The method according to any one of claims 1 to 3, wherein, Generating the sleep stage classification includes categorizing the sleep periods into light sleep, deep sleep, REM sleep, and non-REM sleep.

5. The method according to any one of claims 1 to 4, further comprising: Based on the accelerometer data acquired during the awake period, the activity events during the awake period are indicated; Based on the intensity of the physical activity performed, an activity level classification of the activity event is generated; as well as Based on the activity level classification and the corresponding concurrent heart rhythm classification, an additional arrhythmia correlation is generated, which describes the temporal relationship between the detected arrhythmia and a specific activity level in the activity level classification.

6. The method according to claim 5, wherein, The activity level classification is based on motions that are equivalent to at least a threshold velocity, determined according to the accelerometer data, categorizing the physical activity as mild, moderate, or vigorous.

7. The method according to claim 5 or 6, further comprising: Temporal dysfunction is detected by comparing the expected heart rate response with the actual heart rate response during the physical activity.

8. The method according to any one of claims 1 to 7, further comprising: Based on the arrhythmia correlation, a ratio analysis is generated, which quantifies the probability of a specific arrhythmia occurring during the sleep period and the wakefulness period.

9. The method according to claim 8, wherein, The ratio analysis calculates the statistical correlation between the specific arrhythmia and the sleep period in the user group.

10. The method according to any one of claims 1 to 9, wherein, The health report includes visualizations of the correlation between the heart arrhythmia and the sleep and wake periods.

11. A processing apparatus, comprising: One or more processors; as well as A memory storing computer-readable instructions executable by the one or more processors to perform operations, including: Acquire user measurements generated by wearable monitoring devices during the observation period, including accelerometer data and potential measurements; Sleep and wake periods are detected based on the accelerometer data; A sleep stage classification is generated based on the accelerometer data during the sleep period; Activity level classification is generated based on the number of steps detected during the awake period; The potential measurements are processed using a machine learning model, which is trained to associate patterns in the potential measurements with heart rhythm classification. The sleep stage classification and the activity level classification are correlated with the heart rhythm classification during the same period to generate arrhythmia correlations; Based on the aforementioned arrhythmia correlation, a prediction of cardiac health status is generated; and The output includes a health report predicting the heart health status.

12. The processing apparatus according to claim 11, wherein, The accelerometer data is acquired at a sampling rate of less than 5 Hz.

13. The processing apparatus according to claim 11 or 12, wherein, Generating the activity level classification includes categorizing physical activity as sedentary, mild, moderate, or vigorous based on the movement speed determined from the accelerometer data.

14. The processing apparatus according to claim 13, wherein, The operation also includes: Temporal dysfunction is detected by comparing the user’s expected heart rate response during an activity event within the awake period with the actual heart rate response.

15. The processing apparatus according to claim 14, wherein, The operation also includes: The user's heart rate over time is determined based on the measured potential value; Generate a time-series plot of the heart rate during the detected activity event; The time series plot is stratified based on the activity level classification; and The stratified time series plot is output as part of the health report.

Images