Illness detection based on nervous system metrics
Inventors
Pho, Gerald • Aschbacher, Kirstin • Altini, Marco • Rai, Harpreet • Chapp, Michael
Assignees
Publication Number
US-12268530-B2
Publication Date
2025-04-08
Expiration Date
2041-06-24
Interested in licensing this patent?
MTEC can help explore whether this patent might be available for licensing for your application.
Abstract
Methods, systems, and devices for illness detection are described. A method may include receiving heart rate variability (HRV) data associated with a user from a wearable device, the HRV data collected via the wearable device throughout a first time interval and a second time interval subsequent to the first time interval. The method may include inputting the HRV data into a classifier, and identifying a satisfaction of deviation criteria between a first subset of the HRV data collected throughout the first time interval and a second subset of the HRV data collected throughout the second time interval. The method may include causing a graphical user interface (GUI) of a user device to display an illness risk metric for the user based on the satisfaction of the deviation criteria, the illness risk metric associated with a relative probability that the user will transition from a healthy state to an unhealthy state.
Core Innovation
The invention relates to methods, systems, and devices for illness detection based on physiological data collected from users via wearable devices, particularly focusing on nervous system metrics such as heart rate variability (HRV). The system can detect a user's transition from a healthy state to an unhealthy state, specifically identifying illness during the pre-symptomatic stage before symptom onset. This early detection uses physiological parameters including HRV data, temperature data, respiration rate, movement/activity data, and modifiable behavioral predictors like sleep and physical activity.
The problem being solved is the desire among users for more insight into their physical health, including early detection and prediction of illness to enable interventions before symptom onset. Current wearable devices collect physiological data but do not adequately detect transitions into illness, particularly during pre-symptomatic periods. The disclosed techniques improve illness detection accuracy by using classifiers that analyze physiological data over multiple time intervals, contextualize these data with cyclical biological rhythms, and factor in user location and behaviors to identify satisfying deviation criteria indicative of impending illness.
Claims Coverage
The claims disclose a system with wearable and user devices alongside processors configured to detect illness by analyzing heart rate variability data over specified time intervals using machine learning classifiers.
Detection of illness using dual machine learning classifiers on heart rate variability data
A wearable device measures heart rate variability data over two time intervals. A first machine learning classifier extracts features including RMSSD and resting heart rate data. A second classifier evaluates these features for deviations, specifically decreases in RMSSD and resting heart rate occurring concurrently, to identify potential illness.
Determination of HRV frequency content as a detection feature
The system determines frequency content, including low and high frequency bands, of HRV data for respective time intervals. The classifiers use these frequency domain features to identify deviation criteria indicative of illness.
Identification of divergence between low and high frequency HRV content
The system detects illness based on divergence between low-frequency and high-frequency HRV components, correlating these changes with decreases in RMSSD and resting heart rate to enhance predictive accuracy.
Use of sleep stage data in HRV analysis
Physiological data is classified into sleep stages. The classifier compares HRV data subsets corresponding to the same sleep stage across time intervals to detect deviations indicating illness risk.
Accounting for circadian rhythm in HRV data processing
The system segments time intervals based on a user's circadian rhythm and inputs corresponding HRV data portions into classifiers, improving deviation detection by aligning analysis with user's biological cycles.
User feedback to train illness prediction classifiers
User inputs such as positive illness tests or symptom onset reports are utilized to train and refine machine learning classifiers responsible for illness detection.
Integration of demographic illness data for enhanced prediction
Demographic illness data relevant to the user's geographical location is input into classifiers to contextualize and improve the detection of deviation criteria associated with illness risk.
Multi-user plurality-based illness risk metrics presentation
The system processes HRV data from multiple users to generate illness risk metrics and displays pertinent illness risk information for users on an administrator's device for monitoring purposes.
The independent claims encompass inventive features for illness detection using dual machine learning classifiers analyzing heart rate variability, frequency content, sleep stages, circadian rhythm segments, and user demographic data. The claims cover systems enabling early illness detection with personalized and group data inputs and support administrator monitoring of multiple users' illness risk metrics.
Stated Advantages
Enables early detection and prediction of illness during pre-symptomatic stages, allowing users to take precautionary measures before symptom onset.
Improves accuracy of illness detection by analyzing multiple physiological parameters and rhythmic biological cycles.
Provides personalized illness risk metrics by using individualized baseline physiological data and cyclical behavioral models.
Incorporates location and demographic data to adjust predictive weightings for increased detection reliability across diverse user populations.
Supports feedback from users to improve classifier performance over time.
Facilitates health management for groups by presenting illness risk metrics to administrators for resource allocation and containment of outbreaks.
Documented Applications
Continuous health monitoring of individual users using wearable devices to predict transitions from healthy to unhealthy states based on nervous system parameters.
Providing early illness warnings and illness risk metrics via graphical user interfaces on user devices.
Generating individualized sleep, readiness, and health scores incorporating physiological and behavioral data for health insight.
Tailoring illness detection models with user-specific biological rhythms such as circadian and menstrual cycles.
Health management platforms delivering aggregated illness risk data to administrators for monitoring cohorts, risk stratification, and intervention planning.
Interested in licensing this patent?