# US 11,439,343 B1 · Non-contact identification of sleep and wake periods for elderly care > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 11,439,343 B1 - Document ID: US11439343B1 - Status: Granted - Kind: B1 - Application number: 16/856,116 - Filed: 2020-04-23 - Granted: 2022-09-13 - Earliest claimed: 2019-05-17 - Assignees: Tellus You Care, Inc. - Applicants: Tellus You Care, Inc. - Inventors: Huiyuan Tan; Kevin Hsu; Tania Abedian Coke - Web page: https://patentgrep.com/patent/US11439343B1 ## CPC - A61B 5/4809: {Sleep detection, i.e. determining whether a subject is asleep or not} - G08B 21/0423: {detecting deviation from an expected pattern of behaviour or schedule} - A61B 5/1115: {Monitoring leaving of a patient support, e.g. a bed or a wheelchair} - A61B 5/150809: {by audible feedback} - A61B 5/7267: {involving training the classification device} - G08B 21/06: indicating a condition of sleep, e.g. anti-dozing alarms - A61B: DIAGNOSIS; SURGERY; IDENTIFICATION (usefulness limited to only animals A61D) - G08B: SIGNALLING SYSTEMS, e.g. PERSONAL CALLING SYSTEMS; ORDER TELEGRAPHS; ALARM SYSTEMS ## Abstract Determining sleep patterns of a user includes detecting a plurality of point clouds, each corresponding to a different position of the user at different times, forming a plurality of bounding boxes, each corresponding to coordinates of captured points of one of the point clouds, creating a wake/sleep classifier based on features of the point clouds, determining sleep positions of the user as a function of time based on the bounding boxes, and determining sleep patterns of the user based on the sleep positions of the user and on results of the sleep/wake classifier. Detecting a plurality of point clouds may include using a tracking device to capture movements of the user. The features of the point clouds may include intermediate data that is determined using scalar velocities of points in the point clouds, absolute velocities of points in the point clouds, and/or counts of points in the point clouds. ## Claims ### Claim 1 A method of determining sleep patterns of a user, comprising: detecting a plurality of point clouds, each corresponding to a different position of the user at different times; forming a plurality of bounding boxes, each corresponding to coordinates of captured points of one of the point clouds; creating a wake/sleep classifier based on features of the point clouds; determining sleep positions of the user as a function of time based on the bounding boxes; and determining sleep patterns of the user based on the sleep positions of the user and on results of the sleep/wake classifier. ### Claim 2 (depends on 1) A method, according to claim 1, wherein detecting a plurality of point clouds includes using a tracking device to capture movements of the user. ### Claim 3 (depends on 2) A method, according to claim 2, wherein the tracking device uses radar. ### Claim 4 (depends on 1) A method, according to claim 1, wherein the features of the point clouds include intermediate data that is determined using at least one of: scalar velocities of points in the point clouds, absolute velocities of points in the point clouds, and counts of points in the point clouds. ### Claim 5 (depends on 4) A method, according to claim 4, wherein at least some of the features are filtered according to distance from a tracking device that is used to detect the plurality of point clouds. ### Claim 6 (depends on 5) A method, according to claim 5, wherein the intermediate data includes at least one of: a bag of point counts corresponding to a set of point counts at a series of sequential time frames, a bag of velocities corresponding to a set of point velocities at a series of sequential time frames, and a bag of absolute velocities corresponding to a set of absolute velocities at a series of sequential time frames. ### Claim 7 (depends on 6) A method, according to claim 6, wherein a set of aggregating, scaling and filtering functions are applied to the intermediate data to provide short-term feature aggregation values and mid-term feature aggregation values. ### Claim 8 (depends on 7) A method, according to claim 7, wherein the short-term feature aggregation values are determined based on time slots corresponding to a number of sequential time frames. ### Claim 9 (depends on 8) A method, according to claim 8, wherein the feature aggregation values include at least one of: mean values, median values, sum of values, minimum values and maximum values, and scaling function include logarithmic scaling function values. ### Claim 10 (depends on 8) A method, according to claim 8, wherein the mid-term feature aggregation values are derived from the short-term feature aggregation values. ### Claim 11 (depends on 10) A method, according to claim 10, wherein the mid-term feature aggregation values are determined based on epochs that represent contiguous collections of time slots. ### Claim 12 (depends on 1) A method, according to claim 1, wherein the features of the point clouds are used with truth information as training data for machine learning to provide an assessment of relative feature importance. ### Claim 13 (depends on 12) A method, according to claim 12, wherein the assessment of relative feature importance is determined using random forest machine learning. ### Claim 14 (depends on 1) A method, according to claim 1, wherein determining sleep positions includes determining if a breathing direction of the user is vertical or horizontal. ### Claim 15 (depends on 14) A method, according to claim 14, wherein, if the breathing direction is vertical, the sleep position is determined to be that the user is lying on the back of the user in response to a heart area of the user being detected on a left side of the user and the sleep position is determined to be that the user is lying on the stomach of the user in response to the heart area of the user being detected on a right side of the user. ### Claim 16 (depends on 14) A method, according to claim 14, wherein, if the breathing direction is horizontal, the sleep position is determined to be that the user is lying on a left of the user in response to a heart area of the user being detected in a relatively lower disposition and the sleep position is determined to be that the user is lying on a right side of the user in response to the heart area of the user being detected in a relatively upper disposition. ### Claim 17 (depends on 1) A method, according to claim 1, wherein sleep patterns of the user are determined based on correspondence of the sleep positions of the user with the results of the sleep/wake classifier as a function of time. ### Claim 18 (depends on 17) A method, according to claim 17, further comprising: tracking daily sleep patterns for the user. ### Claim 19 (depends on 18) A method, according to claim 18, further comprising: detecting a significant deviation from the daily sleep patterns. ### Claim 20 (depends on 19) A method, according to claim 19, further comprising: providing an alarm in response to detecting the significant deviation from the daily sleep patterns. ### Claim 21 A non-transitory computer readable medium containing software that determines sleep patterns of a user, the software comprising: executable code that detects a plurality of point clouds, each corresponding to a different position of the user at different times; executable code that forms a plurality of bounding boxes, each corresponding to coordinates of captured points of one of the point clouds; executable code that creates a wake/sleep classifier based on features of the point clouds; executable code that determines sleep positions of the user as a function of time based on the bounding boxes; and executable code that determines sleep patterns of the user based on the sleep positions of the user and on results of the sleep/wake classifier, wherein the features of the point clouds include intermediate data that is determined using at least one of: scalar velocities of points in the point clouds, absolute velocities of points in the point clouds, and counts of points in the point clouds. ### Claim 22 A non-transitory computer readable medium containing software that determines sleep patterns of a user, the software comprising: executable code that detects a plurality of point clouds, each corresponding to a different position of the user at different times; executable code that forms a plurality of bounding boxes, each corresponding to coordinates of captured points of one of the point clouds; executable code that creates a wake/sleep classifier based on features of the point clouds; executable code that determines sleep positions of the user as a function of time based on the bounding boxes; and executable code that determines sleep patterns of the user based on the sleep positions of the user and on results of the sleep/wake classifier, wherein determining sleep positions includes determining if a breathing direction of the user is vertical or horizontal.