Granted Patent B2

US 11,439,344 · App. 16/945,837 · Granted Sep 13, 2022

Inventors: Feng Zhang (Greenbelt, MD); Beibei Wang (Clarksville, MD); Chenshu Wu (Greenbelt, MD); Min Wu (Clarksville, MD); Dan Bugos (Washington, DC); Hangfang Zhang (Greenbelt, MD); K. J. Ray Liu (Potomac, MD); Oscar Chi-Lim Au (San Jose, CA)

Assignee: ORIGIN WIRELESS, INC.

A61B5/4809A61B5/0004A61B5/113A61B5/4818A61B5/6887

Abstract

Claims (176)

1. A system for monitoring a sleep motion of a user in a venue, comprising:

at least one sensor in the venue, wherein the at least one sensor comprises a wireless non-contact sensor having no physical contact with the user;

a processor communicatively coupled to the at least one sensor;

a memory communicatively coupled to the processor; and

a set of instructions stored in the memory which, when executed by the processor, causes the processor to perform:

obtaining, based on the at least one sensor, a plurality of time series of sensing features (TSSF) associated with the sleep motion of the user in the venue, wherein at least one TSSF of the plurality of TSSF is obtained by:

communicating, based on the wireless non-contact sensor, a wireless signal in a wireless multipath channel of the venue,

extracting a time series of channel information (TSCI) of the wireless multipath channel of the venue from the wireless signal, and

obtaining the at least one TSSF based on the TSCI,

wherein obtaining the plurality of TSSF comprises:

computing a first TSSF based on the TSCI, wherein each sensing feature (SF) of the first TSSF comprises one of: a distance score, a Euclidean distance, a similarity score, a correlation, a covariance, or an inner product, based on two adjacent groups of channel information (CI) of the TSCI,

computing a second TSSF comprising periodicity features associated with a periodicity of the sleep motion of the user, wherein the second TSSF is associated with a breathing rate of the user,

computing a time series of trend function of the second TSSF by lowpass filtering the second TSSF,

computing a time series of detrended function of the second TSSF by subtracting the time series of trend function from the second TSSF; and

monitoring the sleep motion of the user jointly based on the plurality of TSSF and the time series of detrended function of the second TSSF,

wherein monitoring the sleep motion of the user comprises:

training a sleep classifier based on data related to breathing rate variance and breathing rate deviation, using machine learning,

computing a first time function of breathing rate variance by computing variance of a detrended breathing rate function within a first sliding time window,

computing a second time function of breathing rate deviation by computing a distance between an average non-rapid-eye-movement (NREM) breathing rate and a percentile of breathing rate within a second sliding time window,

recognizing jointly, based on the sleep classifier, a sleep stage of the user as one of rapid-eye-movement (REM) or NREM based on the first time function of breathing rate variance and the second time function of breathing rate deviation.

2. The system of claim 1 , wherein:

monitoring the sleep motion comprises monitoring at least one of the following of the user: sleep timings, sleep durations, sleep stages, sleep states, sleep quality, sleep apnea, sleep problems, or sleep disorders.

3. The system of claim 1 , wherein the set of instructions, when executed by the processor, further causes the processor to perform:

computing, based on the plurality of TSSF, at least one of: a time series of sleep stages, or a time series of sleep sub-stages, wherein at least one of the sleep stages is related to: rapid-eye-movement (REM) or non-REM (NREM).

4. The system of claim 3 , wherein the set of instructions, when executed by the processor, further causes the processor to perform:

computing each motion feature of the first TSSF based on two adjacent groups of channel information (CI) of the TSCI in a sliding time window, wherein each motion feature comprises at least one of: a distance score, a Euclidean distance, a similarity score, a correlation, a covariance, an inner product, an outer product, or a transformation.

5. The system of claim 3 , wherein the set of instructions, when executed by the processor, further causes the processor to perform:

computing each periodicity feature of the second TSSF based on at least one of the following of channel information (CI) of the TSCI in a sliding time window: an autocorrelation function (ACF), a frequency spectrum, or a frequency transform, wherein:

each periodicity feature is set to a default value when the periodicity is not detected in the sleep motion of the user,

each periodicity feature comprises at least one of: a frequency, a phase, a rate, a frequency index, a time period, or a time index.

6. The system of claim 3 , wherein the set of instructions, when executed by the processor, further causes the processor to perform:

monitoring the sleep motion of the user by recognizing a sleep state of the user jointly based on the first TSSF and the second TSSF;

computing a motion statistics based on a motion feature of the first TSSF, wherein the motion statistics comprises at least one of:

a motion ratio, a percentage of time that a function of the motion feature exceeds a threshold, or

a percentage of time that a function of the motion feature exceeds the threshold in a sliding time window;

computing a periodicity statistics based on a periodicity feature of the second TSSF, wherein the periodicity statistics comprises at least one of:

a periodicity feature ratio, a breathing ratio, a heartbeat ratio,

a percentage of time that the periodicity feature is not a default value (non-default), or

a percentage of time that the periodicity feature is non-default in a sliding time window; and

recognizing the sleep state jointly as either ASLEEP or AWAKE based on the motion statistics and the periodicity statistics.

7. The system of claim 6 , wherein:

the sleep state is recognized jointly as ASLEEP when the motion statistics or the periodicity statistics satisfy a first joint criterion; and

the sleep state is recognized jointly as AWAKE when the motion statistics or the periodicity statistics satisfy a second joint criterion.

8. The system of claim 7 , wherein:

the first joint criterion is that: the motion statistics is less than a first threshold and the periodicity statistics is greater than a second threshold; and

the second joint criterion is that: the motion statistics is greater than the first threshold or the periodicity statistics is less than the second threshold.

9. The system of claim 7 , wherein:

the first joint criterion is that: the motion statistics is less than a first threshold or the periodicity statistics is greater than a second threshold; and

the second joint criterion is that: the motion statistics is greater than the first threshold and the periodicity statistics is less than the second threshold.

10. The system of claim 1 , wherein

the average NREM breathing rate is computed by identifying a peak of a histogram of a time function of breathing rate in an ASLEEP stage in an overnight period; and

the sleep classifier is further trained based on scored polysomnography data.

11. The system of claim 1 , wherein the set of instructions, when executed by the processor, further causes the processor to perform:

computing at least one detrended statistics based on the time series of detrended function of the second TSSF in a respective sliding time window, wherein the at least one detrended statistics comprises at least one of: a mean, a weighted mean, a variance, or a deviation; and

recognizing jointly, based on the at least one detrended statistics, the sleep stage as at least one of: REM, NREM, light sleep, deep sleep, sleep apnea, insomnia, hypersomnia, parasomnia, sleep disruption, nightmare, sleep walking, toss-and-turn, a sleep problem, a sleep condition, or a sleep behavior.

12. The system of claim 1 ,

wherein the machine learning comprises at least one of: supervised learning, unsupervised learning, semi-supervised learning, active learning, reinforcement learning, support vector machine, deep learning, feature learning, clustering, regression, or dimensionality reduction.

13. The system of claim 11 , wherein the set of instructions, when executed by the processor, further causes the processor to perform:

recognizing jointly the sleep stage as at least one of: REM, NREM, light sleep, deep sleep, sleep apnea, insomnia, hypersomnia, parasomnia, sleep disruption, nightmare, sleep walking, toss-and-turn, the sleep problem, the sleep condition, or the sleep behavior, when a sleep state of the user is recognized as ASLEEP.

14. The system of claim 11 , wherein the set of instructions, when executed by the processor, further causes the processor to perform:

computing a sleep analytics based on at least one of: any sleep stage, any sleep state, the first TSSF, the second TSSF, the plurality of TSSF, or any associated timing,

wherein the sleep analytics comprises at least one of: a sleep quality score, a sleep quantity score, a timing, a duration, or a linear combination of ASLEEP periods with positive coefficients and AWAKE periods with negative coefficients.

15. A system for monitoring a sleep motion of a user in a venue, comprising:

at least one sensor in the venue, wherein the at least one sensor comprises a non-contact sensor having no physical contact with the user;

a processor communicatively coupled to the at least one sensor;

a memory communicatively coupled to the processor; and

a set of instructions stored in the memory which, when executed by the processor, causes the processor to perform:

communicating a wireless signal in a wireless multipath channel of the venue in a non-contact manner based on the non-contact sensor;

extracting a time series of channel information (TSCI) of the wireless multipath channel of the venue from the wireless signal;

obtaining, based on the at least one sensor, a plurality of time series of sensing features (TSSF) associated with the sleep motion of the user in the venue, wherein:

the plurality of TSSF comprises a first TSSF and a second TSSF,

each sensing feature (SF) of the first TSSF comprises one of: a distance score, a Euclidean distance, a similarity score, a correlation, a covariance, or an inner product, based on two adjacent groups of channel information (CI) of the TSCI,

the second TSSF comprises periodicity features associated with a periodicity of the sleep motion of the user,

the second TSSF is associated with a breathing rate of the user,

both the first TSSF and the second TSSF are obtained based on the non-contact sensor; and

monitoring the sleep motion of the user jointly based on the plurality of TSSF,

wherein monitoring the sleep motion of the user comprises:

training a sleep classifier based on data related to breathing rate variance and breathing rate deviation, using machine learning,

computing a first time function of breathing rate variance by computing variance of a detrended breathing rate function within a first sliding time window,

computing a second time function of breathing rate deviation by computing a distance between an average non-rapid-eye-movement (NREM) breathing rate and a percentile of breathing rate within a second sliding time window,

recognizing jointly, based on the sleep classifier, a sleep stage of the user as one of rapid-eye-movement (REM) or NREM based on the first time function of breathing rate variance and the second time function of breathing rate deviation.

16. The system of claim 15 , wherein the set of instructions, when executed by the processor, further causes the processor to perform:

communicating a wireless signal in a wireless multipath channel of the venue in a non-contact manner based on the non-contact sensor;

extracting a time series of channel information (TSCI) of the wireless multipath channel of the venue from the wireless signal; and

obtaining the first TSSF and the second TSSF based on the TSCI.

17. The system of claim 15 , wherein the set of instructions, when executed by the processor, further causes the processor to perform:

monitoring the sleep motion of the user by recognizing a sleep state of the user jointly based on the first TSSF and the second TSSF;

computing a motion statistics based on motion features of the first TSSF, wherein the motion statistics comprises at least one of:

a motion ratio, a motion intensity, a motion power, a motion strength, or a percentage of time that the motion feature exceeds a threshold;

computing a periodicity statistics based on periodicity features of the second TSSF, wherein each periodicity feature of the second TSSF is equal to a default value when the periodicity is not detected in the sleep motion of the user, wherein the periodicity statistics comprises at least one of:

a periodicity feature ratio, a breathing ratio, a heartbeat ratio,

a percentage of time that the periodicity feature is not the default value (non-default),

a percentage of time that a breathing rate is non-default, or

a percentage of time that a heartbeat is non-default,

wherein the sleep state is recognized jointly as either ASLEEP or AWAKE based on the motion statistics and the periodicity statistics.

18. The system of claim 15 , wherein the set of instructions, when executed by the processor, further causes the processor to perform:

computing a time series of trend function of the second TSSF by lowpass filtering the second TSSF;

computing a time series of detrended function of the second TSSF by subtracting the time series of trend function from the second TSSF;

computing at least one detrended statistics based on the time series of detrended function of the second TSSF in a respective sliding time window, wherein the at least one detrended statistics comprises: a mean, a weighted mean, a variance, or a deviation; and

monitoring the sleep motion of the user jointly by recognizing jointly a sleep stage of the user as at least one of: REM, NREM, light sleep, deep sleep, sleep apnea, insomnia, hypersomnia, parasomnia, sleep disruption, nightmare, sleep walking, toss-and-turn, a sleep problem, a sleep condition, or a sleep behavior, based on the at least one detrended statistics.

19. The system of claim 15 , wherein the set of instructions, when executed by the processor, further causes the processor to perform:

monitoring the sleep motion of the user jointly by computing a sleep analytics based on at least one of: any sleep stage, any sleep state, the first TSSF, the second TSSF, the plurality of TSSF, or any associated timing,

wherein the sleep analytics comprises at least one of: a sleep quality score, a sleep quantity score, a timing, or a duration.

20. A method of a sleep monitoring system, comprising:

obtaining, based on at least one sensor in a venue, a plurality of time series of sensing features (TSSF) associated with a sleep motion of a user in the venue;

communicating, based on a non-contact sensor having no physical contact with the user, a wireless signal in a wireless multipath channel of the venue;

extracting a time series of channel information (TSCI) of the wireless multipath channel of the venue from the wireless signal, wherein:

the plurality of TSSF comprises a first TSSF and a second TSSF,

the second TSSF comprises periodicity features associated with a periodicity of the sleep motion of the user,

the second TSSF is associated with a breathing rate of the user; and

monitoring the sleep motion of the user jointly based on the plurality of TSSF,

wherein monitoring the sleep motion of the user comprises:

training a sleep classifier based on data related to breathing rate variance and breathing rate deviation, using machine learning,

computing a first time function of breathing rate variance by computing variance of a detrended breathing rate function within a first sliding time window,

computing a second time function of breathing rate deviation by computing a distance between an average non-rapid-eye-movement (NREM) breathing rate and a percentile of breathing rate within a second sliding time window,

recognizing jointly, based on the sleep classifier, a sleep stage of the user as one of rapid-eye-movement (REM) or NREM based on the first time function of breathing rate variance and the second time function of breathing rate deviation.

21. The method of claim 20 , further comprising:

monitoring the sleep motion of the user by recognizing a sleep state of the user jointly based on the first TSSF and the second TSSF;

computing a motion statistics based on motion features of the first TSSF in a first sliding time window, wherein the motion statistics comprises at least one of:

a motion ratio, a motion intensity, a motion power, a motion strength, or

a percentage of time that the motion feature exceeds a threshold;

computing a periodicity statistics based on periodicity features of the second TSSF in a second sliding time window, wherein each periodicity feature of the second TSSF is set to a default value when the periodicity is not detected in the sleep motion of the user,

wherein the periodicity statistics comprises at least one of:

a periodicity feature ratio, a breathing ratio, a heartbeat ratio,

a percentage of time that the periodicity feature is not the default value (non-default),

a percentage of time that a breathing rate is non-default,

a percentage of time that a heartbeat is non-default,

wherein the sleep state is recognized jointly as either ASLEEP or AWAKE based on the motion statistics and the periodicity statistics.

22. The method of claim 20 , further comprising:

computing a time series of trend function of the second TSSF by lowpass filtering the second TSSF;

computing a time series of detrended function of the second TSSF by subtracting the time series of trend function from the second TSSF;

computing at least one detrended statistics based on the time series of detrended function of the second TSSF in a respective sliding time window, wherein the at least one detrended statistics comprises: a mean, a weighted mean, a variance, or a deviation; and

monitoring the sleep motion of the user jointly by recognizing jointly a sleep stage of the user as at least one of: REM, NREM, light sleep, deep sleep, sleep apnea, insomnia, hypersomnia, parasomnia, sleep disruption, nightmare, sleep walking, toss-and-turn, a sleep problem, a sleep condition, or a sleep behavior, based on the at least one detrended statistics.

23. The method of claim 22 ,

wherein the average NREM breathing rate is computed by identifying a peak of a histogram of a time function of breathing rate in an ASLEEP stage in an overnight period.

24. The method of claim 20 , further comprising:

monitoring the sleep motion of the user jointly by computing a sleep analytics based on at least one of: any sleep stage, any sleep state, the first TSSF, the second TSSF, the plurality of TSSF, or any associated timing,

wherein the sleep analytics comprises at least one of: a sleep quality score, a sleep quantity score, a timing, or a duration.

25. A server device of a sleep monitoring system, comprising:

a processor communicatively coupled to at least one sensor in a venue;

a memory communicatively coupled to the processor; and

a set of instructions stored in the memory which, when executed by the processor, causes the processor to perform:

obtaining a plurality of time series of sensing features (TSSF) associated with a sleep motion of a user in the venue, wherein:

each of the plurality of TSSF is associated with a respective sensor of the at least one sensor,

the plurality of TSSF comprises a first TSSF and a second TSSF,

the second TSSF comprises periodicity features associated with a periodicity of the sleep motion of the user,

the second TSSF is associated with a breathing rate of the user,

at least one of the first TSSF or the second TSSF is obtained based on a time series of channel information (TSCI) of a wireless multipath channel of the venue,

the TSCI is extracted from a wireless signal communicated in the wireless multipath channel of the venue by one of the at least one sensor in the venue; and monitoring the sleep motion of the user jointly based on the plurality of TSSF,

wherein monitoring the sleep motion of the user comprises:

training a sleep classifier based on data related to breathing rate variance and breathing rate deviation, using machine learning,

computing a first time function of breathing rate variance by computing variance of a detrended breathing rate function within a first sliding time window,

computing a second time function of breathing rate deviation by computing a distance between an average non-rapid-eye-movement (NREM) breathing rate and a percentile of breathing rate within a second sliding time window,

recognizing jointly, based on the sleep classifier, a sleep stage of the user as one of rapid-eye-movement (REM) or NREM based on the first time function of breathing rate variance and the second time function of breathing rate deviation.

26. The server device of claim 25 , wherein the set of instructions, when executed by the processor, further causes the processor to perform:

monitoring the sleep motion of the user by recognizing a sleep state of the user jointly based on the first TSSF and the second TSSF;

computing a motion statistics based on motion features of the first TSSF in a first sliding time window, wherein the motion statistics comprises at least one of:

a motion ratio, a motion intensity, a motion power, a motion strength, or

a percentage of time that the motion feature exceeds a threshold;

computing a periodicity statistics based on periodicity features of the second TSSF in a second sliding time window, wherein each periodicity feature is set to a default value when the periodicity is not detected in the sleep motion of the user,

wherein the periodicity statistics comprises at least one of:

a periodicity feature ratio, a breathing ratio, a heartbeat ratio,

a percentage of time that the periodicity feature is not the default value (non-default),

a percentage of time that a breathing rate is non-default,

a percentage of time that a heartbeat is non-default,

wherein the sleep state is recognized jointly as either ASLEEP or AWAKE based on the motion statistics and the periodicity statistics.

27. The server device of claim 25 , wherein the set of instructions, when executed by the processor, further causes the processor to perform:

computing a time series of trend function of the second TSSF by lowpass filtering the second TSSF;

computing a time series of detrended function of the second TSSF by subtracting the time series of trend function from the second TSSF;

computing at least one detrended statistics based on the time series of detrended function of the second TSSF in a respective sliding time window, wherein the at least one detrended statistics comprises: a mean, a weighted mean, a variance, or a deviation; and

monitoring the sleep motion of the user jointly by recognizing jointly a sleep stage of the user as at least one of: REM, NREM, light sleep, deep sleep, sleep apnea, insomnia, hypersomnia, parasomnia, sleep disruption, nightmare, sleep walking, toss-and-turn, a sleep problem, a sleep condition, or a sleep behavior, based on the at least one detrended statistics.

28. The server device of claim 27 ,

wherein the average NREM breathing rate is computed by identifying a peak of a histogram of a time function of breathing rate in an ASLEEP stage in an overnight period.

29. The server device of claim 25 , wherein the set of instructions, when executed by the processor, further causes the processor to perform:

monitoring the sleep motion of the user jointly by computing a sleep analytics based on at least one of: any sleep stage, any sleep state, the first TSSF, the second TSSF, the plurality of TSSF, or any associated timing,

wherein the sleep analytics comprises at least one of: a sleep quality score, a sleep quantity score, a timing, or a duration.

Assignments (2)

Continuity (46)

Cited By (2)

US 12,690,806 US 12,713,352