Estimating heart rate by tracking optical signal frequency components

• US 10,342,441 B2
• Filed: 02/01/2016
• Issued: 07/09/2019
• Est. Priority Date: 02/27/2015
• Status: Active Grant

• Current Assignee: Qualcomm, Inc.
• Original Assignee: Qualcomm, Inc.
• Inventors: Puig, Carlos, Samadani, Ramin, Martin, Russel
• Primary Examiner(s): Remaly, Mark D
• Application Number:
  US15/012,673
• Publication Number:
  US 20160249820A1
• Time in Patent Office: 1,254 Days
• CPC Class Codes: A61B 2562/0219 Inertial sensors, e.g. accelerometers, gyroscopes, tilt switches
• CPC Class Codes: A61B 5/02416 using photoplethysmograph signals, e.g. generated by infrared radiation A61B5/14552 takes precedence
• CPC Class Codes: A61B 5/02438 with portable devices, e.g. worn by the patient
• CPC Class Codes: A61B 5/681 Wristwatch-type devices
• CPC Class Codes: A61B 5/721 using a separate sensor to detect motion or using motion information derived from signals other than the physiological signal to be measured
• CPC Class Codes: A61B 5/7257 using Fourier transforms

Abstract

Methods, systems, computer-readable media, and apparatuses for estimating a user'"'"'s heart rate using a PG signal are presented. In some implementations, the heart rate is estimated by computing a frequency-domain PG, identifying one or more features in the frequency-domain PG, selecting qualified features from the one or more features, and constructing one or more traces. In some implementations, an accelerometer signal can be used for motion cancellation to eliminate traces that are motion artifacts.

Claims

1. A method of estimating heart rate of a user comprising:

obtaining a plethysmogram (PG) signal;

computing a frequency-domain PG based on the obtained PG signal, wherein the frequency-domain PG is an amplitude-versus-frequency representation of the PG signal;

identifying a plurality of features from the frequency-domain PG, wherein the plurality of features are peaks or troughs;

applying a set of criteria to identify a subset of the plurality of features as being qualified features, wherein the set of criteria are applied to the frequency-domain PG and include an amplitude based threshold and a bandwidth based threshold, and wherein applying the set of criteria comprises selecting, as qualified features, those features that meet the amplitude based threshold and the bandwidth based threshold;

after identifying the qualified features, constructing a plurality of traces in a frequency-versus-time mapping using the qualified features, each qualified feature representing a position in the frequency-versus-time mapping, wherein each trace comprises a group of related positions, and wherein the constructing of the plurality of traces comprises, for each qualified feature;

adding the qualified feature to any trace whose most recent frequency differs from the qualified feature'"'"'s frequency by less than a threshold frequency difference; and

starting a new trace using the qualified feature when there are no existing traces for which the most recent frequency differs from the qualified feature'"'"'s frequency by less than the threshold frequency difference;

wherein each trace is started as a new trace using a single qualified feature, and wherein at least one of the traces includes an additional qualified feature added based on a most recent frequency of the trace differing from a frequency of the additional qualified feature by less than the threshold frequency difference;

selecting a trace from the plurality of traces as representative of the heart rate for a given period of time; and

using the selected trace to estimate the heart rate of the user.

2. The method of claim 1, wherein each of the plurality of features corresponds to a peak.

3. The method of claim 1, further comprising:

calculating a frequency-domain accelerometer (ACL) signal based on an obtained ACL signal;

identifying a plurality of ACL features from the frequency-domain ACL signal;

selecting qualified ACL features from the plurality of ACL features from the frequency-domain ACL signal;

constructing ACL traces in the frequency-versus-time mapping using the qualified ACL features, each qualified ACL feature representing an ACL position in the frequency-versus-time mapping, wherein an ACL trace comprises a group of related ACL positions;

determining whether to eliminate one or more PG traces based at least in part on the ACL traces; and

upon determining that one or more PG traces should be eliminated, eliminating the one or more PG traces.

4. The method of claim 1, further comprising:

maintaining a set of active traces, wherein maintaining the set of active traces comprises removing from the plurality of traces any trace whose trace age is older than a maximum age.

5. The method of claim 1, further comprising estimating a quality metric for the heart rate.

6. The method of claim 1, further comprising modifying the estimated heart rate based on prior information.

7. The method of claim 6, wherein the prior information comprises known probabilities of pairs of accelerometer power and heart rate for a specific demographic of population.

8. The method of claim 1, wherein the heart rate is estimated at a given time for a requested time, wherein the requested time is earlier than the given time.

9. The method of claim 3, wherein the determining whether to eliminate one or more PG traces comprises:

eliminating PG traces if an inter-trace distance between an ACL trace and a PG trace is lesser than a threshold distance.

10. The method of claim 8, wherein estimating the heart rate for a requested time comprises back-tracking through a heart rate trace to the requested time.

11. The method of claim 1, wherein the frequency-domain PG is computed by applying a Fourier Transform to the PG signal, the amplitudes of the frequency-domain PG corresponding to Fourier Transform values, and wherein the plurality of traces are constructed by processing the qualified features in an order determined based on the Fourier Transform values.

12. The method of claim 1, wherein the frequency-domain PG comprises a power spectrum of the PG signal, wherein the qualified features are peaks, and wherein each qualified feature is a peak for which the ratio of the peak'"'"'s amplitude to a maximum amplitude of the power spectrum of the PG signal exceeds the amplitude based threshold.

13. The method of claim 12, wherein each qualified feature is a peak for which the peak'"'"'s half-power bandwidth is less than the bandwidth based threshold.

14. The method of claim 1, wherein selecting the trace from the plurality of traces comprises:

ranking the plurality of traces based on at least one of length or average amplitude, wherein a highest ranking trace is selected as representative of the heart rate.

15. A mobile device comprising:

a sensor configured to generate a plethysmogram (PG) signal based on a condition of a user; and

a processor configured to;

compute a frequency domain PG signal based on the generated PG signal, wherein the frequency-domain PG is an amplitude-versus-frequency representation of the PG signal;

identify a plurality of features from the frequency-domain PG signal, wherein the plurality of features are peaks or troughs;

apply a set of criteria to identify a subset of the plurality of features as being qualified features, wherein the set of criteria are applied to the frequency-domain PG and include an amplitude based threshold and a bandwidth based threshold, and wherein applying the set of criteria comprises selecting, as qualified features, those features that meet the amplitude based threshold and the bandwidth based threshold;

after identifying the qualified features, construct a plurality of traces in a frequency-versus-time mapping using the qualified features, each qualified feature representing a position in the frequency-versus-time mapping, wherein each trace comprises a group of related positions, and wherein the constructing of the plurality of traces comprises, for each qualified feature;

adding the qualified feature to any trace whose most recent frequency differs from the qualified feature'"'"'s frequency by less than a threshold frequency difference; and

starting a new trace using the qualified feature when there are no existing traces for which the most recent frequency differs from the qualified feature'"'"'s frequency by less than the threshold frequency difference;

wherein each trace is started as a new trace using a single qualified feature, and wherein at least one of the traces includes an additional qualified feature added based on a most recent frequency of the trace differing from a frequency of the additional qualified feature by less than the threshold frequency difference;

select a trace from the plurality of traces as representative of the heart rate for a given period of time; and

use the selected trace to estimate the heart rate of the user.

16. The mobile device of claim 15, wherein each of the plurality of features corresponds to a peak.

17. The mobile device of claim 15

wherein the computing the frequency-domain PG signal based on the generated PG signal comprises applying a Fast Fourier Transform to the generated PG signal.

18. The mobile device of claim 15, further comprising:

an accelerometer configured to generate an accelerometer (ACL) signal based on a motion of the user;

wherein the processor is further configured to;

calculate a frequency-domain ACL signal based on the generated ACL signal;

identify a plurality of ACL features from the frequency-domain ACL signal;

select qualified ACL features from the plurality of ACL features from the frequency-domain ACL signal;

construct ACL traces in the frequency versus time mapping using the qualified ACL features, each qualified ACL feature representing an ACL position in the frequency-versus-time mapping, wherein an ACL trace comprises a group of related ACL positions;

determine whether to eliminate one or more PG traces based at least in part on the ACL traces; and

upon determining that one or more PG traces should be eliminated, eliminate the one or more PG traces.

19. The mobile device of claim 15, wherein the processor is further configured to maintain a set of active traces, wherein maintaining the set of active traces comprises removing from the plurality of traces any trace whose trace age is older than a maximum age.

20. The mobile device of claim 15, wherein the processor is further configured to estimate a quality metric for the heart rate.

21. The mobile device of claim 15, wherein the processor is further configured to modify the estimate heart rate based on prior information, and wherein the prior information comprises known probabilities of pairs of accelerometer power and heart rate for a specific demographic of population.

22. The mobile device of claim 15, wherein the processor is further configured to estimate the heart rate at a given time for a requested time, wherein the requested time is earlier than the given time.

23. The mobile device of claim 18, wherein the determining whether to eliminate one or more PG traces comprises:

eliminating PG traces if an inter-trace distance between an ACL trace and a PG trace is lesser than a threshold distance.

24. An apparatus comprising:

means to generate a plethysmogram (PG) signal based on a condition of a user; and

processing means comprising;

means to compute a frequency domain PG signal based on the generated PG signal, wherein the frequency-domain PG is an amplitude-versus-frequency representation of the PG signal;

means to identify a plurality of features from the frequency-domain PG signal, wherein the plurality of features are peaks or troughs;

means to apply a set of criteria to identify a subset of the plurality of features as being qualified features, wherein the set of criteria are applied to the frequency-domain PG and include an amplitude based threshold and a bandwidth based threshold, and wherein applying the set of criteria comprises selecting, as qualified features, those features that meet the amplitude based threshold and the bandwidth based threshold;

means to, after identifying the qualified features, construct a plurality of traces in a frequency-versus-time mapping using the qualified features, each qualified feature representing a position in the frequency-versus-time mapping, wherein each trace comprises a group of related positions, and wherein the constructing of the plurality of traces comprises, for each qualified feature;

adding the qualified feature to any trace whose most recent frequency differs from the qualified feature'"'"'s frequency by less than a threshold frequency difference; and

starting a new trace using the qualified feature when there are no existing traces for which the most recent frequency differs from the qualified feature'"'"'s frequency by less than the threshold frequency difference;

wherein each trace is started as a new trace using a single qualified feature, and wherein at least one of the traces includes an additional qualified feature added based on a most recent frequency of the trace differing from a frequency of the additional qualified feature by less than the threshold frequency difference;

means to select a trace from the plurality of traces as representative of the heart rate for a given period of time; and

means to use the selected trace to estimate the heart rate of the user.

25. The apparatus of claim 24, further comprising:

means to generate an accelerometer (ACL) signal based on a motion of the user;

wherein the processing means further comprises;

means to calculate a frequency-domain ACL signal based on the generated ACL signal;

means to identify a plurality of ACL features from the frequency-domain ACL signal;

means to select qualified ACL features from the plurality of ACL features from the frequency-domain ACL signal;

means to construct ACL traces in the frequency versus time mapping using the qualified ACL features, each qualified ACL feature representing an ACL position in the frequency-versus-time mapping, wherein an ACL trace comprises a group of related ACL positions;

means to determine whether to eliminate one or more PG traces based at least in part on the ACL traces; and

upon determining that one or more PG traces should be eliminated, means to eliminate the one or more PG traces.

26. The apparatus of claim 24, wherein the processing means further comprises means to modify the estimate heart rate based on prior information.

27. The apparatus of claim 24, wherein the processing means further comprises means to estimate the heart rate at a given time for a requested time, wherein the requested time is earlier than the given time.

28. A non-transitory computer-readable medium having stored instructions thereon, which when executed by a processor, perform a method comprising:

obtaining a plethysmogram (PG) signal;

computing a frequency-domain PG based on the obtained PG signal, wherein the frequency-domain PG is an amplitude-versus-frequency representation of the PG signal;

identifying a plurality of features from the frequency-domain PG, wherein the plurality of features are peaks or troughs;

applying a set of criteria to identify a subset of the plurality of features as being qualified features, wherein the set of criteria are applied to the frequency-domain PG and include an amplitude based threshold and a bandwidth based threshold, and wherein applying the set of criteria comprises selecting, as qualified features, those features that meet the amplitude based threshold and the bandwidth based threshold;

after identifying the qualified features, constructing a plurality of traces in a frequency-versus-time mapping using the qualified features, each qualified feature representing a position in the frequency-versus-time mapping, wherein each trace comprises a group of related positions, and wherein the constructing of the plurality of traces comprises, for each qualified feature;

adding the qualified feature to any trace whose most recent frequency differs from the qualified feature'"'"'s frequency by less than a threshold frequency difference; and

starting a new trace using the qualified feature when there are no existing traces for which the most recent frequency differs from the qualified feature'"'"'s frequency by less than the threshold frequency difference;

wherein each trace is started as a new trace using a single qualified feature, and wherein at least one of the traces includes an additional qualified feature added based on a most recent frequency of the trace differing from a frequency of the additional qualified feature by less than the threshold frequency difference;

selecting a trace from the plurality of traces as representative of the heart rate for a given period of time; and

using the selected trace to estimate the heart rate of the user.