ON-DEMAND HEART RATE ESTIMATION BASED ON OPTICAL MEASUREMENTS

• US 20180000424A1
• Filed: 06/30/2016
• Published: 01/04/2018
• Est. Priority Date: 06/30/2016
• Status: Active Grant

• Current Assignee: Analog Devices, Inc.
• Original Assignee: Analog Devices, Inc.
• Inventors: DEMIRTAS, SEFA, KING, JASON D., ADAMS, ROBERT, AKL, TONY JOSEPH, BERNSTEIN, JEFFREY G.
• Granted Patent: US 10,285,651 B2
• Time in Patent Office: Days
• CPC Class Codes: A61B 5/02416 using photoplethysmograph signals, e.g. generated by infrared radiation A61B5/14552 takes precedence
• CPC Class Codes: A61B 5/1118 Determining activity level
• CPC Class Codes: A61B 5/1123 Discriminating type of movement, e.g. walking or running A61B5/1116, A61B5/112 take precedence
• CPC Class Codes: A61B 5/6802 Sensor mounted on worn items
• CPC Class Codes: A61B 5/681 Wristwatch-type devices
• CPC Class Codes: A61B 5/7207 of noise induced by motion artifacts
• CPC Class Codes: A61B 5/7221 Determining signal validity, reliability or quality A61B5/7207 and A61B5/7217 take precedence
• CPC Class Codes: A61B 5/7257 using Fourier transforms
• CPC Class Codes: G16H 50/30 for calculating health indices; for individual health risk assessment
• CPC Class Codes: G16H 50/70 for mining of medical data, e.g. analysing previous cases of other patients

Abstract

Activity monitors and smart watches utilizing optical measurements are becoming widely popular, and users expect to get an increasingly accurate estimate of their heart rate (HR) from these devices. These devices are equipped with a light source and an optical sensor which enable estimation of HR using a technique called photoplethysmography (PPG). One of the main challenges of HR estimation using PPG is the coupling of motion into the optical PPG signal when the user is moving randomly or exercising. The present disclosure describes a computationally feasible and fast HR estimation algorithm to be executed at instances of little or no motion. Resulting HR readings may be useful on their own, or be provided to systems that monitor HR continuously to prevent the problem of such systems being locked on an incorrect HR for long periods of time. Implementing techniques described herein leads to more accurate HR measurements.

Claims

1. A computer-implemented method for determining a heart rate from a heartbeat signal present in a first signal generated by a heartbeat sensor, the method comprising:

computing an autocorrelation of data samples of the first signal over a first time period;

processing a second signal to evaluate motion of the heartbeat sensor during the first time period, the second signal indicative of a motion of the heartbeat sensor with respect to one or more directions;

determining whether the data samples of the first signal over the first time period pass a first set of confidence checks, the first set of confidence checks comprising at least

determining whether the motion of the heartbeat sensor during the first time period satisfies a motion-related criteria;

upon determination that the data samples of the first signal over the first time period pass the first set of confidence checks, computing a first estimate of the heart rate from the autocorrelation of the data samples of the first signal over the first time period;

computing a Discrete Fourier Transform (DFT) for data samples of the first signal over a second time period;

computing a second estimate of the heart rate from the DFT; and

determining the heart rate based on the first estimate and the second estimate.

2. The method according to claim 1, wherein the second time period is greater than and includes the first time period.

3. The method according to claim 2, further comprising determining whether the data samples of the first signal over the second time period pass the first set of confidence checks, wherein the DFT is computed upon determination that the data samples of the first signal over the second time period pass the first set of confidence checks.

4. The method according to claim 3, further comprising computing a series of autocorrelations of the data samples of the first signal within the second time period, each autocorrelation computed over a period that is equal to the first time period,

wherein the first set of confidence checks further comprises determining whether the outcome of the each autocorrelation satisfies an autocorrelation-related criteria.

5. The method according to claim 4, wherein computing the autocorrelation of the data samples of the first signal over the second time period comprises performing iterations of adding a contribution due to a new data sample to an autocorrelation of a previous iteration and subtracting a contribution due to a former data sample for each new data sample of data samples between a last data sample of the data samples of the first signal over the first time period and a last data sample of the data samples of the first signal over the second time period.

6. The method according to claim 1, further comprising determining whether the data samples of the first signal over the first time period pass a second set of confidence checks, wherein the DFT is computed upon determination that the data samples of the first signal over the first time period pass the second set of confidence checks.

7. The method according to claim 6, further comprising determining an expected number of local peaks in the autocorrelation of the data samples of the first signal over the first time period based on the first heart rate estimate,

wherein the second set of confidence checks comprises determining whether the expected number of local peaks is equal to a number of local peaks present in the autocorrelation of the data samples of the first signal over the first time period.

8. The method according to claim 1, further comprising computing a series of autocorrelations of the data samples of the first signal within the second time period, each autocorrelation computed over a period that is equal to the first time period,

wherein the first set of confidence checks further comprises determining whether the outcome of the each autocorrelation satisfies autocorrelation-related criteria or a predetermined number of autocorrelations satisfy autocorrelation-related criteria.

9. The method according to claim 8, determining whether the each autocorrelation satisfies the autocorrelation-related criteria is carried out only for a portion of the each autocorrelation for which an autocorrelation lag is between 0.25 seconds and 2 seconds.

10. The method according to claim 9, wherein the autocorrelation-related criteria comprise determining whether an autocorrelation peak that occurs first within the portion of the each autocorrelation is larger than any other autocorrelation value within the portion of the each autocorrelation.

11. The method according to claim 9, wherein the autocorrelation-related criteria comprises determining whether autocorrelation peaks occur within the portion of the each autocorrelation in order of decreasing values.

12. The method according to claim 1, further comprising:

using the data samples of the second signal to determine a parameter indicative of the motion of the heartbeat sensor,

wherein the motion-related criteria comprises determining whether the parameter is below a predefined threshold.

13. The method according to claim 1, further comprising determining whether an absolute value of the DFT has a local peak in a feasible range of heart rates, wherein the second estimate of the heart rate is computed upon determination that the absolute value of the DFT has the local peak in the feasible range of heart rates.

14. The method according to claim 1, wherein determining the heart rate based on the first estimate and the second estimate comprises:

determining whether the first estimate and the second estimate are within a predefined number of beats per minute (bpm) from one another, and

upon determining that the first estimate and the second estimate are within the predefined number of bpm from one another, determining the heart rate as equal to one of the first estimate, the second estimate, or an estimate computed based on the first estimate and the second estimate.

15. The method according to claim 1, further comprising decimating the data samples of the first signal over the second time period prior to computing the DFT and computing the DFT on the decimated data samples.

16. The method according to claim 1, further comprising:

processing the first signal and/or the second signal with a pre-processing filter to substantially attenuate signal content outside of a reasonable frequency band of interest corresponding to a range of reasonable heart rates.

17. The method according to claim 16, wherein:

the pre-processing filter is a low-pass filter or a band-pass filter; and

the reasonable frequency band of interest comprises frequencies between 0.5 Hertz and 4 Hertz.

18. An apparatus for determining a heart rate from a heartbeat signal present in a first signal generated by a heartbeat sensor, the apparatus comprising:

at least one memory configured to store computer executable instructions, and

at least one processor coupled to the at least one memory and configured, when executing the instructions, to;

compute an autocorrelation function (AF) of data samples of the first signal over a first time period;

compute an squared difference function (SDF) of the data samples of the first signal over the first time period;

compute a combination function representative of the computed autocorrelation function and the computed squared difference function;

process a second signal to evaluate motion of the heartbeat sensor during the first time period, the second signal indicative of a motion of the heartbeat sensor with respect to one or more directions;

determine whether the data samples of the first signal over the first time period pass a first set of confidence checks, the first set of confidence checks comprising at least determining whether the motion of the heartbeat sensor during the first time period satisfies a motion-related criteria;

upon determination that the data samples of the first signal over the first time period pass the first set of confidence checks, compute a first estimate of the heart rate from the combination function;

compute a Discrete Fourier Transform (DFT) for data samples of the first signal over a second time period;

compute a second estimate of the heart rate from the DFT; and

determine the heart rate based on the first estimate and the second estimate.

19. The apparatus according to claim 18, wherein combination function is computed as, for each time lag x, a value of the combination function CF(x) is computed as:

1−

SDF(x)/(SDF(x)+2*AF(x)), wherein SDF(x) is a value of the SDF for the time lag x and AF(x) is a value of the AF for the time lag x.

20. A computer-implemented method for determining a heart rate from a heartbeat signal present in a first signal generated by a heartbeat sensor, the method comprising:

processing a second signal to evaluate motion of the heartbeat sensor during a first time period, the second signal indicative of a motion of the heartbeat sensor with respect to one or more directions;

determining whether data samples of the first signal over the first time period pass a first set of confidence checks, the first set of confidence checks comprising at least

determining whether the motion of the heartbeat sensor during the first time period satisfies a motion-related criteria;

upon determination that the data samples of the first signal over the first time period pass the first set of confidence checks,

computing a first estimate of the heart rate from the data samples of the first signal over the first time period using a first heart rate estimation algorithm,

computing a second estimate of the heart rate from the data samples of the first signal over the first time period using a second heart rate estimation algorithm, and

determining the heart rate based on the first estimate and the second estimate.