Probability-based usage of multiple estimators of a physiological signal
A physiological signal such as a heart rate acquired from a monitoring device is processed to reduce interference, ambiguity, or artifacts arising during various activities. For example, the system can process a physiological signal to account for motion artifacts in the physiological signal and, thus, reduce the impact of movement on the physiological signal. Additionally, or alternatively, the system can process a physiological signal based on one or more measurement contexts associated with a wearable device. In general, the physiological signal processed as described herein can be useful as a reliable, continuous indication of a physiological parameter and, thus, can serve as the basis for other physiological assessments (e.g., heart rate variability) derived from the physiological parameter.
1. A method for concurrent use of multiple heart rate estimation techniques, the method comprising:
providing a plurality of heart rate estimators for estimating a heart rate, each one of the plurality of heart rate estimators optimized for one of a number of predetermined measurement contexts for measuring the heart rate with a wearable physiological monitor;
providing a plurality of probability estimators, each one of the plurality of probability estimators corresponding to one of the plurality of heart rate estimators, and each one of the plurality of probability estimators providing a likelihood that a corresponding heart rate estimator is accurately estimating the heart rate based on a physiological signal from the wearable physiological monitor;
acquiring a physiological signal from the wearable physiological monitor over an interval having segments;
for each segment of the interval, assigning one or more selected ones of the plurality of heart rate estimators to the segment according to the likelihood of accurately estimating the heart rate in the segment; and
providing a heart rate signal over the interval based upon the physiological signal and the selected ones of the plurality of heart rate estimators.
2. The method of claim 1 , wherein assigning one or more selected ones of the plurality of heart rate estimators includes fusing multiple heart rate estimators to create a fused estimator and applying the fused estimator to determine a heart rate for the segment.
3. The method of claim 1 , further comprising determining a heart rate variability over the interval based upon the heart rate signal.
4. The method of claim 1 , wherein the heart rate signal is a heart rate variability signal.
5. The method of claim 1 , wherein the number of predetermined measurement contexts include one or more of an indoor activity, an outdoor activity, one or more physiological parameters, and an ambient weather condition.
6. The method of claim 1 , wherein the number of predetermined measurement contexts include one or more of active, sedentary, and sleeping.
7. The method of claim 1 , wherein the number of predetermined measurement contexts include one or more types of physical exercise.
8. The method of claim 1 , wherein the number of predetermined measurement contexts include one or more types of motion of the wearable physiological monitor based on motion data from one or more motion sensors in the wearable physiological monitor.
9. The method of claim 1 , wherein the number of predetermined measurement contexts include a current heart rate estimated for the wearable physiological monitor.
10. The method of claim 1 , wherein the heart rate estimators include a frequency domain peak detector for detecting at least one of a fundamental peak and one or more harmonic peaks.
11. The method of claim 1 , wherein the heart rate estimators include a fundamental frequency of a harmonic product spectrum for the physiological signal.
12. The method of claim 1 , further comprising creating a preliminary estimate of a most probable heart rate based on one or more factors, and after creating the preliminary estimate, applying the plurality of heart rate estimators based on the probability estimators to increase a probability of an accurate estimate.
13. The method of claim 12 , wherein the preliminary estimate is based on one or more of a prior heart rate, a subsequent heart rate, and a history for a user of the wearable physiological monitor.
14. The method of claim 12 , wherein the preliminary estimate is filtered to remove one or more motion artifacts before applying the plurality of probability estimators.
15. The method of claim 1 , wherein the heart rate estimators include at least one estimator with a high-Q time domain filter for the physiological signal around a predetermined frequency of interest.
16. The method of claim 1 , wherein the heart rate estimators include an estimator that tracks peaks in the physiological signal relative to a motion signal.
17. A computer program product comprising non-transitory computer executable code embodied in a non-transitory computer-readable medium that, when executing on one or more computing devices, performs the steps of:
providing a plurality of heart rate estimators for estimating a heart rate, each one of the plurality of heart rate estimators corresponding to one of a number of predetermined measurement contexts for measuring the heart rate with a wearable physiological monitor;
providing a plurality of probability estimators, each one of the plurality of probability estimators corresponding to one of the plurality of heart rate estimators, and each one of the plurality of probability estimators providing a likelihood that a corresponding heart rate estimator is accurately estimating the heart rate based on a physiological signal from the wearable physiological monitor;
acquiring a physiological signal from the wearable physiological monitor over an interval having segments;
for each segment, assigning one or more selected ones of the plurality of heart rate estimators to the segment based upon the likelihood of accurately estimating the heart rate over in the segment; and
providing a heart rate signal over the interval based upon the physiological signal and the selected ones of the plurality of heart rate estimators.
18. The computer program product of claim 17 , wherein the segments are discrete segments of the interval.
19. The computer program product of claim 17 , wherein the heart rate signal is continuously provided over the interval.
20. The computer program product of claim 17 , wherein the number of predetermined measurement contexts include one or more types of physical activity.
21. The computer program product of claim 17 , wherein the plurality of heart rate estimators include at least one or more of a frequency domain peak detector, a fundamental frequency of a harmonic product spectrum for the physiological signal, a fundamental frequency of a complex cepstrum for the physiological signal, a high-Q time domain filter for the physiological signal around a predetermined frequency of interest, and an estimator that tracks peaks in the physiological signal relative to a motion signal.
22. A system comprising:
a memory configured to store data corresponding to a physiological signal over an interval, the physiological signal acquired by a wearable physiological monitor; and
a server configured to assign, for each segment of the interval, one or more selected heart rate estimators of a plurality of heart rate estimators based on a likelihood of accurately estimating the heart rate in the segment and to provide a continuous heart rate signal over the interval based upon the physiological signal and the one or more selected heart rate estimators, each heart rate estimator of the plurality of heart rate estimators optimized for one of a number of predetermined measurement contexts for measuring the heart rate with the wearable physiological monitor.
23. The system of claim 22 , wherein the likelihood of accurately estimating the heart rate in the segment is based on a plurality of probability estimators, each one of the plurality of probability estimators corresponding to one of the plurality of heart rate estimators, and each one of the plurality of probability estimators providing a likelihood that a corresponding heart rate estimator is accurately estimating the heart rate based on the physiological signal from the wearable physiological monitor.
24. The system of claim 22 , wherein the plurality of heart rate estimators include at least one or more of a frequency domain peak detector, a fundamental frequency of a harmonic product spectrum for the physiological signal, a fundamental frequency of a complex cepstrum for the physiological signal, a high-Q time domain filter for the physiological signal around a predetermined frequency of interest, and an estimator that tracks peaks in the physiological signal relative to a motion signal.