Methods for quantifying the risk of cardiac death using exercise induced heart rate variability metrics

• US 7,708,683 B2
• Filed: 03/01/2007
• Issued: 05/04/2010
• Est. Priority Date: 03/03/2006
• Status: Expired due to Fees

• Current Assignee: Mortara Instrument, Inc. (Baxter International Inc.)
• Original Assignee: Cardiac Science Operating Company (Cardica Incorporated)
• Inventors: Hadley, David M.
• Primary Examiner(s): Manuel; George
• Assistant Examiner(s): Wieland; Robert N
• Application Number:
  US11/681,099
• Publication Number:
  US 20070208266A1
• Time in Patent Office: 1,160 Days
• Field of Search: 340/576, 340/572.1, 600/515, 600/519, 702/181, 702/188, 702/189, 702/193, 375/130, 375/241
• US Class Current: 600/15
• CPC Class Codes: A61B 5/02405 Determining heart rate variability
• CPC Class Codes: A61B 5/30 Input circuits therefor
• CPC Class Codes: A61B 5/349 Detecting specific parameters of the electrocardiograph cycle
• CPC Class Codes: A61B 5/352 Detecting R peaks, e.g. for synchronising diagnostic apparatus; Estimating R-R interval
• CPC Class Codes: A61B 5/364 Detecting abnormal ECG interval, e.g. extrasystoles, ectopic heartbeats
• CPC Class Codes: A61B 5/4884 inducing physiological or psychological stress, e.g. applications for stress testing

Abstract

Methods and apparatus for assessing cardiac risks in a specific patient. One embodiment of a method in accordance with the invention comprises providing heart rate activity of a specific patient including a windowed time series relating to heart rate variability during a heart rate test. This method further includes determining a frequency domain value based on energy values of frequency bands of the heart rate variability in the windowed time series, and/or determining an aggregate power for a frequency band of the windowed time series. This method further includes assessing the risk of a cardiac event based on the frequency value and/or the aggregate power.

Claims

1. A method for assessing cardiac risks in a specific patient based on heart rate variability, comprising:

providing heart activity data of a specific patient including a windowed time series related to heart rate variability during a heart rate test;

determining a frequency domain value based on at least one of energy values of frequency bands and spectral s1ope(s) across selected frequency ranges of the heart activity data in the windowed time series; and

assessing the risk of a cardiac event based on the frequency domain value;

wherein providing heart activity data comprises determining an instantaneous heart rate sequence from an ECG waveform obtained during an exercise period of increasing heart rate and a recovery period of decreasing heart rate, providing a uniform sample series of the instantaneous heart rate sequence, and reducing the uniform sample series by filtering and/or detrending via fitting a polynomial function to the heart activity data across a selected time window to obtain a heart rate variability waveform associated with variances in the heart rate between beats in a desired frequency band.

2. The method of claim 1 wherein:

determining an instantaneous heart rate sequence comprises ascertaining R-R intervals of the ECG waveform; and

reducing the uniform sample series comprises at least one of processing the uniform sample series through a high-pass filter and detrending by subtraction of a fit polynomial function to the heart activity data that provides the heart rate variability waveform for shorter frequencies.

3. The method of claim 1 wherein:

determining an instantaneous heart rate sequence comprises ascertaining R-R intervals of the ECG waveform and expressing the instantaneous heart rate sequence in beats/minute;

providing a uniform sample series of the instantaneous heart rate sequence comprises interpolating between adjacent beats at a uniform sample rate; and

reducing the uniform sample series comprises at least one of processing the uniform sample series through a high-pass filter and detrending by subtraction of a fit polynomial function to the heart activity data that provides the heart rate variability waveform for shorter frequencies.

4. The method of claim 1 wherein reducing the uniform sample series comprise filtering in the forward direction through at least a portion of the exercise period of the heart rate test to approximately an end of a peak exercise phase and reverse filtering through at least a portion of the recovery period to approximately the end of the peak exercise phase.

5. The method of claim 1 wherein detrending comprises fitting a polynomial function to the heart activity data either before or after interpolation for the windowed time series and subtracting the resulting function from the heart activity data.

6. The method of claim 1, further comprising performing a Fourier transform of the windowed time series into a spectrum for frequency domain analysis.

7. A method for assessing cardiac risks in a specific patient based on heart rate variability, comprising:

providing heart activity data of a specific patient including a windowed time series related to heart rate variability during a heart rate test;

determining a frequency domain value based on at least one of energy values of frequency bands and spectral slope(s) across selected frequency ranges of the heart activity data in the windowed time series; and

assessing the risk of a cardiac event based on the frequency domain value;

wherein determining a frequency domain value based on at least one of the energy values comprises (a) fitting a line to the spectral power across a selected frequency range, and (b) determining a slope of the line to obtain a spectral slope, wherein the frequency domain value comprises the spectral slope.

8. A method for assessing cardiac risks in a specific patient based on heart rate variability, comprising:

providing heart activity data of a specific patient including a windowed time series related to heart rate variability during a heart rate test;

determining a frequency domain value based on at least one of energy values of frequency bands and spectral slope(s) across selected frequency ranges of the heart activity data in the windowed time series; and

assessing the risk of a cardiac event based on the frequency domain value;

wherein assessing the risk of a cardiac event comprises comparing the determined frequency domain value with a reference frequency domain value range associated with a population that has experienced the cardiac event.

9. The method of claim 8 wherein providing the heart activity data comprises:

identifying a peak exercise phase and a mid-recovery phase of the heart rate test;

determining instantaneous heart rate sequences for the peak exercise phase and the mid-recovery phase by ascertaining R-R intervals of an ECG waveform of the peak exercise phase and the mid-recovery phase;

providing uniform sample series of the instantaneous heart rate sequences by interpolating between adjacent beats at a uniform sample rate; and

filtering the uniform sample series by high-pass filtering the uniform sample series in a forward direction through at least a portion of the exercise period and reverse filtering through at least a portion of the recovery period to provide the heart rate variability waveform for shorter frequencies during the peak exercise phase and the mid-recovery phase.

10. The method of claim 9 wherein:

for the peak exercise phase, the energy values comprise a peak exercise low frequency average power, a peak exercise middle frequency average power, and a peak exercise high frequency average power of the windowed time series, and wherein determining the frequency domain value comprises determining a peak exercise frequency domain value by (a) fitting a peak exercise line to the peak exercise low frequency average power, the peak exercise middle frequency average power, and peak exercise the high frequency average power, and (b) determining a slope of the peak exercise line to obtain a peak exercise spectral slope, wherein the peak exercise frequency domain value comprises the peak exercise spectral slope; and

for the mid-recovery phase, the energy values further comprise a mid-recovery low frequency average power, a mid-recovery middle frequency average power, and a mid-recovery high frequency average power of the windowed time series, and wherein the method further comprises determining a mid-recovery frequency domain value by (a) fitting a mid-recovery line to the mid-recovery low frequency average power, the mid-recovery middle frequency average power, and the mid-recovery high frequency average power, and (b) determining a slope of the mid-recovery line to obtain a mid-recovery spectral slope, wherein the mid-recovery frequency domain value comprises the mid-recovery spectral slope.

11. The method of claim 8 wherein:

for the peak exercise phase, the energy values comprise a peak exercise low frequency average power, a peak exercise middle frequency average power, and a peak exercise high frequency average power of the windowed time series, and wherein determining the frequency domain value comprises determining a peak exercise frequency domain value by (a) fitting a peak exercise line to the peak exercise low frequency average power, the peak exercise middle frequency average power, and peak exercise the high frequency average power, and (b) determining a slope of the peak exercise line to obtain a peak exercise spectral slope, wherein the peak exercise frequency domain value comprises the peak exercise spectral slope; and

for the mid-recovery phase, the energy values further comprise a mid-recovery low frequency average power, a mid-recovery middle frequency average power, and a mid-recovery high frequency average power of the windowed time series, and wherein the method further comprises determining a mid-recovery frequency domain value by (a) fitting a mid-recovery line to the mid-recovery low frequency average power, the mid-recovery middle frequency average power, and the mid-recovery high frequency average power, and (b) determining a slope of the mid-recovery line to obtain a mid-recovery spectral slope, wherein the mid-recovery frequency domain value comprises the mid-recovery spectral slope, and the method further comprises fitting a line across selected frequency ranges for mid-exercise phase and for the start recovery phase, and combining the two determined slopes.

12. The method of claim 11 wherein combining the two determined slopes comprises subtracting the slopes.

13. The method of claim 11 wherein combining the two determined slopes comprises averaging the slopes.

14. The method of claim 11 wherein assessing the risk of a cardiac event comprises at least one of (a) comparing a difference between the determined peak exercise frequency domain value and the determined mid-recovery exercise frequency value with a difference between a reference peak exercise frequency domain value associated with a population that has experienced the cardiac event and a reference mid-recovery frequency domain value associated with the population that has experienced the cardiac event, and (b) fitting a line to the Log (power) versus Log (frequency) over a selected frequency range, determining spectral slope estimates from different time windows of the stress test, and combining the spectral slope estimates to provide a good signal to noise ratio.

15. A method for assessing cardiac risks in a specific patent based on heart rate variability, comprising:

providing heart activity data of a specific patient including a windowed time series related to heart rate variability during a heart rate test, the windowed time series including ecotopic beats;

determining an aggregate power for a frequency band of the windowed time series; and

assessing the risk of a cardiac event based on the aggregate power;

wherein providing heart activity data comprises determining an instantaneous heart rate sequence from an ECG waveform obtained during an exercise period of increasing heart rate and a recovery period of decreasing heart rate, providing a uniform sample series of the instantaneous heart rate sequence, and reducing the uniform sample series via filtering the heart activity data and/or detrending by subtraction of a fit polynomial function of the heart rate variability data to obtain a heart rate variability waveform associated with variances in the heart rate between beats in a desired frequency band; and

determining the aggregate power comprise performing (a) a Fourier transform of the windowed time series into a spectrum for frequency domain analysis and/or (b) a root-means-square computation of the filtered or detrended time series.

16. The method of claim 15 wherein:

determining an instantaneous heart rate sequence comprises ascertaining R-R intervals of the ECG waveform;

reducing the uniform sample series comprises at least one of processing the uniform sample series through a selected band filter and detrending by subtraction of a fit polynomial function of the heart rate variability data that provides the heart rate variability waveform for selected frequencies; and

expressing the instantaneous heart rate sequence in beats/minute.

17. The method of claim 16 wherein:

determining an instantaneous heart rate sequence comprises ascertaining R-R intervals of the ECG waveform;

providing a uniform sample series of the instantaneous heart rate sequence comprises interpolating between adjacent beats at a uniform sample rate; and

reducing the uniform sample series comprises at least one of processing the uniform sample series through a selected band filter and detrending by subtraction of a fit polynomial function of the heart rate variability data that provides the heart rate variability waveform for selected frequencies.

18. A method for assessing cardiac risks in a specific patent based on heart rate variability, comprising:

providing heart activity data of a specific patient including a windowed time series related to heart rate variability during a heart rate test, the windowed time series including ecotopic beats;

determining an aggregate power for a frequency band of the windowed time series; and

assessing the risk of a cardiac event based on the aggregate power;

wherein providing the heart activity data comprises;

identifying a peak exercise phase and a mid-recovery phase of the heart rate test;

determining instantaneous heart rate sequences for the peak exercise phase and the mid-recovery phase by ascertaining R-R intervals of an ECG waveform of the peak exercise phase and the mid-recovery phase;

providing uniform sample series of the instantaneous heart rate sequences by interpolating between adjacent beats at a uniform sample rate;

reducing the uniform sample series comprises filtering the uniform sample series through a selected band filter in a forward direction through at least a portion of the exercise period and a reverse direction through at least a portion of the recovery period to provide the heart rate variability waveform for a selected frequency band during the peak exercise phase and the mid-recovery phase, and/or determining an average of slopes of the mid-exercise and start-recovery phases.

19. The method of claim 18 wherein the aggregate power comprises a sum of the power of the spectral components within the selected frequency band.

20. A method for assessing cardiac risks in a specific patent based on heart rate variability, comprising:

providing heart activity data of a specific patient including a windowed time series related to heart rate variability during a heart rate test, the windowed time series including ecotopic beats;

determining an aggregate power for a frequency band of the windowed time series; and

assessing the risk of a cardiac event based on the aggregate power;

wherein determining an aggregate power comprises performing a root-means-squared computation of corresponding filtered time series.

21. The method of claim 20 wherein assessing the risk of a cardiac event comprises comparing the determined aggregate power with a reference aggregate power range associated with a population that has experienced the cardiac event.

22. The method of claim 20 wherein assessing the risk of a cardiac event comprises comparing corresponding aggregate powers for the peak exercise low frequency, peak exercise middle frequency, peak exercise high frequency, mid-recovery low frequency, mid-recovery middle frequency, and mid-recovery high frequency with corresponding reference aggregate powers associated with a population that has experienced the cardiac event.

23. A method for assessing cardiac risks in a specific patient based on heart rate variability, comprising:

providing heart activity data of a specific patient including a windowed time series related to heart rate variability during a heart rate test;

determining a frequency domain value based on energy values of frequency bands of the heart rate variability in the windowed time series;

determining an aggregate power for a frequency band of the windowed time series; and

assessing the risk of a cardiac event based on the frequency value and the aggregate power;

wherein providing heart activity data comprises determining an instantaneous heart rate sequence from an ECG waveform obtained during an exercise period of increasing heart rate and a recovery period of decreasing heart rate, providing a uniform sample series of the instantaneous heart rate sequence, and reducing the uniform sample series to obtain a heart rate variability waveform associated with variances in the heart rate between beats in a desired frequency band.

24. The method of claim 23 wherein:

determining an instantaneous heart rate sequence comprises ascertaining R-R intervals of the ECG waveform and expressing the instantaneous heart rate sequence in beats/minute;

reducing the uniform sample series comprises processing the uniform sample series through a selected band filter that provides the heart rate variability waveform for selected frequencies;

performing a Fourier transform of the windowed time series into a spectrum for frequency domain analysis; and

computing an aggregate power of the windowed time series using a root-means-square computation.

25. The method of claim 23 wherein:

determining an instantaneous heart rate sequence comprises ascertaining R-R intervals of the ECG waveform;

providing a uniform sample series of the instantaneous heart rate sequence comprises interpolating between adjacent beats at a uniform sample rate; and

reducing the uniform sample series comprises processing the uniform sample series through a selected band filter that provides the heart rate variability waveform for selected frequencies.

26. The method of claim 23 wherein reducing the uniform sample series comprise forward filtering through at least a portion of the exercise period of the heart rate test to approximately an end of a peak exercise phase and reverse filtering through at least a portion of the recovery period of the heart rate test to approximately the end of the peak exercise phase.

27. A method for assessing cardiac risks in a specific patient based on heart rate variability, comprising:

providing heart activity data of a specific patient including a windowed time series related to heart rate variability during a heart rate test;

determining a frequency domain value based on energy values of frequency bands of the heart rate variability in the windowed time series;

determining an aggregate power for a frequency band of the windowed time series; and

assessing the risk of a cardiac event based on the frequency value and the aggregate power;

wherein the energy values comprise a low frequency average power, a middle frequency average power, and a high frequency average power of the windowed time series, and wherein determining the frequency domain value comprises (a) fitting a line to the low frequency average power, the middle frequency average power, and the high frequency average power, and (b) determining a slope of the line to obtain a spectral slope, wherein the frequency domain value comprises the spectral slope.

28. The method of claim 27 wherein assessing the risk of a cardiac event comprises comparing the determined frequency domain value with a reference frequency domain value range associated with a population that has experienced the cardiac event.

29. A method for assessing cardiac risks in a specific patient based on heart rate variability, comprising:

providing heart activity data of a specific patient including a windowed time series related to heart rate variability during a heart rate test;

determining a frequency domain value based on energy values of frequency bands of the heart rate variability in the windowed time series;

determining an aggregate power for a frequency band of the windowed time series; and

assessing the risk of a cardiac event based on the frequency value and the aggregate power;

wherein providing the heart activity data comprises;

identifying a peak exercise phase and a mid-recovery phase of the heart rate test;

determining instantaneous heart rate sequences for the peak exercise phase and the mid-recovery phase by ascertaining R-R intervals of an ECG waveform of the peak exercise phase and the mid-recovery phase;

providing uniform sample series of the instantaneous heart rate sequences by interpolating between adjacent beats at a uniform sample rate; and

reducing the uniform sample series by filtering the uniform sample series through a selected band filter in a forward direction through at least the exercise period and in a reverse direction through at least a portion of the recovery period to provide the heart rate variability waveform for selected frequencies during the peak exercise phase and the mid-recovery phase.

30. The method of claim 29 wherein the aggregate power comprises a sum of the power of the spectral components within the selected frequency band.

31. The method of claim 30 wherein assessing the risk of a cardiac event comprises comparing (a) a difference between the determined peak exercise frequency domain value and the determined mid-recovery exercise frequency value with (b) a difference between a reference peak exercise frequency domain value range associated with a population that has experienced the cardiac event and a reference mid-recovery frequency domain value range associated with the population that has experienced the cardiac event.

32. A method for assessing cardiac risks in a specific patient based on heart rate variability, comprising:

providing heart activity data of a specific patient including a windowed time series related to heart rate variability during a heart rate test;

determining a frequency domain value based on energy values of frequency bands of the heart rate variability in the windowed time series;

determining an aggregate power for a frequency band of the windowed time series; and

assessing the risk of a cardiac event based on the frequency value and the aggregate power;

wherein determining an aggregate power comprises performing a root-means-squared calculation of a corresponding windowed time series.

33. The method of claim 32 wherein assessing the risk of a cardiac event comprises comparing the determined aggregate power with a reference aggregate power range associated with a population that has experienced the cardiac event.

34. The method of claim 32 wherein assessing the risk of a cardiac event comprises (a) comparing the determined frequency domain value with a reference frequency domain value range associated with a population that has experienced the cardiac event and (b) comparing the determined aggregate power with a reference aggregate power range associated with the population that has experienced the cardiac event.

35. An apparatus for assessing cardiac risks in a specific patient based on heart rate variability, comprising:

at least one sensor configured to obtain cardiac signals from a specific patient; and

a computer configured to receive the cardiac signals from the sensor, and the computer including a computer operable medium having instructions that cause the computer to (a) provide heart activity data of the specific patient including a windowed time series related to heart rate variability during a heart rate test, (b) determine a frequency domain value based on at least one of energy values of frequency bands and spectral slope(s) across selected frequency ranges of the heart activity data in the windowed time series, and (c) assess the risk of a cardiac event based on the frequency domain value;

wherein the heart activity data comprises an instantaneous heart rate sequence from an ECG waveform obtained during an exercise period of increasing heart rate and a recovery period of decreasing heart rate, and the instructions cause the computer to provide a uniform sample series of the instantaneous heart rate sequence, and reduce the uniform sample series by filtering and/or detrending via fitting a polynomial function to the heart activity data across a selected time window to obtain a heart rate variability waveform associated with variances in the heart rate between beats in a desired frequency band.