Medical decision support system
At least one swing value is determined responsive to a difference between maximum and minimum amplitude values of an auscultatory sound signal within a temporal region of a heart-cycle segment spanning an entire heart cycle of an auscultatory sound signal, wherein a location of at least one temporal region is responsive to a duration of the heart-cycle segment. S 4 sound presence is detected responsive to a ratio of S 4 SWING to S 2 SWING in relation an associated median value thereof from a population of test-subjects. A Support Vector Machine trained responsive to age, sex, S 4 presence and a plurality of heart sound swing measures provides for detecting CAD. Unsupervised classification of an S 3 swing and median and mean values of a Short Time Fourier Transform within associated frequency intervals, based upon data from a plurality of heart cycles of a plurality of test-subject provides for detecting presence of an S 3 sound.
1 . A computer-implemented method of detecting an S 4 heart sound in an auscultatory sound signal, comprising:
a. receiving an electrographic signal from an electrocardiogram (ECG) sensor;
b. receiving at least one auscultatory sound signal responsive to a corresponding at least one auscultatory sound-or-vibration sensor operatively associated with a test-subject;
c. segmenting said at least one auscultatory sound signal responsive to said electrographic signal so as to associate a plurality of heart-cycle time series with said at least one auscultatory sound signal, wherein each heart-cycle time series of said plurality of heart-cycle time series spans a single corresponding heart cycle;
d. for said each heart-cycle time series of said plurality of heart-cycle time series:
i. locating an S 2 -sound time interval within said each heart-cycle time series, wherein an end of said S 2 -sound time interval is assumed to occur at a start of a diastasis time interval, and a start of said S 2 -sound time interval is assumed to occur at a midpoint between a start of said each heart-cycle time series and said end of said S 2 -sound time interval;
ii. determining an S 2 swing value of said each heart-cycle time series, wherein said S 2 swing value is given by an absolute value of a difference between maximum and minimum values of said each heart-cycle time series within said S 2 -sound time interval;
iii. locating an S 4 -sound time interval within said each heart-cycle time series, wherein an end of said S 4 -sound time interval is assumed to occur at an end of said each heart-cycle time series, and a start of said S 4 -sound time interval is assumed to occur at a point in time that is 80 percent of a duration of said each heart-cycle time series;
iv. determining an S 4 swing value of said each heart-cycle time series, wherein said S 4 swing value is given by an absolute value of a difference between maximum and minimum values of said each heart-cycle time series within said S 4 -sound time interval; and
v. determining an S 4 S 2 swing ratio of said each heart-cycle time series, wherein said S 4 S 2 swing ratio is given by a ratio of said S 4 swing value and said S 2 swing value; and
e. for said each heart-cycle time series of said plurality of heart-cycle time series, determining a value of a fraction of said each heart-cycle time series of said plurality of heart-cycle time series for which a corresponding said S 4 S 2 swing ratio exceeds a predetermined mean value of said S 4 S 2 swing ratio, wherein said fraction of said each heart-cycle time series is in relation to a total number of said each heart-cycle time series of said plurality of heart-cycle time series, if said value of said fraction of said each heart-cycle time series is in excess of a first threshold, then indicating that said test-subject exhibits the S 4 heart sound, otherwise indicating that said test-subject does not exhibit said S 4 heart sound.
2 . A computer-implemented method of detecting an S 4 heart sound in an auscultatory sound signal as recited in claim 1 , wherein said start of said diastasis time interval is assumed to occur at a temporal offset from the beginning of said each heart-cycle time series, wherein said temporal offset is given in milliseconds by 350 plus 0 . 3 times the quantity (TRR- 350 ), wherein said TRR is said duration of said each heart-cycle time series in milliseconds.
3 . A computer-implemented method of detecting an S 4 heart sound in an auscultatory sound signal as recited in claim 1 , wherein said ratio of said S 4 S 2 swing ratio is given by said S 4 swing value divided by said S 2 swing value.
4 . A computer-implemented method of detecting an S 4 heart sound in an auscultatory sound signal as recited in claim 1 , further comprising:
a. repeating steps a-d of claim 1 for each training-test-subject of a plurality of training-test-subjects so as to generate a composite set of S 4 S 2 swing ratios, wherein said composite set of S 4 S 2 swing ratios comprises a composite of said S 4 S 2 swing ratios from each said each heart-cycle time series of said plurality of heart-cycle time series from said each training-test-subject of said plurality of training-test-subjects, a first non-null subset of said plurality of training-test-subjects is known a priori to exhibit said S 4 heart sound, and a remaining subset of said plurality of training-test-subjects is known a priori to not exhibit said S 4 heart sound; and
b. determining said predetermined mean value of said S 4 S 2 swing ratio responsive to a mean value of said composite set of S 4 S 2 swing ratios.
5 . A computer-implemented method of detecting an S 4 heart sound in an auscultatory sound signal as recited in claim 1 , further comprising:
a. setting a value of a first flag responsive to whether or not said test-subject exhibits said S 4 heart sound;
b. for each of a selected frequency range selected from a group of frequency ranges consisting of a first frequency range and a second frequency range:
i. bandpass filtering said at least one auscultatory sound signal with a bandpass filter responsive to said selected frequency range so as to generate a corresponding at least one bandpass-filtered auscultatory sound signal;
ii. segmenting said corresponding at least one bandpass-filtered auscultatory sound signal responsive to said electrographic signal so as to associate a corresponding plurality of heart-cycle time series with said corresponding at least one bandpass-filtered auscultatory sound signal, wherein each corresponding heart-cycle time series of said corresponding plurality of heart-cycle time series spans a single corresponding heart cycle; and
iii. for said each corresponding heart-cycle time series of said corresponding plurality of heart-cycle time series:
a) determining a plurality of corresponding swing values, wherein each corresponding swing value of said plurality of corresponding swing values is selected from the group of corresponding swing values consisting of a corresponding S 1 swing value, a corresponding S 2 swing value, a corresponding S 3 swing value, and a corresponding S 4 swing value that are respectively determined within a corresponding S 1 -sound time interval, a corresponding S 2 -sound time interval, a corresponding S 3 -sound time interval, and a corresponding S 4 -sound time interval, respectively, wherein an end of said corresponding S 4 -sound time interval is assumed to occur at an end of said each corresponding heart-cycle time series, a start of said corresponding S 4 -sound time interval is assumed to occur at a point in time that is 80 percent of a duration of said each corresponding heart-cycle time series, said corresponding S 4 swing value is given by an absolute value of a difference between maximum and minimum values of said each corresponding heart-cycle time series within said corresponding S 4 -sound time interval, an end of said corresponding S 2 -sound time interval is assumed to occur at said start of said diastasis time interval, a start of said corresponding S 2 -sound time interval is assumed to occur at a midpoint between the beginning of said each corresponding heart-cycle time series and said end of said corresponding S 2 -sound time interval, said corresponding S 2 swing value is given by an absolute value of a difference between maximum and minimum values of said each corresponding heart-cycle time series within said corresponding S 2 -sound time interval, a start of said corresponding S 3 -sound time interval is coincident said end of said corresponding S 2 -sound time interval, and an end of said corresponding S 3 -sound time interval is coincident with said start of said corresponding S 4 -sound time interval, said corresponding S 3 swing value is given by an absolute value of a difference between maximum and minimum values of said each corresponding heart-cycle time series within said corresponding S 3 -sound time interval, a start of said corresponding S 1 -sound time interval is coincident with said beginning of said each corresponding heart-cycle time series, and an end of said corresponding S 1 -sound time interval is coincident with said start of said corresponding S 2 -sound time interval, and said corresponding S 1 swing value is given by an absolute value of a difference between maximum and minimum values of said each corresponding heart-cycle time series within said corresponding S 1 -sound time interval; and
b) determining at least one corresponding swing ratio of said each corresponding heart-cycle time series, wherein each corresponding swing ratio of said at least one corresponding swing ratio is selected from the group of corresponding swing ratios consisting of a corresponding S 4 S 2 swing ratio of said corresponding S 4 swing value and said corresponding S 2 swing value; a corresponding S 4 S 3 swing ratio of said corresponding S 4 swing value and said corresponding S 3 swing value; a corresponding S 3 S 1 swing ratio of said corresponding S 3 swing value and said corresponding S 1 swing value; a corresponding S 3 S 2 swing ratio of said corresponding S 3 swing value and said corresponding S 2 swing value; a corresponding S 2 S 1 swing ratio of said corresponding S 2 swing value and said corresponding S 1 swing value; and a corresponding S 4 S 1 swing ratio of said corresponding S 4 swing value and said corresponding S 1 swing value; and
c. using a support vector machine (SVM) to generate an estimate of whether or not said test-subject exhibits a cardiac condition responsive to a feature set of data, wherein said feature set of data comprises:
i. said first flag; and
ii. for each of said first and second frequency ranges, a median value of each of said at least one corresponding swing ratio across each said each corresponding heart-cycle time series of said corresponding plurality of heart-cycle time series.
6 . A computer-implemented method of detecting an S 4 heart sound in an auscultatory sound signal as recited in claim 5 , wherein said start of said diastasis time interval is assumed to occur at a temporal offset from the beginning of said each corresponding heart-cycle time series, wherein said temporal offset is given in milliseconds by 350 plus 0.3 times the quantity (TRR-350), wherein said TRR is said duration of said each corresponding heart-cycle time series in milliseconds.
7 . A computer-implemented method of detecting an S 4 heart sound in an auscultatory sound signal as recited in claim 5 , wherein said feature set of data is further responsive to an age of said test-subject.
8 . A computer-implemented method of detecting an S 4 heart sound in an auscultatory sound signal as recited in claim 5 , wherein said feature set of data is further responsive to a gender of said test-subject.
9 . A computer-implemented method of detecting an S 4 heart sound in an auscultatory sound signal as recited in claim 5 , wherein said first frequency range is between 1 and 20 Hertz.
10 . A computer-implemented method of detecting an S 4 heart sound in an auscultatory sound signal as recited in claim 5 , wherein a low end of said second frequency range is between 20 and 25 Hertz, and a high end of said second frequency range is between 30 and 40 Hertz.
11 . A computer-implemented method of detecting an S 4 heart sound in an auscultatory sound signal as recited in claim 5 , wherein the operation of bandpass filtering is implemented with a third-order Butterworth bandpass filter.
12 . A computer-implemented method of detecting an S 4 heart sound in an auscultatory sound signal as recited in claim 5 , further comprising for each said selected frequency range and for said each corresponding heart-cycle time series of said corresponding plurality of heart-cycle time series, determining at least one supplemental swing measure, wherein each supplemental swing measure of said at least one supplemental swing measure is selected from the group of supplemental swing measures consisting of a ratio of said corresponding S 3 swing value and a product of said S 1 and S 2 corresponding swing values; a ratio of said corresponding S 4 swing value and said product of said S 1 and S 2 corresponding swing values; a ratio of said corresponding S 3 swing value and a sum of said S 1 and S 2 corresponding swing values; a ratio of said corresponding S 4 swing value and said sum of said S 1 and S 2 corresponding swing values; a sum of said corresponding S 3 S 2 swing ratio and said corresponding S 3 S 1 swing ratio; and a sum of said corresponding S 4 S 2 swing ratio and said corresponding S 4 S 1 swing ratio, wherein for each of said first and second frequency ranges, said feature set of data further comprise a median value of each of said at least one supplemental swing measures across each said each corresponding heart-cycle time series of said corresponding plurality of heart-cycle time series.
13 . A computer-implemented method of detecting an S 4 heart sound in an auscultatory sound signal as recited in claim 5 , further comprising:
a. repeating steps a and b of claim 5 for each training-test-subject of a plurality of training-test-subjects so as to generate a composite feature set of data, wherein said composite feature set of data comprises a composite of said feature set of data from said each training-test-subject of said plurality of training-test-subjects, a first non-null subset of said plurality of training-test-subjects is known a priori to exhibit said cardiac condition, and a remaining subset of said plurality of training-test-subjects is known a priori to not exhibit said cardiac condition;
b. for each said each training-test-subject of said plurality of training-test-subjects, including in said composite feature set of data an indication of whether or not said each training-test-subject exhibits said cardiac condition; and
c. training said support vector machine (SVM) with said composite feature set of data from said plurality of training-test-subjects so as to provide for estimating whether or not said test-subject exhibits said cardiac condition responsive to a corresponding feature set of data for the same features as had been incorporated for each said each training-test-subject in said composite feature set of data.
14 . A computer-implemented method of detecting an S 4 heart sound in an auscultatory sound signal as recited in claim 13 , wherein said composite feature set of data is further responsive to an age of said each training-test-subject.
15 . A computer-implemented method of detecting an S 4 heart sound in an auscultatory sound signal as recited in claim 13 , wherein said composite feature set of data is further responsive to a gender of said each training-test-subject.
16 . A computer-implemented method of detecting an S 4 heart sound in an auscultatory sound signal as recited in claim 13 , wherein for each of said first and second frequency ranges, said feature set of data further comprises a median value of each of at least one supplemental swing measure across each said each corresponding heart-cycle time series of said corresponding plurality of heart-cycle time series for each said training-test-subject of said plurality of training-test-subjects, wherein each supplemental swing measure of said at least one supplemental swing measure is selected from the group of supplemental swing measures consisting of a ratio of said corresponding S 3 swing value and a product of said S 1 and S 2 corresponding swing values; a ratio of said corresponding S 4 swing value and said product of said S 1 and S 2 corresponding swing values; a ratio of said corresponding S 3 swing value and a sum of said S 1 and S 2 corresponding swing values; a ratio of said corresponding S 4 swing value and said sum of said S 1 and S 2 corresponding swing values;
a sum of said corresponding S 3 S 2 swing ratio and said corresponding S 3 S 1 swing ratio; and
a sum of said corresponding S 4 S 2 swing ratio and said corresponding S 4 S 1 swing ratio.
17 . A computer-implemented method of detecting an S 4 heart sound in an auscultatory sound signal as recited in claim 13 , wherein the features incorporated in said feature set of data used in the operation of using said support vector machine (SVM) to generate said estimate of whether or not said test-subject exhibits said cardiac condition are the same features as used in said composite feature set of data to train said support vector machine (SVM).
18 . A computer-implemented method of detecting an S 4 heart sound in an auscultatory sound signal as recited in claim 1 , further comprising:
a. for each of a selected frequency range selected from a group of frequency ranges consisting of a first frequency range and a second frequency range:
i. bandpass filtering said at least one auscultatory sound signal with a bandpass filter responsive to said selected frequency range so as to generate a corresponding at least one bandpass-filtered auscultatory sound signal;
ii. segmenting said corresponding at least one bandpass-filtered auscultatory sound signal responsive to said electrographic signal so as to associate a corresponding plurality of heart-cycle time series with said corresponding at least one bandpass-filtered auscultatory sound signal, wherein each corresponding heart-cycle time series of said corresponding plurality of heart-cycle time series spans a single corresponding heart cycle; and
iii. for said each corresponding heart-cycle time series of said corresponding plurality of heart-cycle time series:
a) determining a plurality of corresponding swing values, wherein each corresponding swing value of said plurality of corresponding swing values is selected from the group of corresponding swing values consisting of a corresponding S 1 swing value, a corresponding S 2 swing value, a corresponding S 3 swing value, and a corresponding S 4 swing value that are respectively determined within a corresponding S 1 -sound time interval, a corresponding S 2 -sound time interval, a corresponding S 3 -sound time interval, and a corresponding S 4 -sound time interval, respectively, wherein an end of said corresponding S 4 -sound time interval is assumed to occur at an end of said each corresponding heart-cycle time series, a start of said corresponding S 4 -sound time interval is assumed to occur at a point in time that is 80 percent of a duration of said each corresponding heart-cycle time series, said corresponding S 4 swing value is given by an absolute value of a difference between maximum and minimum values of said each corresponding heart-cycle time series within said corresponding S 4 -sound time interval, an end of said corresponding S 2 -sound time interval is assumed to occur at said start of said diastasis time interval, a start of said corresponding S 2 -sound time interval is assumed to occur at a midpoint between the beginning of said each corresponding heart-cycle time series and said end of said corresponding S 2 -sound time interval, said corresponding S 2 swing value is given by an absolute value of a difference between maximum and minimum values of said each corresponding heart-cycle time series within said corresponding S 2 -sound time interval, a start of said corresponding S 3 -sound time interval is coincident said end of said corresponding S 2 -sound time interval, and an end of said corresponding S 3 -sound time interval is coincident with said start of said corresponding S 4 -sound time interval, said corresponding S 3 swing value is given by an absolute value of a difference between maximum and minimum values of said each corresponding heart-cycle time series within said corresponding S 3 -sound time interval, a start of said corresponding S 1 -sound time interval is coincident with said beginning of said each corresponding heart-cycle time series, and an end of said corresponding S 1 -sound time interval is coincident with said start of said corresponding S 2 -sound time interval, and said corresponding S 1 swing value is given by an absolute value of a difference between maximum and minimum values of said each corresponding heart-cycle time series within said corresponding S 1 -sound time interval; and
b) determining at least one corresponding swing ratio of said each corresponding heart-cycle time series, wherein each corresponding swing ratio of said at least one corresponding swing ratio is selected from the group of corresponding swing ratios consisting of a corresponding S 4 S 2 swing ratio of said corresponding S 4 swing value and said corresponding S 2 swing value; a corresponding S 4 S 3 swing ratio of said corresponding S 4 swing value and said corresponding S 3 swing value; a corresponding S 3 S 1 swing ratio of said corresponding S 3 swing value and said corresponding S 1 swing value; a corresponding S 3 S 2 swing ratio of said corresponding S 3 swing value and said corresponding S 2 swing value; a corresponding S 2 S 1 swing ratio of said corresponding S 2 swing value and said corresponding S 1 swing value; and a corresponding S 4 S 1 swing ratio of said corresponding S 4 swing value and said corresponding S 1 swing value; and
b. using a support vector machine (SVM) to generate an estimate of whether or not said test-subject exhibits a cardiac condition responsive to a feature set of data, wherein said feature set of data comprises:
i. an indication of whether or not said test-subject exhibits said S 4 heart sound; and
ii. for each of said first and second frequency ranges, a median value of each of said at least one corresponding swing ratio across each said each corresponding heart-cycle time series of said corresponding plurality of heart-cycle time series.
19 . A computer-implemented method of detecting an S 4 heart sound in an auscultatory sound signal as recited in claim 18 , wherein said start of said diastasis time interval is assumed to occur at a temporal offset from the beginning of said each corresponding heart-cycle time series, wherein said temporal offset is given in milliseconds by 350 plus 0.3 times the quantity (TRR-350), wherein said TRR is said duration of said each corresponding heart-cycle time series in milliseconds.
20 . A computer-implemented method of detecting an S 4 heart sound in an auscultatory sound signal as recited in claim 18 , wherein said feature set of data is further responsive to an age of said test-subject.
21 . A computer-implemented method of detecting an S 4 heart sound in an auscultatory sound signal as recited in claim 18 , wherein said feature set of data is further responsive to a gender of said test-subject.
22 . A computer-implemented method of detecting an S 4 heart sound in an auscultatory sound signal as recited in claim 18 , wherein said first frequency range is between 1 and 20 Hertz.
23 . A computer-implemented method of detecting an S 4 heart sound in an auscultatory sound signal as recited in claim 18 , wherein a low end of said second frequency range is between 20 and 25 Hertz, and a high end of said second frequency range is between 30 and 40 Hertz.
24 . A computer-implemented method of detecting an S 4 heart sound in an auscultatory sound signal as recited in claim 18 , wherein the operation of bandpass filtering is implemented with a third-order Butterworth bandpass filter.
25 . A computer-implemented method of detecting an S 4 heart sound in an auscultatory sound signal as recited in claim 18 , further comprising for each said selected frequency range and for said each corresponding heart-cycle time series of said corresponding plurality of heart-cycle time series, determining at least one supplemental swing measure, wherein each supplemental swing measure of said at least one supplemental swing measure is selected from the group of supplemental swing measures consisting of a ratio of said corresponding S 3 swing value and a product of said S 1 and S 2 corresponding swing values; a ratio of said corresponding S 4 swing value and said product of said S 1 and S 2 corresponding swing values; a ratio of said corresponding S 3 swing value and a sum of said S 1 and S 2 corresponding swing values; a ratio of said corresponding S 4 swing value and said sum of said S 1 and S 2 corresponding swing values; a sum of said corresponding S 3 S 2 swing ratio and said corresponding S 3 S 1 swing ratio; and a sum of said corresponding S 4 S 2 swing ratio and said corresponding S 4 S 1 swing ratio, wherein for each of said first and second frequency ranges, said feature set of data further comprise a median value of each of said at least one supplemental swing measures across each said each corresponding heart-cycle time series of said corresponding plurality of heart-cycle time series.
26 . A computer-implemented method of detecting an S 4 heart sound in an auscultatory sound signal as recited in claim 18 , further comprising:
a. repeating step a of claim 18 for each training-test-subject of a plurality of training-test-subjects so as to generate a composite feature set of data, wherein said composite feature set of data comprises a composite of said feature set of data from said each training-test-subject of said plurality of training-test-subjects, a first non-null subset of said plurality of training-test-subjects is known a priori to exhibit said cardiac condition, and a remaining subset of said plurality of training-test-subjects is known a priori to not exhibit said cardiac condition;
b. for each said each training-test-subject of said plurality of training-test-subjects, including in said composite feature set of data an indication of whether or not said each training-test-subject exhibits said cardiac condition; and
c. training said support vector machine (SVM) with said composite feature set of data from said plurality of training-test-subjects so as to provide for estimating whether or not said test-subject exhibits said cardiac condition responsive to a corresponding feature set of data for the same features as had been incorporated for each said each training-test-subject in said composite feature set of data.
27 . A computer-implemented method of detecting an S 4 heart sound in an auscultatory sound signal as recited in claim 26 , wherein said composite feature set of data is further responsive to an age of said each training-test-subject.
28 . A computer-implemented method of detecting an S 4 heart sound in an auscultatory sound signal as recited in claim 26 , wherein said composite feature set of data is further responsive to a gender of said each training-test-subject.
29 . A computer-implemented method of detecting an S 4 heart sound in an auscultatory sound signal as recited in claim 26 , wherein for each of said first and second frequency ranges, said feature set of data further comprises a median value of each of at least one supplemental swing measure across each said each corresponding heart-cycle time series of said corresponding plurality of heart-cycle time series for each said training-test-subject of said plurality of training-test-subjects, wherein each supplemental swing measure of said at least one supplemental swing measure is selected from the group of supplemental swing measures consisting of a ratio of said corresponding S 3 swing value and a product of said S 1 and S 2 corresponding swing values; a ratio of said corresponding S 4 swing value and said product of said S 1 and S 2 corresponding swing values; a ratio of said corresponding S 3 swing value and a sum of said S 1 and S 2 corresponding swing values; a ratio of said corresponding S 4 swing value and said sum of said S 1 and S 2 corresponding swing values;
a sum of said corresponding S 3 S 2 swing ratio and said corresponding S 3 S 1 swing ratio; and
a sum of said corresponding S 4 S 2 swing ratio and said corresponding S 4 S 1 swing ratio.
30 . A computer-implemented method of detecting an S 4 heart sound in an auscultatory sound signal as recited in claim 26 , wherein the features incorporated in said feature set of data used in the operation of using said support vector machine (SVM) to generate said estimate of whether or not said test-subject exhibits said cardiac condition are the same features as used in said composite feature set of data to train said support vector machine (SVM).
31 . A computer-implemented method of segmenting an auscultatory sound signal, comprising:
a. receiving an electrographic signal from an electrocardiogram (ECG) sensor;
b. generating an electrographic envelope signal representing an envelope responsive to an even power of said electrographic signal;
c. locating a plurality of peaks of said electrographic envelope signal corresponding to a corresponding plurality of R-peaks of said electrographic signal;
d. receiving at least one auscultatory sound signal from a corresponding at least one auscultatory sound-or-vibration sensor;
e. segmenting said at least one auscultatory sound signal into at least one heart-cycle segment responsive to said plurality of peaks of said electrographic envelope signal, wherein each said at least one heart-cycle segment spans an entire heart cycle associated with a corresponding single heartbeat;
f. for at least one said at least one heart-cycle segment:
i. identifying at least one temporal region within each said at least one said at least one heart-cycle segment, wherein a corresponding at least one location of said at least one temporal region is responsive to a duration of said at least one said at least one heart-cycle segment, and at least one said at least one temporal region is located relative to a start of a diastasis time interval within diastole of said at least one said at least one heart-cycle segment,, wherein said start of said diastasis time interval is assumed to occur at a temporal offset from the beginning of said at least one heart-cycle segment, wherein said temporal offset is given in milliseconds by 350 plus 0.3 times the quantity (TRR-350), wherein said TRR is the duration of said at least one said at least one heart-cycle segment in milliseconds; and
ii. determining at least one swing value responsive to a difference between maximum and minimum amplitude values of said at least one auscultatory sound signal within a corresponding said at least one temporal region; and
g. utilizing a first swing value of said at least one swing value as a first feature input set to a classifier to provide for determining whether or not a test-subject associated with said at least one auscultatory sound signal likely exhibits a cardiac condition.
32 . A computer-implemented method of segmenting an auscultatory sound signal as recited in claim 31 , wherein at least one said at least one temporal region either spans said start of said diastasis time interval or terminates at said start of said diastasis time interval.
33 . A computer-implemented method of segmenting an auscultatory sound signal as recited in claim 31 , wherein a duration of at least one said at least one temporal region is between 80 and 120 milliseconds.
34 . A computer-implemented method of segmenting an auscultatory sound signal as recited in claim 31 , wherein at least one said at least one temporal region is associated with at least one heart sound selected from the group of heart sounds consisting of an S 3 heart sound, an S 4 heart sound and an S 5 heart sound.
35 . A computer-implemented method of segmenting an auscultatory sound signal, comprising:
a. receiving an electrographic signal from an electrocardiogram (ECG) sensor;
b. generating an electrographic envelope signal representing an envelope responsive to an even power of said electrographic signal;
c. locating a plurality of peaks of said electrographic envelope signal corresponding to a corresponding plurality of R-peaks of said electrographic signal;
d. receiving at least one auscultatory sound signal from a corresponding at least one auscultatory sound-or-vibration sensor;
e. segmenting said at least one auscultatory sound signal into at least one heart-cycle segment responsive to said plurality of peaks of said electrographic envelope signal, wherein each said at least one heart-cycle segment spans an entire heart cycle associated with a corresponding single heartbeat;
f. for at least one said at least one heart-cycle segment:
i. identifying at least one temporal region within each said at least one said at least one heart-cycle segment, wherein a corresponding at least one location of said at least one temporal region is responsive to a duration of said at least one said at least one heart-cycle segment, at least one said at least one temporal region is associated with at least one heart sound selected from the group consisting of an S 3 heart sound, an S 4 heart sound and an S 5 heart sound, and said at least one said at least one temporal region is located relative to a start of a diastasis time interval within diastole of said at least one said at least one heart-cycle segment; and
ii. determining at least one swing value responsive to a difference between maximum and minimum amplitude values of said at least one auscultatory sound signal within a corresponding said at least one temporal region;
g. utilizing a first swing value of said at least one swing value as a first feature input set to a classifier to provide for determining whether or not a test-subject associated with said at least one auscultatory sound signal likely exhibits a cardiac condition;
h. generating a Short-Time Fourier Transform of said at least one auscultatory sound signal within said at least one temporal region; and
i. utilizing at least one frequency-domain feature from or responsive to said Short-Time Fourier Transform of said at least one auscultatory sound signal within said at least one temporal region as a second feature input set to said classifier to provide for determining whether or not said test-subject associated with said at least one auscultatory sound signal likely exhibits a cardiac condition.
36 . A computer-implemented method of segmenting an auscultatory sound signal as recited in claim 35 , wherein at least one said at least one temporal region is located at the end of said at least one said at least one heart-cycle segment.
37 . A computer-implemented method of segmenting an auscultatory sound signal as recited in claim 36 , wherein a duration of said at least one said at least one temporal region is between 10 and 25 percent of the duration of said at least one said at least one heart-cycle segment.
38 . A computer-implemented method of segmenting an auscultatory sound signal as recited in claim 36 , wherein a duration of said at least one said at least one temporal region is between 10 and 20 percent of the duration of said at least one said at least one heart-cycle segment.
39 . A computer-implemented method of segmenting an auscultatory sound signal, comprising:
a. receiving an electrographic signal from an electrocardiogram (ECG) sensor;
b. generating an electrographic envelope signal representing an envelope responsive to an even power of said electrographic signal;
c. locating a plurality of peaks of said electrographic envelope signal corresponding to a corresponding plurality of R-peaks of said electrographic signal;
d. receiving at least one auscultatory sound signal from a corresponding at least one auscultatory sound-or-vibration sensor;
e. segmenting said at least one auscultatory sound signal into at least one heart-cycle segment responsive to said plurality of peaks of said electrographic envelope signal, wherein each said at least one heart-cycle segment spans an entire heart cycle associated with a corresponding single heartbeat;
f. for at least one said at least one heart-cycle segment:
i. identifying at least one temporal region within each said at least one said at least one heart-cycle segment, wherein a corresponding at least one location of said at least one temporal region is responsive to a duration of said at least one said at least one heart-cycle segment, said at least one temporal region is associated with at least one heart sound selected from the group consisting of an S 3 heart sound, an S 4 heart sound and an S 5 heart sound, and at least one said at least one temporal region is located at the end of at least one said at least one heart-cycle segment; and
ii. determining at least one swing value responsive to a difference between maximum and minimum amplitude values of said at least one auscultatory sound signal within a corresponding said at least one temporal region;
g. utilizing a first swing value of said at least one swing value as a first feature input set to a classifier to provide for determining whether or not a test-subject associated with said at least one auscultatory sound signal likely exhibits a cardiac condition; and
h. prior to the operation of determining said at least one swing value, generating a filtered auscultatory sound signal by filtering said at least one auscultatory sound signal within said at least one said at least one heart-cycle segment using a filter having a cutoff frequency that provides for passing either an audible range of frequencies or passing an inaudible range of frequencies below said audible range of frequencies, wherein said at least one swing value is determined responsive to said filtered auscultatory sound signal.
40 . A computer-implemented method of segmenting an auscultatory sound signal as recited in claim 39 , wherein a duration of said at least one said at least one temporal region is between 10 and 25 percent of the duration of said at least one said at least one heart-cycle segment.
41 . A computer-implemented method of segmenting an auscultatory sound signal as recited in claim 39 , wherein a duration of said at least one said at least one temporal region is between 10 and 20 percent of the duration of said at least one said at least one heart-cycle segment.
42 . A computer-implemented method of segmenting an auscultatory sound signal, comprising:
a. receiving an electrographic signal from an electrocardiogram (ECG) sensor;
b. generating an electrographic envelope signal representing an envelope responsive to an even power of said electrographic signal;
c. locating a plurality of peaks of said electrographic envelope signal corresponding to a corresponding plurality of R-peaks of said electrographic signal;
d. receiving at least one auscultatory sound signal from a corresponding at least one auscultatory sound-or-vibration sensor;
e. segmenting said at least one auscultatory sound signal into at least one heart-cycle segment responsive to said plurality of peaks of said electrographic envelope signal, wherein each said at least one heart-cycle segment spans an entire heart cycle associated with a corresponding single heartbeat;
f. for at least one said at least one heart-cycle segment:
i. identifying at least one temporal region within each said at least one said at least one heart-cycle segment, wherein a corresponding at least one location of said at least one temporal region is responsive to a duration of said at least one said at least one heart-cycle segment, and at least one said at least one temporal region is located at the end of said at least one said at least one heart-cycle segment; and
ii. determining a first swing value responsive to a difference between maximum and minimum amplitude values of said at least one auscultatory sound signal within a corresponding said at least one temporal region;
g. utilizing said first swing value as a first feature input set to a classifier to provide for determining whether or not a test-subject associated with said at least one auscultatory sound signal likely exhibits a cardiac condition;
h. determining a second swing value responsive to a difference between maximum and minimum amplitude values of said at least one auscultatory sound signal during an S 2 -sound time interval at the beginning of diastole of said at least one said at least one heart-cycle segment;
i. determining a ratio of said first swing value and said second swing value as a measure of a likelihood that said test-subject exhibits an S 4 heart sound; and
j. utilizing said ratio of said first swing value and said second swing value as a third feature input set to said classifier to provide for determining whether or not said test-subject associated with said at least one auscultatory sound signal likely exhibits a cardiac condition.
43 . A computer-implemented method of segmenting an auscultatory sound signal as recited in claim 42 , wherein a duration of said at least one said at least one temporal region is between 10 and 25 percent of the duration of said at least one said at least one heart-cycle segment.
44 . A computer-implemented method of segmenting an auscultatory sound signal as recited in claim 42 , wherein a duration of said at least one said at least one temporal region is between 10 and 20 percent of the duration of said at least one said at least one heart-cycle segment.
45 . A computer-implemented method of segmenting an auscultatory sound signal, comprising:
a. receiving an electrographic signal from an electrocardiogram (ECG) sensor;
b. generating an electrographic envelope signal representing an envelope responsive to an even power of said electrographic signal;
c. locating a plurality of peaks of said electrographic envelope signal corresponding to a corresponding plurality of R-peaks of said electrographic signal;
d. receiving at least one auscultatory sound signal from a corresponding at least one auscultatory sound-or-vibration sensor;
e. segmenting said at least one auscultatory sound signal into at least one heart-cycle segment responsive to said plurality of peaks of said electrographic envelope signal, wherein each said at least one heart-cycle segment spans an entire heart cycle associated with a corresponding single heartbeat; and
f. for at least one said at least one heart-cycle segment:
i. identifying at least one temporal region within each said at least one said at least one heart-cycle segment, wherein a corresponding at least one location of said at least one temporal region is responsive to a duration of said at least one said at least one heart-cycle segment, and at least one said at least one temporal region is located relative to a start of a diastasis time interval within diastole of said at least one said at least one heart-cycle segment, wherein said start of said diastasis time interval is assumed to occur at a temporal offset from the beginning of said at least one heart-cycle segment, wherein said temporal offset is given in milliseconds by 350 plus 0.3 times the quantity (TRR-350), wherein said TRR is the duration of said at least one said at least one heart-cycle segment in milliseconds; and
ii. determining at least one swing value responsive to a difference between maximum and minimum amplitude values of said at least one auscultatory sound signal within a corresponding said at least one temporal region.
46 . A computer-implemented method of segmenting an auscultatory sound signal as recited in claim 45 , wherein at least one said at least one temporal region either spans said start of said diastasis time interval or terminates at said start of said diastasis time interval.
47 . A computer-implemented method of segmenting an auscultatory sound signal as recited in claim 46 , wherein a duration of at least one said at least one temporal region is between 80 and 120 milliseconds.
48 . A computer-implemented method of segmenting an auscultatory sound signal as recited in claim 45 , wherein a duration of at least one said at least one temporal region is between 80 and 120 milliseconds.
49 . A computer-implemented method of segmenting an auscultatory sound signal as recited in claim 45 , wherein at least one said at least one temporal region is associated with at least one heart sound selected from the group of heart sounds consisting of an S 3 heart sound, an S 4 heart sound and an S 5 heart sound.
50 . A computer-implemented method of segmenting an auscultatory sound signal, comprising:
a. receiving an electrographic signal from an electrocardiogram (ECG) sensor;
b. generating an electrographic envelope signal representing an envelope responsive to an even power of said electrographic signal;
c. locating a plurality of peaks of said electrographic envelope signal corresponding to a corresponding plurality of R-peaks of said electrographic signal;
d. receiving at least one auscultatory sound signal from a corresponding at least one auscultatory sound-or-vibration sensor;
e. segmenting said at least one auscultatory sound signal into at least one heart-cycle segment responsive to said plurality of peaks of said electrographic envelope signal, wherein each said at least one heart-cycle segment spans an entire heart cycle associated with a corresponding single heartbeat;
f. for at least one said at least one heart-cycle segment:
i. identifying at least one temporal region within each said at least one said at least one heart-cycle segment, wherein a corresponding at least one location of said at least one temporal region is responsive to a duration of said at least one said at least one heart-cycle segment; and
ii. determining at least one swing value responsive to a difference between maximum and minimum amplitude values of said at least one auscultatory sound signal within a corresponding said at least one temporal region;
g. utilizing at least one swing value of said at least one swing value as a first feature input set to a classifier to provide for determining whether or not a test-subject associated with said at least one auscultatory sound signal likely exhibits a cardiac condition;
h. generating a Short-Time Fourier Transform of said at least one auscultatory sound signal within at least one said at least one temporal region; and
i. utilizing at least one frequency-domain feature from or responsive to said Short-Time Fourier Transform of said at least one auscultatory sound signal within said at least one said at least one temporal region as a second feature input set to said classifier to provide for determining whether or not said test-subject associated with said at least one auscultatory sound signal likely exhibits a cardiac condition.
51 . A computer-implemented method of segmenting an auscultatory sound signal as recited in claim 50 , wherein at least one said at least one temporal region is associated with at least one heart sound selected from the group of heart sounds consisting of an S 3 heart sound, said S 4 heart sound and an S 5 heart sound.
52 . A computer-implemented method of segmenting an auscultatory sound signal as recited in claim 51 , wherein at least one said at least one temporal region is located relative to a start of a diastasis time interval within said diastole of said at least one said at least one heart-cycle segment.
53 . A computer-implemented method of segmenting an auscultatory sound signal as recited in claim 50 , wherein at least one said at least one temporal region is located at the end of said at least one said at least one heart-cycle segment.
54 . A computer-implemented method of segmenting an auscultatory sound signal as recited in claim 53 , wherein a duration of said at least one said at least one temporal region is between 10 and 20 percent of the duration of said at least one said at least one heart-cycle segment.
55 . A computer-implemented method of segmenting an auscultatory sound signal as recited in claim 53 , wherein a duration of said at least one said at least one temporal region is between 10 and 25 percent of the duration of said at least one said at least one heart-cycle segment.
56 . A computer-implemented method of segmenting an auscultatory sound signal as recited in claim 50 , wherein at least one said at least one temporal region either spans a start of a diastasis time interval or terminates at said start of said diastasis time interval.
57 . A computer-implemented method of segmenting an auscultatory sound signal as recited in claim 50 , wherein a duration of at least one said at least one temporal region is between 80 and 120 milliseconds.
58 . A computer-implemented method of segmenting an auscultatory sound signal, comprising:
a. receiving an electrographic signal from an electrocardiogram (ECG) sensor;
b. generating an electrographic envelope signal representing an envelope responsive to an even power of said electrographic signal;
c. locating a plurality of peaks of said electrographic envelope signal corresponding to a corresponding plurality of R-peaks of said electrographic signal;
d. receiving at least one auscultatory sound signal from a corresponding at least one auscultatory sound-or-vibration sensor;
e. segmenting said at least one auscultatory sound signal into at least one heart-cycle segment responsive to said plurality of peaks of said electrographic envelope signal, wherein each said at least one heart-cycle segment spans an entire heart cycle associated with a corresponding single heartbeat;
f. for at least one said at least one heart-cycle segment:
i. identifying at least one temporal region within each said at least one said at least one heart-cycle segment, wherein a corresponding at least one location of said at least one temporal region is responsive to a duration of said at least one said at least one heart-cycle segment; and
ii. determining at least one swing value responsive to a difference between maximum and minimum amplitude values of said at least one auscultatory sound signal within a corresponding said at least one temporal region; and
g. prior to the operation of determining said at least one swing value, generating a filtered auscultatory sound signal by filtering said at least one auscultatory sound signal within said at least one said at least one heart-cycle segment using a filter having a cutoff frequency that provides for passing either an audible range of frequencies or passing an inaudible range of frequencies below said audible range of frequencies, wherein said at least one swing value is determined responsive to said filtered auscultatory sound signal.
59 . A computer-implemented method of segmenting an auscultatory sound signal as recited in claim 58 , wherein at least one said at least one temporal region is associated with at least one heart sound selected from the group of heart sounds consisting of an S 3 heart sound, an S 4 heart sound and an S 5 heart sound.
60 . A computer-implemented method of segmenting an auscultatory sound signal as recited in claim 59 , wherein said at least one said at least one temporal region is located relative to a start of a diastasis time interval within diastole of said at least one said at least one heart-cycle segment.
61 . A computer-implemented method of segmenting an auscultatory sound signal as recited in claim 58 , wherein at least one said at least one temporal region is located relative to a start of a diastasis time interval within diastole of said at least one said at least one heart-cycle segment.
62 . A computer-implemented method of segmenting an auscultatory sound signal as recited in claim 61 , wherein at least one said at least one temporal region either spans said start of said diastasis time interval or terminates at said start of said diastasis time interval.
63 . A computer-implemented method of segmenting an auscultatory sound signal as recited in claim 58 , wherein a duration of at least one said at least one temporal region is between 80 and 120 milliseconds.
64 . A computer-implemented method of segmenting an auscultatory sound signal as recited in claim 58 , wherein at least one said at least one temporal region is located at the end of said at least one said at least one heart-cycle segment.
65 . A computer-implemented method of segmenting an auscultatory sound signal as recited in claim 64 , wherein a duration of said at least one said at least one temporal region is between 10 and 25 percent of the duration of said at least one said at least one heart-cycle segment.
66 . A computer-implemented method of segmenting an auscultatory sound signal as recited in claim 64 , wherein a duration of said at least one said at least one temporal region is between 10 and 20 percent of the duration of said at least one said at least one heart-cycle segment.
67 . A computer-implemented method of segmenting an auscultatory sound signal, comprising:
a. receiving an electrographic signal from an electrocardiogram (ECG) sensor;
b. generating an electrographic envelope signal representing an envelope responsive to an even power of said electrographic signal;
c. locating a plurality of peaks of said electrographic envelope signal corresponding to a corresponding plurality of R-peaks of said electrographic signal;
d. receiving at least one auscultatory sound signal from a corresponding at least one auscultatory sound-or-vibration sensor;
e. segmenting said at least one auscultatory sound signal into at least one heart-cycle segment responsive to said plurality of peaks of said electrographic envelope signal, wherein each said at least one heart-cycle segment spans an entire heart cycle associated with a corresponding single heartbeat;
f. for at least one said at least one heart-cycle segment:
i. identifying at least one temporal region within each said at least one said at least one heart-cycle segment, wherein a corresponding at least one location of said at least one temporal region is responsive to a duration of said at least one said at least one heart-cycle segment; and
ii. determining at least one swing value responsive to a difference between maximum and minimum amplitude values of said at least one auscultatory sound signal within a corresponding said at least one temporal region;
g. utilizing a first swing value of said at least one swing value as a first feature input set to a classifier to provide for determining whether or not a test-subject associated with said at least one auscultatory sound signal likely exhibits a cardiac condition, wherein said first swing value is associated with a corresponding said at least one temporal region located at an end of said at least one said at least one heart-cycle segment;
h. determining a ratio of said first swing value and a second swing value of said at least one swing value as a measure of a likelihood that said test-subject exhibits an S 4 heart sound, wherein said second swing value is associated with a corresponding said at least one temporal region associated with an S 2 -sound time interval located at a beginning of diastole of said at least one said at least one heart-cycle segment; and
i. utilizing said ratio of said first swing value and said second swing value as a third feature input set to said classifier to provide for determining whether or not said test-subject likely exhibits a cardiac condition.
68 . A computer-implemented method of segmenting an auscultatory sound signal as recited in claim 67 , wherein at least one said at least one temporal region is associated with at least one heart sound selected from the group of heart sounds consisting of an S 3 heart sound, said S 4 heart sound and an S 5 heart sound.
69 . A computer-implemented method of segmenting an auscultatory sound signal as recited in claim 67 , wherein at least one said at least one temporal region is located relative to a start of a diastasis time interval within said diastole of said at least one said at least one heart-cycle segment.
70 . A computer-implemented method of segmenting an auscultatory sound signal as recited in claim 67 , wherein a duration of at least one said at least one temporal region is between 80 and 120 milliseconds.
71 . A computer-implemented method of segmenting an auscultatory sound signal as recited in claim 67 , wherein a duration of said corresponding said at least one temporal region associated with said first swing value is between 10 and 25 percent of the duration of said at least one said at least one heart-cycle segment.
72 . A computer-implemented method of segmenting an auscultatory sound signal as recited in claim 67 , wherein a duration of said corresponding said at least one temporal region associated with said first swing value is between 10 and 20 percent of the duration of said at least one said at least one heart-cycle segment.
73 . A computer-implemented method of segmenting an auscultatory sound signal, comprising:
a. receiving an electrographic signal from an electrocardiogram (ECG) sensor;
b. generating an electrographic envelope signal representing an envelope responsive to an even power of said electrographic signal;
c. locating a plurality of peaks of said electrographic envelope signal corresponding to a corresponding plurality of R-peaks of said electrographic signal;
d. receiving at least one auscultatory sound signal from a corresponding at least one auscultatory sound-or-vibration sensor;
e. segmenting said at least one auscultatory sound signal into at least one heart-cycle segment responsive to said plurality of peaks of said electrographic envelope signal, wherein each said at least one heart-cycle segment spans an entire heart cycle associated with a corresponding single heartbeat;
f. for at least one said at least one heart-cycle segment:
i. identifying at least one temporal region within each said at least one said at least one heart-cycle segment, wherein a corresponding at least one location of said at least one temporal region is responsive to a duration of said at least one said at least one heart-cycle segment; and
ii. determining at least one swing value responsive to a difference between maximum and minimum amplitude values of said at least one auscultatory sound signal within a corresponding said at least one temporal region; and
g. determining a ratio of a first swing value of said at least one swing value and a second swing value of said at least one swing value as a measure of a likelihood that a test-subject associated with said at least one auscultatory sound signal exhibits an S 4 heart sound, wherein said first swing value is associated with a corresponding said at least one temporal region associated with an S 4 -sound time interval located at an end of said at least one said at least one heart-cycle segment, and said second swing value is associated with a corresponding said at least one temporal region associated with an S 2 -sound time interval located at a beginning of diastole of said at least one said at least one heart-cycle segment.
74 . A computer-implemented method of segmenting an auscultatory sound signal as recited in claim 73 , wherein at least one said at least one temporal region is associated with at least one heart sound selected from the group of heart sounds consisting of an S 3 heart sound, said S 4 heart sound and an S 5 heart sound.
75 . A computer-implemented method of segmenting an auscultatory sound signal as recited in claim 73 , wherein at least one said at least one temporal region is located relative to a start of a diastasis time interval within said diastole of said at least one said at least one heart-cycle segment.
76 . A computer-implemented method of segmenting an auscultatory sound signal as recited in claim 73 , wherein a duration of at least one said at least one temporal region is between 80 and 120 milliseconds.
77 . A computer-implemented method of segmenting an auscultatory sound signal as recited in claim 73 , wherein a duration of said corresponding said at least one temporal region associated with said first swing value is between 10 and 25 percent of the duration of said at least one said at least one heart-cycle segment.
78 . A computer-implemented method of segmenting an auscultatory sound signal as recited in claim 73 , wherein a duration of said corresponding said at least one temporal region associated with said first swing value is between 10 and 20 percent of the duration of said at least one said at least one heart-cycle segment.