Signal feature extraction and data compression method based on combined acquisition of electrocardiogram and seismocardiogram

Abstract

The signal feature extraction and data compression method based on electrocardiogram and heart vibration diagram joint acquisition disclosed by the application realizes feature point labeling of electrocardiogram signals and heart vibration signals by using a light algorithm, has high real-time performance, and is suitable for real-time portable heart health monitoring mode. The algorithm realizes R wave recognition by using the Shannon information entropy of the normalized first-order differential signal, realizes IM point and AC point positioning by using R wave positioning and the envelope of the high-frequency component of the heart vibration signal, improves the accuracy of feature point recognition, and is suitable for measurement results of electrocardiogram signals and heart vibration signals in various scenes. The method extracts complete signal features according to the feature point recognition results of the electrocardiogram signals and the heart vibration signals, realizes data compression by information granulation processing on the time series of different signal features, can meet the light weight of calculation and storage while providing comprehensive information for heart health condition diagnosis, and is beneficial to improving the accuracy of diagnosis.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of medical equipment information processing, and particularly relates to a signal feature extraction and data compression method based on joint collection of electrocardiograms and seismocardiograms. BACKGROUND

[0002] Cardiovascular diseases seriously threaten human life and health, and the prevention and treatment of cardiovascular diseases has always been one of the key works in the medical and health field. Electrocardiograms and seismocardiograms both contain rich medical information, and reflect the activity state of the human heart from the electrical dimension and the mechanical dimension, respectively. Since the activity of the heart has the characteristics of electrical-mechanical coupling, joint collection of electrocardiograms and seismocardiograms can obtain more rich physiological parameters of the heart, which is conducive to improving the accuracy of cardiovascular disease diagnosis, thereby better serving the prevention and treatment of cardiovascular diseases.

[0003] In order to adapt to future new medical models such as remote medical treatment, intelligent medical treatment and distributed medical treatment, portable intelligent heart health monitoring devices are continuously developed, and real-time portable heart health monitoring models put forward new requirements for intelligent processing and analysis of signals. A large number of medical practices show that characteristic wave bands in the electrocardio signal, such as P wave, QRS wave and T wave, are closely related to the activity of specific parts in the heart, and characteristic points in the seismocardiogram, such as MC point, AO point, RE point, AC point, MO point, IM point, IC point and RF point, also correspond to specific physiological events in the heart movement process, so the joint analysis of these characteristic waves can realize accurate cardiovascular disease diagnosis.

[0004] Due to the limited running speed and storage capacity of ordinary single-chip microcomputers, light-weight algorithms are required to realize real-time signal processing and analysis, and traditional signal feature extraction algorithms often need to occupy a large amount of computing resources, which cannot meet the requirements of intelligent processing and analysis of signals. On the other hand, the repeatability of real-time signal collection results will lead to a large amount of redundancy of stored data, which brings pressure to data storage. SUMMARY

[0005] In order to overcome the above-mentioned shortcomings of the prior art, the purpose of the present application is to provide a signal feature extraction and data compression method based on joint collection of electrocardiograms and seismocardiograms, which has high accuracy, integrity and real-time performance, and can meet the requirements of intelligent processing and analysis of signals for real-time heart health monitoring mode of portable intelligent heart health monitoring devices under new medical models, and provide support for intelligent diagnosis of heart health status.

[0006] In order to achieve the above-mentioned purpose, the technical scheme adopted by the present application is as follows:

[0007] A signal feature extraction and data compression method based on electrocardiogram and seismocardiogram joint acquisition, comprising the following steps:

[0008] S1, using a filtering method to process the original electrocardiogram signal and the seismocardiogram signal synchronously collected to filter out the noise components contained in the original signal, to obtain the denoised electrocardiogram signal and the seismocardiogram signal, and to obtain the high-frequency component of the seismocardiogram signal at the same time;

[0009] S2, performing first-order differential calculation on the denoised electrocardiogram signal in S1, and obtaining the peak envelope line of the Shannon information entropy of the first-order differential signal;

[0010] S3, locating the position of the R wave of the electrocardiogram signal by using the peak envelope line of the Shannon information entropy in S2;

[0011] S4, substituting the R wave position of the electrocardiogram signal determined in S3 into the denoised electrocardiogram signal in S1 to realize the positioning of the P wave, the Q wave, the S wave and the T wave in the electrocardiogram signal;

[0012] S5, obtaining the peak envelope line of the high-frequency component of the seismocardiogram signal in S1, and combining the position information of the R wave of the electrocardiogram signal determined in S3 to realize the positioning of the IM point and the AC point of the seismocardiogram signal;

[0013] S6, substituting the positions of the IM point and the AC point of the seismocardiogram signal determined in S5 into the denoised seismocardiogram signal in S1 to realize the positioning of the MC point, the AO point, the RE point, the MO point, the IC point and the RF point in the seismocardiogram signal;

[0014] S7, using the P wave, the Q wave, the R wave, the S wave and the T wave of the electrocardiogram signal and the MC point, the AO point, the RE point, the AC point, the MO point, the IM point, the IC point and the RF point of the seismocardiogram signal determined in the position to respectively obtain the time interval feature and the amplitude feature of the electrocardiogram signal and the seismocardiogram signal and the time interval feature of the combination of the electrocardiogram signal and the seismocardiogram signal, to use them as the input data of the subsequent heart health condition diagnosis algorithm;

[0015] S8, using the column storage method to store the time interval feature and the amplitude feature of the electrocardiogram signal and the seismocardiogram signal and the time interval feature of the combination of the electrocardiogram signal and the seismocardiogram signal, and performing information granulation processing on the time sequence of different signal features to remove bad points and highly similar data to realize data compression.

[0016] Preferably, the S1 step specifically comprises: performing smoothing processing on the original electrocardiogram signal, and then performing high-pass filtering, power trap wave filtering and low-pass filtering processing respectively to obtain the denoised electrocardiogram signal; performing smoothing processing on the original seismocardiogram signal, and then performing high-pass filtering and low-pass filtering processing respectively to obtain the denoised seismocardiogram signal; and obtaining the high-frequency component of the seismocardiogram signal by performing high-pass filtering on the original seismocardiogram signal.

[0017] Further preferably, the S2 specific step is: performing first-order differential calculation on the denoised electrocardio signal, and performing normalization processing on the calculated first-order differential signal, calculating the Shannon information entropy of the normalized first-order differential signal and obtaining the peak envelope line thereof.

[0018] Further preferably, the S3 step is specifically: identifying the peak point of the envelope line and multiplying the amplitude of the peak point by a fixed coefficient a (0.01 < a < 0.99), and the position of the point corresponding to the amplitude on the left side of the peak point is the detection starting position of the R wave, and the first maximum value point after substituting the starting position into the denoised electrocardio signal is the position of the R wave.

[0019] Further preferably, the S4 step is specifically: the first minimum value point before each R wave corresponds to the Q wave, the first minimum value point after each R wave corresponds to the S wave, the first maximum value point before each Q wave corresponds to the P wave, and the first maximum value point after each S wave corresponds to the T wave.

[0020] Further preferably, the S5 step is specifically: obtaining the peak envelope line of the high-frequency component of the heart shock signal, and the fixed window of 100 ms range before and after the point corresponding to the R wave of the electrocardio signal on the envelope line is the window for identifying the IM point, identifying the peak point in the window range and multiplying the amplitude of the peak point by a fixed coefficient β1 (0.01 < β1 < 0.99), and the position of the point corresponding to the amplitude on the left side of the peak point is the detection starting position of the IM point, and the first minimum value point after substituting the starting position into the denoised heart shock signal is the IM point, and the first peak point after the fixed window corresponding to the R wave is identified and the amplitude of the peak point is multiplied by a fixed coefficient β2 (0.01 < β2 < 0.99), and the position of the point corresponding to the amplitude on the left side of the peak point is the detection starting position of the AC point, and the first maximum value point after substituting the starting position into the denoised heart shock signal is the AC point.

[0021] Further preferably, the S6 step is specifically: the first maximum value point before each IM point corresponds to the MC point, the first maximum value point after each IM point corresponds to the AO point, the first minimum value point after each AO wave corresponds to the IC point, the first maximum value point after each IC point corresponds to the RE point, the first minimum value point after each AC point corresponds to the MO point, and the first maximum value point after each MO point corresponds to the RF point.

[0022] Further preferably, the S7 step specifically comprises: marking the P wave, Q wave, R wave, S wave and T wave of the determined position electrocardio signal and the MC point, AO point, RE point, AC point, MO point, IM point, IC point and RF point of the heart shock signal on the denoised electrocardio signal and heart shock signal respectively, and acquiring the time interval features and amplitude features of the electrocardio signal and heart shock signal and the time interval features of the electrocardio signal and heart shock signal in combination as input data of the subsequent heart health condition diagnosis algorithm.

[0023] Further preferably, the S8 step specifically comprises: adopting a column storage method to store the time interval features and amplitude features of the electrocardio signal and heart shock signal and the time interval features of the electrocardio signal and heart shock signal in combination, wherein the signal features exceeding the specified threshold are determined as bad points and are not stored, the fluctuation points of different signal feature time sequences are identified, the signal feature time sequences between adjacent fluctuation points form information particles, the first point in the information particle is selected as the feature point of the information particle, the feature time of all signal feature information particles is extracted and a new time sequence is formed, the corresponding of all signal features at each time point is performed, and the data compression is completed.

[0024] Compared with the prior art, the present application has the following beneficial technical effects:

[0025] The signal feature extraction and data compression method based on electrocardiogram and heart shock map joint acquisition disclosed in the present application realizes the feature point labeling of the electrocardio signal and heart shock signal by using a lightweight algorithm, avoids the problem that the traditional signal feature extraction algorithm often needs to occupy a large amount of computing resources and cannot meet the intelligent processing and analysis requirements of the signal, has high real-time performance, and is suitable for real-time portable heart health monitoring mode. The algorithm realizes the recognition of the R wave by using the Shannon information entropy of the normalized first-order differential signal, realizes the positioning of the IM point and AC point by using the positioning of the R wave and the envelope of the high-frequency component of the heart shock signal, improves the accuracy of the feature point recognition, and is suitable for the measurement results of the electrocardio signal and heart shock signal in various scenes. The method extracts complete signal features according to the feature point recognition results of the electrocardio signal and heart shock signal, including the time interval features and amplitude features of the electrocardio signal and heart shock signal and the time interval features of the electrocardio signal and heart shock signal in combination, realizes the data compression by performing the information particle processing on the time sequences of different signal features, can meet the lightweight computing and storage while providing comprehensive information for the heart health condition diagnosis, and is beneficial to improving the accuracy of the diagnosis. BRIEF DESCRIPTION OF DRAWINGS

[0026] Figure 1 The method flowchart of the present application is shown in the figure;

[0027] Figure 2 The processing process and feature point labeling results of the electrocardio signal and heart shock signal of the embodiment are shown in the figure. DETAILED DESCRIPTION

[0028] The application will be further described in the following with the attached drawings and specific examples, which are an explanation of the application rather than a limitation.

[0029] As Figure 1 The signal feature extraction and data compression method based on electrocardiogram and seismocardiogram joint acquisition of the application includes:

[0030] S1, using a filtering method to process the original electrocardiogram signal and the seismocardiogram signal synchronously collected to filter out the noise components contained in the original signal, to obtain the denoised electrocardiogram signal and the seismocardiogram signal, and to obtain the high-frequency component of the seismocardiogram signal at the same time. Specifically, for the original electrocardiogram signal, smooth processing is performed, followed by high-pass filtering, power frequency notch filtering and low-pass filtering to obtain the denoised electrocardiogram signal. For the original seismocardiogram signal, smooth processing is performed, followed by high-pass filtering and low-pass filtering to obtain the denoised seismocardiogram signal. At the same time, the original seismocardiogram signal is subjected to high-pass filtering to obtain the high-frequency part of the seismocardiogram signal.

[0031] S2, first-order differential calculation is performed on the denoised electrocardiogram signal in S1, and the peak envelope line of the Shannon information entropy of the first-order differential signal is obtained. Specifically, first-order differential calculation is performed on the denoised electrocardiogram signal, and the calculated first-order differential signal is subjected to normalization processing. The Shannon information entropy of the normalized first-order differential signal is calculated and the peak envelope line thereof is obtained.

[0032] S3, the peak envelope line of the Shannon information entropy in S2 is used to locate the position of the R wave of the electrocardiogram signal. Specifically, the peak points of the envelope line are identified and the amplitude of the peak points is multiplied by a fixed coefficient α (0.01 < α < 0.99). The position of the point on the left side of the peak point whose amplitude corresponds to the peak point is the starting position of the detection of the R wave. The starting position is substituted into the denoised electrocardiogram signal, and the first maximum value point thereafter is the position of the R wave.

[0033] S4, the determined R wave position of the electrocardiogram signal in S3 is substituted into the denoised electrocardiogram signal in S1 to realize the positioning of the P wave, the Q wave, the S wave and the T wave in the electrocardiogram signal. Specifically, the first minimum value point before each R wave corresponds to the Q wave, the first minimum value point after each R wave corresponds to the S wave, the first maximum value point before each Q wave corresponds to the P wave, and the first maximum value point after each S wave corresponds to the T wave.

[0034] S5, the peak envelope of the high frequency component of the S1 center shock signal is obtained, and the positions of the IM point and the AC point of the heart shock signal are located by combining the position information of the R wave of the S3 electrocardio signal. Specifically, the peak envelope of the high frequency component of the heart shock signal is obtained, and the fixed window of 100 ms before and after the point corresponding to the R wave of the electrocardio signal on the envelope line is the window for identifying the IM point. The peak point in the identification window range is identified, and the amplitude of the peak point is multiplied by a fixed coefficient β1 (0.01 < β1 < 0.99). The position of the point on the left side of the peak point corresponding to the amplitude is the starting position of the IM point detection. The starting position is substituted into the denoised heart shock signal, and the first minimum value point after the starting position is the IM point. The first peak point after the fixed window corresponding to the R wave is identified, and the amplitude of the peak point is multiplied by a fixed coefficient β2 (0.01 < β2 < 0.99). The position of the point on the left side of the peak point corresponding to the amplitude is the starting position of the AC point detection. The starting position is substituted into the denoised heart shock signal, and the first maximum value point after the starting position is the AC point.

[0035] S6, the positions of the IM point and the AC point of the heart shock signal determined in S5 are substituted into the denoised heart shock signal in S1, and the positions of the MC point, the AO point, the RE point, the MO point and the IC point in the heart shock signal are located. Specifically, the first maximum value point before each IM point corresponds to the MC point, the first maximum value point after each IM point corresponds to the AO point, the first minimum value point after each AO wave corresponds to the IC point, the first maximum value point after each IC point corresponds to the RE point, the first minimum value point after each AC point corresponds to the MO point, and the first maximum value point after each MO point corresponds to the RF point.

[0036] S7, the P wave, the Q wave, the R wave, the S wave and the T wave of the electrocardio signal with the determined positions and the MC point, the AO point, the RE point, the AC point, the MO point, the IM point and the IC point of the heart shock signal are used to obtain the time interval features and the amplitude features of the electrocardio signal and the heart shock signal and the time interval features of the combination of the electrocardio signal and the heart shock signal, which are used as the input data of the subsequent heart health condition diagnosis algorithm. Specifically, the P wave, the Q wave, the R wave, the S wave and the T wave of the electrocardio signal with the determined positions and the MC point, the AO point, the RE point, the AC point, the MO point, the IM point, the IC point and the RF point of the heart shock signal are marked on the denoised electrocardio signal and the denoised heart shock signal respectively, and the corresponding amplitude information is obtained. The time interval features and the amplitude features of the electrocardio signal and the heart shock signal and the time interval features of the combination of the electrocardio signal and the heart shock signal are obtained, which are used as the input data of the subsequent heart health condition diagnosis algorithm.

[0037] S8, the method of column storage is used to store the time interval features and amplitude features of the electrocardio signals and heart shock signals and the time interval features of the combination of the electrocardio signals and heart shock signals, wherein the signal features exceeding the specified threshold are determined as bad points and are not stored, the fluctuation points of different signal feature time sequences are identified, the signal feature time sequences between adjacent fluctuation points form information particles, the first point in the information particle is selected as the feature point of the information particle, the feature time of all signal feature information particles is extracted and a new time sequence is formed, the corresponding of all signal features is performed at each time point, and the data compression is completed.

[0038] The application is further explained in the following with a specific embodiment:

[0039] As Figure 2 In (a)-(e), the original electrocardio signals are subjected to secondary exponential smoothing, and then subjected to 1Hz high-pass filtering, 50Hz power frequency notch filtering and 100Hz low-pass filtering to obtain the denoised electrocardio signals. The original heart shock signals are subjected to secondary exponential smoothing, and then subjected to 1Hz high-pass filtering and 30Hz low-pass filtering to obtain the denoised heart shock signals. At the same time, the original heart shock signals are subjected to 20Hz high-pass filtering to obtain the high-frequency components of the heart shock signals.

[0040] As Figure 2 In (f), the denoised electrocardio signals are subjected to first-order differential calculation, and the calculated first-order differential signals are subjected to normalization processing. The specific method of normalization is to obtain the absolute value of the first-order differential signals and convert them to the range of [0, 1] using linearization method. The Shannon information entropy of the normalized first-order differential signals is calculated and the peak envelope line thereof is obtained. Since the amplitude changes greatly near the R wave of the electrocardio signals, the peak value of the envelope line appears near the R wave.

[0041] Further, the peak points of the envelope line are identified, and the amplitudes of the peak points are multiplied by a fixed coefficient α, and α is 0.5. As Figure 2 In (f), the point on the left side of the peak point with the amplitude consistent with the calculated amplitude is marked on the envelope line, and the position of the point is the starting position of the detection of the R wave. The starting position is substituted into the denoised electrocardio signals, and the first maximum value point after the starting position is the R wave.

[0042] As Figure 2 In (h), the first minimum value point before each R wave on the denoised electrocardio signals corresponds to the Q wave, the first minimum value point after each R wave corresponds to the S wave, the first maximum value point before each Q wave corresponds to the P wave, and the first maximum value point after each S wave corresponds to the T wave. The identified R wave, P wave, Q wave, R wave, S wave and T wave are marked on the denoised electrocardio signals.

[0043] AsFigure 2 In the middle (g), the peak envelope of the high-frequency component of the heart shock signal is obtained. Since the amplitude of the heart shock signal changes obviously near the IM point and the AC point, the envelope line appears corresponding peaks near the IM point and the AC point. The peak point in the fixed window of 100 ms before and after the point corresponding to the R wave of the electrocardiogram signal on the envelope line is identified. The amplitude of the peak point is multiplied by a fixed coefficient β1, and the value of β1 is 0.9. The position of the point with the amplitude consistent with the calculated amplitude on the left side of the peak point is the starting position of the IM detection. The first minimum point after the starting position in the denoised heart shock signal is the IM point. The first peak point after the fixed window corresponding to the R wave is identified, and the amplitude of the peak point is multiplied by a fixed coefficient β2, and the value of β2 is 0.6. The position of the point with the amplitude consistent with the calculated amplitude on the left side of the peak point is the starting position of the AC detection. The first maximum point after the starting position in the denoised heart shock signal is the AC point.

[0044] As Figure 2 In the middle (i), the first maximum point before each IM point corresponds to the MC point, the first maximum point after each IM point corresponds to the AO point, the first minimum point after each AO wave corresponds to the IC point, the first maximum point after each IC point corresponds to the RE point, the first minimum point after each AC point corresponds to the MO point, and the first maximum point after each MO point corresponds to the RF point. The identified MC point, AO point, RE point, AC point, MO point, IM point, IC point, and RF point are marked on the denoised heart shock signal.

[0045] Further, according to the feature point marking of the electrocardiogram signal and the heart shock signal, the time interval features and amplitude features of the electrocardiogram signal and the heart shock signal and the time interval features of the combination of the electrocardiogram signal and the heart shock signal are extracted, wherein the time interval features of the electrocardiogram signal include R-R, P-Q, P-R, P-T, Q-R, R-S, R-T, S-T, etc., the amplitude features include the amplitudes of R wave, P wave, Q wave, R wave, S wave, and T wave, and |P-Q / R-S|, |P-Q / Q-R|, |P-Q / S-T|, |Q-R / R-S|, |Q-R / S-T|, |R-S / S-T|, etc. The time interval features of the heart shock signal include AO-AO, MC-AO, MC-MO, AO-AC, AO-RE, AC-MO, MO-RF, etc., the amplitude features include the amplitudes of MC point, AO point, RE point, AC point, MO point, IM point, IC point, and RF point, and |IM-AO / AO-IC|, |IM-AO / IC-RE|, |AO-IC / IC-RE|, |AC-MO / MO-RF|, etc. The time interval features of the combination of the electrocardiogram signal and the heart shock signal include R-AO, R-RE, etc. These features will be used as input data for the subsequent heart health condition diagnosis algorithm.

[0046] Further, the time interval features and amplitude features of the electrocardiogram signals and heart shock signals and the time interval features of the combination of the electrocardiogram signals and heart shock signals are classified and stored by using the column storage method, the threshold values of different information features are set and compared, the signal features exceeding the specified threshold value are determined as bad points and are not stored, the extreme points of different time sequences are identified, the adjacent extreme points in different time sequences are compared to identify the candidate fluctuation points, if the properties of the adjacent candidate fluctuation points are consistent, the front candidate fluctuation point is not considered as a fluctuation point, if the properties of the adjacent candidate fluctuation points are inconsistent, both of the candidate fluctuation points are considered as fluctuation points,

[0047] The signal feature time sequence between adjacent fluctuation points constitutes an information particle, and the first point in the information particle is selected as the feature point of the information particle, if the length of the time sequence in the information particle exceeds a certain threshold value, the threshold value is 1 hour here, the corresponding fluctuation point is inserted when the threshold value is reached, the further division of the information particle is realized, the feature time of all signal feature information particles is extracted and constitutes a new time sequence, for each time point in the time sequence, the feature value of the closest signal feature information particle before the time point is selected as the signal feature value of the time point, and the compression of the overall data is completed in this way.

Claims

1. A signal feature extraction and data compression method based on combined acquisition of electrocardiogram and seismocardiogram, characterized in that, The method comprises the following steps: S1, using a filtering method to process the original electrocardio signals and the original heart shock signals collected synchronously to filter out the noise components contained in the original signals, to obtain the denoised electrocardio signals and the denoised heart shock signals, and to obtain the high-frequency components of the heart shock signals; S2, performing first-order differential calculation on the denoised electrocardio signals in S1, and obtaining the peak envelope line of the Shannon information entropy of the first-order differential signals; S3, locating the positions of the R waves of the electrocardio signals by using the peak envelope line of the Shannon information entropy in S2; S4, substituting the determined positions of the R waves of the electrocardio signals in S3 into the denoised electrocardio signals in S1 to realize the positioning of the P waves, the Q waves, the S waves and the T waves in the electrocardio signals; S5, obtaining the peak envelope line of the high-frequency components of the heart shock signals in S1, and combining the position information of the determined R waves of the electrocardio signals in S3 to realize the positioning of the IM points and the AC points of the heart shock signals; S6, substituting the determined positions of the IM points and the AC points of the heart shock signals in S5 into the denoised heart shock signals in S1 to realize the positioning of the MC points, the AO points, the RE points, the MO points, the IC points and the RF points in the heart shock signals; S7, using the P waves, the Q waves, the R waves, the S waves and the T waves of the electrocardio signals and the MC points, the AO points, the RE points, the AC points, the MO points, the IM points, the IC points and the RF points of the heart shock signals to obtain the time interval features and the amplitude features of the electrocardio signals and the heart shock signals and the time interval features of the combination of the electrocardio signals and the heart shock signals, which are used as the input data of the subsequent heart health condition diagnosis algorithm; S8, using the column storage method to store the time interval features and the amplitude features of the electrocardio signals and the heart shock signals and the time interval features of the combination of the electrocardio signals and the heart shock signals, performing information granulation processing on the time sequences of different signal features, removing bad points and highly similar data to realize data compression; The step S5 specifically comprises: obtaining the peak envelope line of the high-frequency components of the heart shock signals, the fixed window of 100 ms in front of and behind the point corresponding to the R wave of the electrocardio signals on the envelope line is the window for identifying the IM point, identifying the peak point in the window and multiplying the amplitude of the peak point by a fixed coefficient β1, the position of the point on the left side of the peak point corresponding to the amplitude is the detection starting position of the IM point, substituting the detection starting position of the M point into the denoised heart shock signals, and the first minimum value point thereafter is the IM point, identifying the first peak point after the fixed window corresponding to the R wave and multiplying the amplitude of the peak point by a fixed coefficient β2, the position of the point on the left side of the peak point corresponding to the amplitude is the detection starting position of the AC point, substituting the detection starting position of the AC point into the denoised heart shock signals, and the first maximum value point thereafter is the AC point; wherein 0.01<β1<0.99 and 0.01<β2<0.

99. S6 step is specifically: the first maximum point before each IM point corresponds to the MC point, the first maximum point after each IM point corresponds to the AO point, the first minimum point after each AO wave corresponds to the IC point, the first maximum point after each IC point corresponds to the RE point, the first minimum point after each AC point corresponds to the MO point, and the first maximum point after each MO point corresponds to the RF point; S7 step is specifically: the P wave, Q wave, R wave, S wave and T wave of the determined position of the electrocardio signal and the MC point, AO point, RE point, AC point, MO point, IM point, IC point and RF point of the heart shock signal are marked on the denoised electrocardio signal and heart shock signal respectively, the time interval characteristics and amplitude characteristics of the electrocardio signal and heart shock signal and the time interval characteristics of the combination of the electrocardio signal and heart shock signal are obtained respectively, which are used as the input data of the subsequent heart health condition diagnosis algorithm; S8 step is specifically: the time interval characteristics and amplitude characteristics of the electrocardio signal and heart shock signal and the time interval characteristics of the combination of the electrocardio signal and heart shock signal are classified and stored by using column storage method, wherein the signal characteristics exceeding the specified threshold are determined as bad points and are not stored, the fluctuation points of different signal characteristic time sequences are identified, the signal characteristic time sequences between adjacent fluctuation points constitute information particles, the first point in the information particle is selected as the characteristic point of the information particle, the characteristic time of all signal characteristic information particles is extracted and a new time sequence is formed, the corresponding of all signal characteristics is performed at each time point, and the data compression is completed.

2. The signal feature extraction and data compression method based on combined acquisition of electrocardiogram and seismocardiogram according to claim 1, characterized in that, S1 step is specifically: the original electrocardio signal is smoothed, then high-pass filtering, power trap wave and low-pass filtering are performed respectively to obtain the denoised electrocardio signal, the original heart shock signal is smoothed, then high-pass filtering and low-pass filtering are performed respectively to obtain the denoised heart shock signal, and at the same time, the high-frequency component of the heart shock signal is obtained by high-pass filtering the original heart shock signal.

3. The signal feature extraction and data compression method based on combined acquisition of electrocardiogram and seismocardiogram according to claim 1, characterized in that, S2 step is specifically: the first-order differential signal of the denoised electrocardio signal is calculated, and the calculated first-order differential signal is normalized, the Shannon information entropy of the normalized first-order differential signal is calculated, and the peak envelope line thereof is obtained.

4. The signal feature extraction and data compression method based on combined acquisition of electrocardiogram and seismocardiogram according to claim 1, characterized in that, S3 step is specifically: the peak points of the envelope line are identified, and the amplitude of the peak points is multiplied by a fixed coefficient α, the position of the point corresponding to the amplitude on the left side of the peak point is the detection starting position of the R wave, the detection starting position of the R wave is substituted into the denoised electrocardio signal, and the first maximum point after the detection starting position is the position of the R wave, wherein 0.01<α<0.

99.

5. The signal feature extraction and data compression method based on combined acquisition of electrocardiogram and seismocardiogram according to claim 1, characterized in that, S4 step is specifically: the first minimum point before each R wave corresponds to the Q wave, the first minimum point after each R wave corresponds to the S wave, the first maximum point before each Q wave corresponds to the P wave, and the first maximum point after each S wave corresponds to the T wave.

Images