CN115363598A - Electrocardiosignal processing method and system - Google Patents
Electrocardiosignal processing method and system Download PDFInfo
- Publication number
- CN115363598A CN115363598A CN202211135001.8A CN202211135001A CN115363598A CN 115363598 A CN115363598 A CN 115363598A CN 202211135001 A CN202211135001 A CN 202211135001A CN 115363598 A CN115363598 A CN 115363598A
- Authority
- CN
- China
- Prior art keywords
- feature
- characteristic
- electrocardiosignals
- subset
- electrocardiosignal
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000003672 processing method Methods 0.000 title claims abstract description 12
- 238000004458 analytical method Methods 0.000 claims abstract description 51
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 35
- 238000012549 training Methods 0.000 claims abstract description 33
- 238000012545 processing Methods 0.000 claims abstract description 29
- 238000000034 method Methods 0.000 claims abstract description 28
- 238000007637 random forest analysis Methods 0.000 claims abstract description 16
- 238000007781 pre-processing Methods 0.000 claims description 9
- 230000000747 cardiac effect Effects 0.000 claims description 8
- 238000010586 diagram Methods 0.000 claims description 8
- 238000004364 calculation method Methods 0.000 claims description 3
- 239000013598 vector Substances 0.000 description 15
- 238000003066 decision tree Methods 0.000 description 9
- 230000008859 change Effects 0.000 description 7
- 201000010099 disease Diseases 0.000 description 4
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 4
- 230000002159 abnormal effect Effects 0.000 description 3
- 230000000694 effects Effects 0.000 description 3
- 230000006870 function Effects 0.000 description 3
- 238000005070 sampling Methods 0.000 description 3
- 239000000126 substance Substances 0.000 description 3
- 238000012360 testing method Methods 0.000 description 3
- 206010003658 Atrial Fibrillation Diseases 0.000 description 2
- 210000005036 nerve Anatomy 0.000 description 2
- 230000002889 sympathetic effect Effects 0.000 description 2
- 230000001515 vagal effect Effects 0.000 description 2
- IYLGZMTXKJYONK-ACLXAEORSA-N (12s,15r)-15-hydroxy-11,16-dioxo-15,20-dihydrosenecionan-12-yl acetate Chemical compound O1C(=O)[C@](CC)(O)C[C@@H](C)[C@](C)(OC(C)=O)C(=O)OCC2=CCN3[C@H]2[C@H]1CC3 IYLGZMTXKJYONK-ACLXAEORSA-N 0.000 description 1
- 238000012935 Averaging Methods 0.000 description 1
- 208000024172 Cardiovascular disease Diseases 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 210000003403 autonomic nervous system Anatomy 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 238000013145 classification model Methods 0.000 description 1
- 230000007423 decrease Effects 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 230000016507 interphase Effects 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 230000036651 mood Effects 0.000 description 1
- QVRVXSZKCXFBTE-UHFFFAOYSA-N n-[4-(6,7-dimethoxy-3,4-dihydro-1h-isoquinolin-2-yl)butyl]-2-(2-fluoroethoxy)-5-methylbenzamide Chemical compound C1C=2C=C(OC)C(OC)=CC=2CCN1CCCCNC(=O)C1=CC(C)=CC=C1OCCF QVRVXSZKCXFBTE-UHFFFAOYSA-N 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 230000000717 retained effect Effects 0.000 description 1
- IYLGZMTXKJYONK-UHFFFAOYSA-N ruwenine Natural products O1C(=O)C(CC)(O)CC(C)C(C)(OC(C)=O)C(=O)OCC2=CCN3C2C1CC3 IYLGZMTXKJYONK-UHFFFAOYSA-N 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
Images
Classifications
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/24—Detecting, measuring or recording bioelectric or biomagnetic signals of the body or parts thereof
- A61B5/316—Modalities, i.e. specific diagnostic methods
- A61B5/318—Heart-related electrical modalities, e.g. electrocardiography [ECG]
- A61B5/346—Analysis of electrocardiograms
- A61B5/349—Detecting specific parameters of the electrocardiograph cycle
- A61B5/352—Detecting R peaks, e.g. for synchronising diagnostic apparatus; Estimating R-R interval
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/02—Detecting, measuring or recording pulse, heart rate, blood pressure or blood flow; Combined pulse/heart-rate/blood pressure determination; Evaluating a cardiovascular condition not otherwise provided for, e.g. using combinations of techniques provided for in this group with electrocardiography or electroauscultation; Heart catheters for measuring blood pressure
- A61B5/024—Detecting, measuring or recording pulse rate or heart rate
- A61B5/02405—Determining heart rate variability
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/24—Detecting, measuring or recording bioelectric or biomagnetic signals of the body or parts thereof
- A61B5/316—Modalities, i.e. specific diagnostic methods
- A61B5/318—Heart-related electrical modalities, e.g. electrocardiography [ECG]
- A61B5/346—Analysis of electrocardiograms
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
- A61B5/7235—Details of waveform analysis
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
- A61B5/7235—Details of waveform analysis
- A61B5/7264—Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems
- A61B5/7267—Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems involving training the classification device
Landscapes
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Cardiology (AREA)
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Animal Behavior & Ethology (AREA)
- Public Health (AREA)
- Biomedical Technology (AREA)
- Heart & Thoracic Surgery (AREA)
- Medical Informatics (AREA)
- Molecular Biology (AREA)
- Surgery (AREA)
- Biophysics (AREA)
- General Health & Medical Sciences (AREA)
- Pathology (AREA)
- Veterinary Medicine (AREA)
- Artificial Intelligence (AREA)
- Physiology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Psychiatry (AREA)
- Signal Processing (AREA)
- Evolutionary Computation (AREA)
- Fuzzy Systems (AREA)
- Mathematical Physics (AREA)
- Measurement And Recording Of Electrical Phenomena And Electrical Characteristics Of The Living Body (AREA)
Abstract
The invention relates to an electrocardiosignal processing method and an electrocardiosignal processing system, in particular to the technical field of electrocardiosignal analysis. The method comprises the steps of carrying out heart rate variability analysis on the electrocardiosignals to obtain values of all characteristics in a characteristic set of the electrocardiosignals; calibrating and classifying the electrocardiosignals according to the characteristic values to obtain the heart states corresponding to the electrocardiosignals; processing the values of the features by sequentially adopting a reliefF algorithm and a multiple collinearity analysis algorithm to obtain an optimal feature subset; training the random forest model to obtain a weight model by taking the optimal feature subset of each electrocardiosignal in the sample set and the heart state corresponding to each electrocardiosignal as a training set; obtaining the weight value of each feature in the optimal feature subset according to the weight model; and determining the weighted sum of the weight value of each feature in the optimal feature subset and the value of each feature in the optimal feature subset as a feature index formula, wherein the feature index formula is used for calculating the feature index. The invention can improve the accuracy of the electrocardiosignal processing result.
Description
Technical Field
The invention relates to the technical field of electrocardiosignal analysis, in particular to an electrocardiosignal processing method and an electrocardiosignal processing system.
Background
The electrocardiosignal is one of the important physiological signals of human body. A large number of effective characteristics can be extracted from the electrocardiosignals and used for reflecting some diseases of human bodies. Heart Rate Variability (HRV) is based on the number of variations in the interaction of the body's sympathetic and vagal nerves that cause a corresponding change in Heart Rate. Heart rate variability, which reflects the activity of the autonomic nervous system, is assessed for the balance of tone between the sympathetic and vagal nerves of the heart, and may be judged to include: cardiovascular disease, fatigue, mood, and the like. Time-frequency domain analysis is always a common method for HRV analysis, and electrocardiosignals are extracted by extracting time-frequency domain characteristics. However, only analyzing the time domain and the frequency domain can not extract some nonlinear information of the electrocardiosignals.
Aiming at the problem of analysis of electrocardiosignals, a patent with the domestic application number of 201410852128.0 and the name of 'an electrocardiosignal processing method' provides an electrocardiosignal processing method based on a neural network. The method does not extract the characteristics of the signals, and adopts an end-to-end learning method to directly process the signals. This approach makes it difficult to see what the components in the model contribute to the final goal. In other words, the model becomes more black-boxed, which reduces the interpretability of the network. And the method is only suitable for the condition that the data volume of the electrocardiosignal is large. The patent with the application number of 201811429915.9 and the name of 'a method, a device and equipment for measuring heart rate variability based on time-frequency analysis' provides a method for extracting time-frequency domain characteristics of electrocardiosignals, and the method is suitable for various data volumes, but the method only carries out mathematical statistics analysis on characteristic parameters to construct a disease monitoring model based on characteristic parameter analysis, does not completely mine the characteristic parameters, and has low accuracy and poor actual test effect.
Disclosure of Invention
The invention aims to provide an electrocardiosignal processing method and an electrocardiosignal processing system, which can improve the accuracy of an electrocardiosignal processing result.
In order to achieve the purpose, the invention provides the following scheme:
an electrocardiosignal processing method, comprising:
obtaining a sample set, the sample set comprising a plurality of cardiac electrical signals;
for any electrocardiosignal in the sample set, carrying out heart rate variability analysis on the electrocardiosignal to obtain the value of each characteristic in the characteristic set of the electrocardiosignal; the characteristic set comprises a time domain index, a frequency domain index and a nonlinear domain index;
calibrating and classifying the electrocardiosignals according to the values of the characteristics in the characteristic set of the electrocardiosignals to obtain the heart states corresponding to the electrocardiosignals;
processing the values of all the characteristics in the characteristic set of the electrocardiosignals by sequentially adopting a reliefF algorithm and a multiple collinearity analysis algorithm to obtain an optimal characteristic subset of the electrocardiosignals; the optimal feature subset is a subset of the feature set;
training a random forest model by taking the optimal characteristic subset of each electrocardiosignal in the sample set and the heart state corresponding to each electrocardiosignal as a training set to obtain a weight model;
obtaining a weight value of each feature in the optimal feature subset of each electrocardiosignal according to the weight model;
and determining the weighted sum of the weight value of each characteristic in the optimal characteristic subset of each electrocardiosignal and the value of each characteristic in the optimal characteristic subset of each electrocardiosignal as a characteristic index formula, wherein the characteristic index formula is used for calculating a characteristic index.
Optionally, the analyzing the heart rate variability of the electrocardiographic signal to obtain values of each feature in the feature set of the electrocardiographic signal specifically includes:
performing signal preprocessing on the electrocardiosignals to obtain an electrocardio oscillogram of the electrocardiosignals;
obtaining an RR interval sequence of the electrocardiosignal according to an electrocardio waveform diagram of the electrocardiosignal;
adopting a3 sigma principle to remove the RR interval sequence of the electrocardiosignal to obtain a processed RR interval sequence;
and obtaining the value of each characteristic in the characteristic set of the electrocardiosignals according to the processed RR interphase sequence.
Optionally, the processing, by sequentially using a reliefF algorithm and a multiple collinearity analysis algorithm, values of each feature in the feature set of the electrocardiograph signal to obtain an optimal feature subset of the electrocardiograph signal specifically includes:
inputting the value of each feature in the feature set of the electrocardiosignals into a Relieff analysis model to obtain the selection weight of each feature in the feature set;
deleting the features with the weight smaller than a first set threshold value in the feature set to obtain a first feature set;
and processing the first feature set by adopting a multiple collinearity analysis algorithm to obtain an optimal feature subset.
Optionally, the processing the first feature set by using a multiple collinearity analysis algorithm to obtain an optimal feature subset specifically includes:
calculating mutual information between every two features in the first feature set;
and deleting the two features of which the mutual information is greater than a second set threshold value, and selecting the feature with small weight to obtain an optimal feature subset.
A cardiac electrical signal processing system comprising:
an obtaining module, configured to obtain a sample set, where the sample set includes a plurality of cardiac electrical signals;
the heart rate variability analysis module is used for carrying out heart rate variability analysis on any one electrocardiosignal in the sample set to obtain the value of each characteristic in the characteristic set of the electrocardiosignals; the feature set comprises a time domain index, a frequency domain index and a nonlinear domain index;
the calibration classification module is used for performing calibration classification on the electrocardiosignals according to values of all features in the feature set of the electrocardiosignals to obtain heart states corresponding to the electrocardiosignals;
the optimal characteristic subset determining module is used for processing the values of all characteristics in the characteristic set of the electrocardiosignals by sequentially adopting a reliefF algorithm and a multiple collinearity analysis algorithm to obtain an optimal characteristic subset of the electrocardiosignals; the optimal feature subset is a subset of the feature set;
the training module is used for training a random forest model to obtain a weight model by taking the optimal characteristic subset of each electrocardiosignal in the sample set and the heart state corresponding to each electrocardiosignal as a training set;
the weight value determining module is used for obtaining the weight value of each feature in the optimal feature subset of each electrocardiosignal according to the weight model;
and the characteristic index formula determining module is used for determining a weighted sum of a weight value of each characteristic in the optimal characteristic subset of each electrocardiosignal and a value of each characteristic in the optimal characteristic subset of each electrocardiosignal as a characteristic index formula, and the characteristic index formula is used for calculating a characteristic index.
Optionally, the heart rate variability analysis module specifically includes:
the preprocessing unit is used for preprocessing the electrocardiosignals to obtain an electrocardio oscillogram of the electrocardiosignals;
the RR interval sequence determining unit is used for obtaining an RR interval sequence of the electrocardiosignals according to an electrocardio waveform diagram of the electrocardiosignals;
the eliminating unit is used for eliminating the RR interval sequence of the electrocardiosignal by adopting a3 sigma principle to obtain a processed RR interval sequence;
and the characteristic value determining unit is used for obtaining the value of each characteristic in the characteristic set of the electrocardiosignals according to the processed RR interval sequence.
Optionally, the optimal feature subset determining module specifically includes:
a selection weight calculation unit, configured to input values of each feature in the feature set of the electrocardiographic signal into a ReliefF analysis model, so as to obtain a selection weight of each feature in the feature set;
the first feature set determining unit is used for deleting the features with the weight smaller than a first set threshold value selected from the feature sets to obtain a first feature set;
and the optimal feature subset determining unit is used for processing the first feature set by adopting a multiple collinearity analysis algorithm to obtain an optimal feature subset.
Optionally, the optimal feature subset determining unit specifically includes:
a mutual information calculating subunit, configured to calculate mutual information between every two features in the first feature set;
and the optimal feature subset determining subunit is used for deleting the features with small weight selected from the two features of which the mutual information is greater than the second set threshold value to obtain an optimal feature subset.
According to the specific embodiment provided by the invention, the invention discloses the following technical effects: the invention introduces a Relieff algorithm and a multiple collinearity analysis method to select the optimal characteristic subset, and can improve the accuracy of the electrocardiosignal processing result.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings needed to be used in the embodiments will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings without inventive exercise.
Fig. 1 is a flowchart of an electrocardiograph signal processing method according to an embodiment of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
In order to make the aforementioned objects, features and advantages of the present invention more comprehensible, the present invention is described in detail with reference to the accompanying drawings and the detailed description thereof.
As shown in fig. 1, an embodiment of the present invention provides an electrocardiographic signal processing method, including:
step 101: a sample set is obtained, the sample set comprising a plurality of cardiac electrical signals.
Step 102: for any electrocardiosignal in the sample set, performing Heart Rate Variability (HRV) analysis on the electrocardiosignal to obtain the values of each characteristic in a characteristic set D { f (1), f (2), \8230, f (37 } of the electrocardiosignal, wherein the characteristic set comprises 10 time domain indexes, 7 frequency domain indexes and 20 nonlinear domain indexes, and is specifically shown in Table 1.
TABLE 1 characteristic set table
Step 103: taking the basic value of the electrocardiosignals of the user into consideration, calibrating and classifying the electrocardiosignals according to the values of all the characteristics in the characteristic set of the electrocardiosignals by using a relative ratio method in combination with single-factor variance analysis to obtain the heart states corresponding to the electrocardiosignals, wherein the heart states can be some abnormal states related to the heart. Aiming at the characteristic of continuous change of electrocardiosignals of a human body, the electrocardiosignals are divided into a normal state or an abnormal state, for example, the state can be set to be normal, light fatigue, moderate fatigue and moderate fatigue during fatigue detection; the state can be set to "normal", "atrial fibrillation", etc. when atrial fibrillation is judged. Finally, a feature vector data set with state calibrations is formed, and the calibration is used as a label of the optimal feature subset and is used as a training sample together with the optimal feature subset.
Step 104: processing the values of the features in the feature set of the electrocardiosignals by sequentially adopting a reliefF algorithm and a multiple collinearity analysis algorithm to obtain an optimal feature subset D1{ f (1), f (2) \8230;, f (11) } of the electrocardiosignals; the optimal feature subset is a subset of the feature set.
Step 105: training a random forest model to obtain a weight model by taking the optimal characteristic subset of each electrocardiosignal in the sample set and the heart state corresponding to each electrocardiosignal as a training set, specifically, inputting the training sample into the random forest model, and assigning initial values to parameters of the random forest model; and selecting the optimal number L of decision trees by adopting a learning curve mode, and sampling and training by using a bootstarp algorithm.
Step 106: and obtaining the weight value of each feature in the optimal feature subset of each electrocardiosignal according to the weight model, directly analyzing the feature importance after random forest training, and directly outputting the contribution rate of each feature parameter to the model, namely the weight value of the feature.
Step 107: and determining a weighted sum of a weight value of each feature in the optimal feature subset of each electrocardiosignal and a value of each feature in the optimal feature subset of each electrocardiosignal as a feature index formula, wherein the feature index formula is used for calculating a feature index, the feature index contains all information of the electrocardiosignals, can be changed along with the change of the state, and can be subsequently used for judging diseases related to the heart.
The index formula is as follows:
indict = a1 means hr + a2 LFn + a3 ED2+ a4 ED1+ a5 p nn50+ a6 PI + a7 FE + a8 GI + a9 ED3+ a10 CV + a11 Rpf, where a1, a2, \ 8230 \ 8230;, a11 are the weighting coefficients in the model for these 11 characteristic indices.
As an optional implementation, the analyzing the heart rate variability of the cardiac electrical signal to obtain the value of each feature in the feature set of the cardiac electrical signal specifically includes:
and carrying out signal preprocessing on the electrocardiosignals to obtain an electrocardio oscillogram of the electrocardiosignals.
And obtaining an RR interval sequence RR { RR (1), RR (2),.., RR (n 1) } of the electrocardiosignal according to the electrocardiogram waveform diagram of the electrocardiosignal.
And (3) removing the RR interval sequence of the electrocardiosignal by adopting a3 sigma principle to obtain a processed RR interval sequence, and removing abnormal values by using the 3 sigma principle.
And obtaining the value of each characteristic in the characteristic set of the electrocardiosignals according to the processed RR interval sequence.
As an optional implementation manner, the obtaining values of each feature in the feature set of the electrocardiographic signal according to the processed RR interval sequence specifically includes:
carrying out statistic analysis on RR interval sequences RR { RR (1), RR (2) } to obtain a time domain index: meanHR, MEAN, SDNN, RMSSD, TINN, SDNNindex, pNN50, HRVindex, SDSD, CV.
Performing Fourier transform on the RR interval sequence RR { RR (1), RR (2), RR (n 1) } to convert the RR interval sequence into a frequency domain spectrogram, and calculating a frequency domain index: very Low Frequency (VLF), low Frequency (LF), high Frequency (HF), normalized low frequency power fraction (LFn), normalized high frequency power fraction (HFn), LFn/HFn and Total Power (TP).
Calculating non-linear domain indexes (semi-minor axis (SD 1) of poincare scattergram, semi-major axis (SD 2), ratio of semi-minor axis and semi-major axis (index), poincare scattergram area (S), vector Length Index (VLI), vector Angle Index (VAI), complex Correlation Measure (CCM), guzik Index (GI), porta Index (PI), ehler Index (EI), distribution entropy of four quadrants (ED 1, ED2, ED3, ED 4), positive feedback index (Rpf), negative feedback index (Rnf), total feedback index (Rtf), sample Entropy (SE), approximate Entropy (AE), fuzzy Entropy (FE):
a poincare scattergram is drawn according to RR interval sequences RR { RR (1), RR (2),.. And RR (n 1) }, and a scatter distribution map is drawn according to variation (first order difference) of two adjacent intervals of RR interval sequences RR { RR (1), RR (2),. And.rr (n 1) }, and the transformed scattergram can be divided into four quadrants, so that the poincare scattergram graphically shows the relevance between the adjacent points in the time sequence.
A semi-short axis (SD 1), a semi-long axis (SD 2), a ratio (index) of the semi-short axis and the semi-long axis, a Poincare scattergram area (S), a Vector Length Index (VLI), and a Vector Angle Index (VAI) of the Poincare scattergram are calculated according to the waveform characteristics of the scattergram.
Complex Correlation Measure (CCM): the complex correlation measure is a measure of the time variability in the window of formation of three consecutive points in the scatter plot. Aiming at a scatter diagram coordinate (Rn +1, rn +1Rn + 2), a sliding window method is adopted, the window size is 3, and the window stepping is 1. Wherein Rn is the current R peak point, rn +1 is the next adjacent R peak point, and Rn +2 is the next R peak point. The CCM measure is combined from the results of all overlapping scatter-sliding windows, and the formula is as follows.
Wherein τ represents the amount of delay of the poincare scattergram; n is the number of R peaks; c n Representing the area of the scatter plot fitted ellipse, i.e. C n SD1 SD2, where SD1 is the semiminor axis of the poincare scattergram and SD2 is the semimajor axis of the poincare scattergram; a (i) represents the area of a triangle formed by three consecutive points in the ith sliding window.
Guzik Index (GI), porta Index (PI), ehler Index (EI):
wherein the distance from the ith point to the isoline in the scatter diagram is D i ，
Represents the number of scatter points above the contour at a distance from the contour of
The number of scatter points below the contour line is represented; RR i Representing the ith RR interval.
Positive feedback index (Rpf), negative feedback index (Rnf), total feedback index (Rtf):
the positive feedback index represents the ratio of the number of scattered points to the total number of scattered points distributed in one quadrant and three quadrant, i.e. the change of RR interval is followed by the change of the same direction (the heart rate continuously increases or decreases).
The negative feedback index represents the ratio of the number of scattered points to the total number of scattered points distributed in the two-quadrant and the four-quadrant, i.e. the change of the RR interval follows the change in the opposite direction (the heart rate is accelerated and decelerated alternately).
The total feedback index is defined as positive feedback index/negative feedback index, reflecting the overall mechanism of the heart dynamic system.
And making N concentric circles on the Poincare scattergram by taking the origin as the center to form N annular subregions, wherein the annular subregions are divided according to four quadrants, namely each quadrant is divided into N fan-shaped subregions. Then, the number of scattered points in each divided sector area is counted, and the proportion of the scattered points to the total number of the scattered point areas is calculated, and the proportion values are respectively marked as p1, p2, \ 8230, pN is used for approximately describing the distribution probability of the scattered points in each area.
The distribution entropy of the four quadrants (distribution entropy ED1 of the first quadrant, distribution entropy ED2 of the second quadrant, distribution entropy ED3 of the third quadrant, distribution entropy ED4 of the fourth quadrant):
Sample Entropy (SE) definition:
when the RR interval sequence RR { RR (1), RR (2),.. The RR (n 1) } sequence is an m-dimensional vector, RR (i) and RR (j) are any two values in the sequence, and d [ RR (i), RR (j) ] is the maximum value of the distance between the two values.
Wherein the content of the first and second substances,
is d [ RR (i), RR (j)]The ratio of the number smaller than the threshold r to the total number of vectors; r (r)>0) Is a given threshold value; num { d [ RR (i), RR (j)]<r is d [ RR (i), RR (j)]A number less than r; n1-m is the total number of vectors;
wherein, B m (r) is
Of the average value of (a).
A m (r)＝B m+1 (r)
Wherein, A m (r) is the mean value B m The (r) dimension increases to a value of m + 1.
When N is finite, it can be estimated by:
approximate Entropy (AE) is used to describe the irregularities of complex systems.
When the RR interval sequence RR { RR (1), RR (2),.. Multidot.rr (N1) } sequence is an N-dimensional vector, m is defined as an integer of the vector, and r is a metric of 'similarity'.
An m-dimensional vector X (1), X (2), a.. Once, X (N-m + 1) is reconstructed, where X (i) = [ rr (i), rr (i + 1),. Once, rr (i + m-1) ].
For each value of i, the distance between the vector and the remaining vectors is calculated:
d [ X (i), X (j) ] = max | rr (i), rr (i + 1),.. R, rr (i + m-1) |, where i ranges from 0-m-1;
the number of d [ X (i), X (j) ] < r and the ratio of this number to the total number of vectors N-m +1 are counted for each i value according to a given threshold r (r > 0) and are recorded as:
wherein the content of the first and second substances,
is the average of all i. Firstly, the method is carried out
Taking the logarithm and then averaging the logarithm of all i.
The approximate entropy of this sequence is then:
wherein the content of the first and second substances,
is composed of
The dimension increases to a value of m + 1.
The Fuzzy Entropy (FE) is similar to the physical meanings of the approximate entropy and the sample entropy, the fuzzy entropy measures the probability of the new mode, and the larger the measurement value is, the larger the probability of the new mode is, namely, the sequence complexity is.
When the RR interval sequence RR { RR (1), RR (2) } is an N-dimensional vector.
Dividing RR interval sequences into k = N-m +1 sequences X (i) = [ RR (i), RR (i + 1),.. Multidot.rr (i + m-1) ], with m as a time window.
Defining the distance between the two time series X (i) and X (j) as:
and (3) introducing a fuzzy membership function, and calculating the similarity between the time sequences X (i) and X (j) by using the fuzzy function:
where r is the similarity tolerance, i, j =1, 2.
Defining a function:
then fuzzy entropy
As an optional implementation manner, the processing the values of the features in the feature set of the electrocardiographic signal by sequentially using a reliefF algorithm and a multiple collinearity analysis algorithm to obtain the optimal feature subset of the electrocardiographic signal specifically includes:
and inputting the value of each feature in the feature set of the electrocardiosignals into a Relieff analysis model to obtain the selection weight of each feature in the feature set.
And (3) selecting the features with the weight smaller than a first set threshold (0.2) in the feature set, and deleting the features to obtain a first feature set.
And processing the first feature set by adopting a multiple collinearity analysis algorithm to obtain an optimal feature subset.
As an optional implementation manner, the processing the first feature set by using the multiple collinearity analysis algorithm to obtain an optimal feature subset specifically includes:
and calculating mutual information between every two characteristics in the first characteristic set.
And deleting the two features of which the mutual information is greater than a second set threshold (0.8) and selecting the feature with small weight to obtain an optimal feature subset. Because the two features are considered to have multiple collinearity when the mutual information is greater than a second set threshold, one of the features with a larger weight is retained, and the other feature is deleted.
The essence of the Random Forest (RF) model is a bagging integration algorithm, which averages the predicted results of the base evaluator or determines the results of the integrated evaluator using majority voting.
A bagging method: a part of data is randomly selected from the data set to be used as a training set, so that each tree is trained by using different training sets, and the generated classification decision trees are different, namely the bagging method. The bagging method uses a random sampling technique with a put-back to form each different training set.
As an optional implementation, the specific method of the random forest model is as follows:
assuming that the RF model has L decision trees, a Classification And Regression Tree algorithm (CART) is used to construct a corresponding Classification Regression decision Tree, and each decision Tree is trained by using a boottrap method. And after training is finished, inputting a test set, carrying out classification judgment on the test set by each decision tree, and finally outputting a result depending on voting of a plurality of decision trees.
S0010, inputting the training samples into a random forest model, assigning initial values to model parameters by a Grid Search Grid searching method, and presetting training sample train and training times N.
S0011, traversing the number of decision trees in the random forest from 1-200, drawing a learning curve of training samples, finally obtaining 200 learning rates, selecting the number L of the decision trees with the highest learning rate, then sampling training data by using a Bootstrap algorithm, randomly generating L training sets, and selecting x prediction samples in each training set for verification training.
To the foregoing method, an embodiment of the present invention provides an electrocardiographic signal processing system, including:
the acquisition module is used for acquiring a sample set, wherein the sample set comprises a plurality of electrocardiosignals.
The heart rate variability analysis module is used for carrying out heart rate variability analysis on any one electrocardiosignal in the sample set to obtain the value of each characteristic in the characteristic set of the electrocardiosignals; the feature set includes a time domain index, a frequency domain index, and a nonlinear domain index.
And the calibration and classification module is used for performing calibration and classification on the electrocardiosignals according to the values of the features in the feature set of the electrocardiosignals to obtain the heart states corresponding to the electrocardiosignals.
The optimal characteristic subset determining module is used for processing the values of all characteristics in the characteristic set of the electrocardiosignals by sequentially adopting a reliefF algorithm and a multiple collinearity analysis algorithm to obtain an optimal characteristic subset of the electrocardiosignals; the optimal feature subset is a subset of the feature set.
The training module is used for training a random forest model to obtain a weight model by taking the optimal characteristic subset of each electrocardiosignal in the sample set and the heart state corresponding to each electrocardiosignal as a training set;
and the weight value determining module is used for obtaining the weight value of each characteristic in the optimal characteristic subset of each electrocardiosignal according to the weight model.
And the characteristic index formula determining module is used for determining the weighted sum of the weight value of each characteristic in the optimal characteristic subset of each electrocardiosignal and the value of each characteristic in the optimal characteristic subset of each electrocardiosignal as a characteristic index formula, and the characteristic index formula is used for calculating a characteristic index.
As an optional implementation, the heart rate variability analysis module specifically includes:
and the preprocessing unit is used for preprocessing the electrocardiosignals to obtain an electrocardio oscillogram of the electrocardiosignals.
And the RR interval sequence determining unit is used for obtaining the RR interval sequence of the electrocardiosignal according to the electrocardio waveform diagram of the electrocardiosignal.
And the removing unit is used for removing the RR interval sequence of the electrocardiosignal by adopting a3 sigma principle to obtain the processed RR interval sequence.
And the characteristic value determining unit is used for obtaining the value of each characteristic in the characteristic set of the electrocardiosignals according to the processed RR interval sequence.
As an optional implementation manner, the optimal feature subset determining module specifically includes:
and the selection weight calculation unit is used for inputting the values of all the characteristics in the feature set of the electrocardiosignals into a Relieff analysis model to obtain the selection weights of all the characteristics in the feature set.
And the first feature set determining unit is used for deleting the features with the weight smaller than a first set threshold value selected from the feature sets to obtain a first feature set.
And the optimal feature subset determining unit is used for processing the first feature set by adopting a multiple collinearity analysis algorithm to obtain an optimal feature subset.
As an optional implementation manner, the optimal feature subset determining unit specifically includes:
and the mutual information calculating subunit is used for calculating mutual information between every two characteristics in the first characteristic set.
And the optimal feature subset determining subunit is used for deleting the features with small weight selected from the two features of which the mutual information is greater than the second set threshold value to obtain an optimal feature subset.
The invention has the following beneficial effects:
(1) The electrocardio signal is a common physiological signal, and the electrocardio waveform processing method can effectively preprocess the original waveform signal to obtain the high-quality electrocardio waveform data.
(2) The method can enable the obtained characteristic indexes to more effectively reflect the real characteristics and meanings of the electrocardiographic waveform by using the time domain indexes, the frequency domain indexes and the nonlinear domain indexes.
(3) And establishing a classification model by combining machine learning, calculating a characteristic index according to the weight of the characteristic, wherein the index can be used as a basis for judging the heart-related diseases subsequently.
(4) The optimal subset is selected by introducing a Relieff algorithm and a multiple collinearity analysis method, the processing efficiency and accuracy are improved, and finally a characteristic index indict obtained by weighting the characteristics in the optimal characteristic subset according to the weight in the model is obtained.
In the present specification, the embodiments are described in a progressive manner, each embodiment focuses on differences from other embodiments, and the same and similar parts among the embodiments are referred to each other. For the system disclosed by the embodiment, the description is relatively simple because the system corresponds to the method disclosed by the embodiment, and the relevant points can be referred to the description of the method part.
The principle and the embodiment of the present invention are explained by applying specific examples, and the above description of the embodiments is only used to help understanding the method and the core idea of the present invention; meanwhile, for a person skilled in the art, according to the idea of the present invention, the specific embodiments and the application range may be changed. In view of the above, the present disclosure should not be construed as limiting the invention.
Claims (8)
1. An electrocardiographic signal processing method, comprising:
obtaining a sample set, the sample set comprising a plurality of cardiac electrical signals;
for any electrocardiosignal in the sample set, carrying out heart rate variability analysis on the electrocardiosignal to obtain the value of each characteristic in the characteristic set of the electrocardiosignal; the feature set comprises a time domain index, a frequency domain index and a nonlinear domain index;
calibrating and classifying the electrocardiosignals according to the values of the characteristics in the characteristic set of the electrocardiosignals to obtain the heart states corresponding to the electrocardiosignals;
processing the values of all the characteristics in the characteristic set of the electrocardiosignals by sequentially adopting a reliefF algorithm and a multiple collinearity analysis algorithm to obtain an optimal characteristic subset of the electrocardiosignals; the optimal feature subset is a subset of the feature set;
training a random forest model by taking the optimal characteristic subset of each electrocardiosignal in the sample set and the heart state corresponding to each electrocardiosignal as a training set to obtain a weight model;
obtaining a weight value of each feature in the optimal feature subset of each electrocardiosignal according to the weight model;
and determining the weighted sum of the weight value of each characteristic in the optimal characteristic subset of each electrocardiosignal and the value of each characteristic in the optimal characteristic subset of each electrocardiosignal as a characteristic index formula, wherein the characteristic index formula is used for calculating a characteristic index.
2. The method according to claim 1, wherein the step of performing heart rate variability analysis on the ecg signal to obtain values of features in the feature set of the ecg signal comprises:
performing signal preprocessing on the electrocardiosignals to obtain an electrocardio oscillogram of the electrocardiosignals;
obtaining an RR interval sequence of the electrocardiosignals according to an electrocardio waveform diagram of the electrocardiosignals;
removing the RR interval sequence of the electrocardiosignal by adopting a3 sigma principle to obtain a processed RR interval sequence;
and obtaining the value of each characteristic in the characteristic set of the electrocardiosignals according to the processed RR interval sequence.
3. The method according to claim 1, wherein the processing of the values of the features in the feature set of the electrocardiographic signal by sequentially using a reliefF algorithm and a multiple collinearity analysis algorithm to obtain the optimal feature subset of the electrocardiographic signal specifically comprises:
inputting the value of each feature in the feature set of the electrocardiosignals into a Relieff analysis model to obtain the selection weight of each feature in the feature set;
deleting the features with the weight smaller than a first set threshold value in the feature set to obtain a first feature set;
and processing the first feature set by adopting a multiple collinearity analysis algorithm to obtain an optimal feature subset.
4. The method according to claim 3, wherein the processing the first feature set by using a multiple collinearity analysis algorithm to obtain an optimal feature subset specifically comprises:
calculating mutual information between every two features in the first feature set;
and deleting the two characteristics of which the mutual information is greater than a second set threshold value, and selecting the characteristics with small weight to obtain an optimal characteristic subset.
5. A cardiac signal processing system, comprising:
an obtaining module, configured to obtain a sample set, where the sample set includes a plurality of electrocardiographic signals;
the heart rate variability analysis module is used for carrying out heart rate variability analysis on any one electrocardiosignal in the sample set to obtain the value of each characteristic in the characteristic set of the electrocardiosignals; the feature set comprises a time domain index, a frequency domain index and a nonlinear domain index;
the calibration and classification module is used for performing calibration and classification on the electrocardiosignals according to values of all characteristics in the characteristic set of the electrocardiosignals to obtain the heart states corresponding to the electrocardiosignals;
the optimal characteristic subset determining module is used for processing the values of all characteristics in the characteristic set of the electrocardiosignals by sequentially adopting a reliefF algorithm and a multiple collinearity analysis algorithm to obtain an optimal characteristic subset of the electrocardiosignals; the optimal feature subset is a subset of the feature set;
the training module is used for training a random forest model to obtain a weight model by taking the optimal characteristic subset of each electrocardiosignal in the sample set and the heart state corresponding to each electrocardiosignal as a training set;
the weight value determining module is used for obtaining the weight value of each feature in the optimal feature subset of each electrocardiosignal according to the weight model;
and the characteristic index formula determining module is used for determining the weighted sum of the weight value of each characteristic in the optimal characteristic subset of each electrocardiosignal and the value of each characteristic in the optimal characteristic subset of each electrocardiosignal as a characteristic index formula, and the characteristic index formula is used for calculating a characteristic index.
6. The system of claim 5, wherein the heart rate variability analysis module comprises:
the preprocessing unit is used for preprocessing the electrocardiosignals to obtain an electrocardio oscillogram of the electrocardiosignals;
the RR interval sequence determining unit is used for obtaining an RR interval sequence of the electrocardiosignals according to an electrocardio waveform diagram of the electrocardiosignals;
the removing unit is used for removing the RR interval sequence of the electrocardiosignal by adopting a3 sigma principle to obtain a processed RR interval sequence;
and the characteristic value determining unit is used for obtaining the value of each characteristic in the characteristic set of the electrocardiosignals according to the processed RR interval sequence.
7. The system of claim 5, wherein the optimal feature subset determining module specifically comprises:
a selection weight calculation unit, configured to input values of each feature in the feature set of the electrocardiographic signal into a ReliefF analysis model, so as to obtain a selection weight of each feature in the feature set;
the first feature set determining unit is used for deleting the features with the weight smaller than a first set threshold value selected from the feature set to obtain a first feature set;
and the optimal feature subset determining unit is used for processing the first feature set by adopting a multiple collinearity analysis algorithm to obtain an optimal feature subset.
8. The system according to claim 7, wherein the optimal feature subset determining unit specifically includes:
a mutual information calculating subunit, configured to calculate mutual information between every two features in the first feature set;
and the optimal feature subset determining subunit is used for deleting the features with small weight selected from the two features of which the mutual information is greater than the second set threshold value to obtain an optimal feature subset.
Priority Applications (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211135001.8A CN115363598B (en) | 2022-09-19 | 2022-09-19 | Electrocardiosignal processing method and system |
| GB2307060.0A GB2622458A (en) | 2022-09-19 | 2023-05-12 | Electrocardiographic (ECG) signal processing method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211135001.8A CN115363598B (en) | 2022-09-19 | 2022-09-19 | Electrocardiosignal processing method and system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115363598A true CN115363598A (en) | 2022-11-22 |
| CN115363598B CN115363598B (en) | 2023-03-24 |
Family
ID=84072424
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211135001.8A Active CN115363598B (en) | 2022-09-19 | 2022-09-19 | Electrocardiosignal processing method and system |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN115363598B (en) |
| GB (1) | GB2622458A (en) |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106377247A (en) * | 2016-09-10 | 2017-02-08 | 天津大学 | Feature selection-based arrhythmia classification method |
| WO2018120088A1 (en) * | 2016-12-30 | 2018-07-05 | 中国科学院深圳先进技术研究院 | Method and apparatus for generating emotional recognition model |
| CN110876626A (en) * | 2019-11-22 | 2020-03-13 | 兰州大学 | Depression detection system based on optimal lead selection of multi-lead electroencephalogram |
| CN113180596A (en) * | 2021-04-07 | 2021-07-30 | 中山大学 | Non-contact sleep analysis method and device and storage medium |
| CN114569096A (en) * | 2022-02-28 | 2022-06-03 | 西安交通大学 | Non-contact continuous blood pressure measuring method and system based on video stream |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111067508B (en) * | 2019-12-31 | 2022-09-27 | 深圳安视睿信息技术股份有限公司 | Non-intervention monitoring and evaluating method for hypertension in non-clinical environment |
-
2022
- 2022-09-19 CN CN202211135001.8A patent/CN115363598B/en active Active
-
2023
- 2023-05-12 GB GB2307060.0A patent/GB2622458A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106377247A (en) * | 2016-09-10 | 2017-02-08 | 天津大学 | Feature selection-based arrhythmia classification method |
| WO2018120088A1 (en) * | 2016-12-30 | 2018-07-05 | 中国科学院深圳先进技术研究院 | Method and apparatus for generating emotional recognition model |
| CN110876626A (en) * | 2019-11-22 | 2020-03-13 | 兰州大学 | Depression detection system based on optimal lead selection of multi-lead electroencephalogram |
| CN113180596A (en) * | 2021-04-07 | 2021-07-30 | 中山大学 | Non-contact sleep analysis method and device and storage medium |
| CN114569096A (en) * | 2022-02-28 | 2022-06-03 | 西安交通大学 | Non-contact continuous blood pressure measuring method and system based on video stream |
Non-Patent Citations (1)
| Title |
|---|
| 潘礼正，尹泽明，佘世刚，袁峥峥，赵路: "基于FCA-ReliefF的融合生理信号情绪识别研究", 《计算机测量与控制》 * |
Also Published As
| Publication number | Publication date |
|---|---|
| GB2622458A (en) | 2024-03-20 |
| GB202307060D0 (en) | 2023-06-28 |
| CN115363598B (en) | 2023-03-24 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107951485B (en) | Ambulatory ECG analysis method and apparatus based on artificial intelligence self study | |
| CN108714026B (en) | Fine-grained electrocardiosignal classification method based on deep convolutional neural network and online decision fusion | |
| CN107837082B (en) | Automatic electrocardiogram analysis method and device based on artificial intelligence self-learning | |
| CN110693489B (en) | Myocardial infarction detection method based on multi-classifier reinforcement learning | |
| CN112656427A (en) | Electroencephalogram emotion recognition method based on dimension model | |
| WO2019161610A1 (en) | Electrocardiogram information processing method and electrocardiogram workstation system | |
| CN110598676B (en) | Deep learning gesture electromyographic signal identification method based on confidence score model | |
| CN111110224A (en) | Electrocardiogram classification method and device based on multi-angle feature extraction | |
| CN110680313A (en) | Epileptic period classification method based on pulse group intelligent algorithm and combined with STFT-PSD and PCA | |
| CN112754502A (en) | Automatic music switching method based on electroencephalogram signals | |
| CN112617851A (en) | Mental load classification method and system based on electrocardiosignals | |
| CN111956210B (en) | Electrocardiosignal R wave identification method based on EEMD and signal structure analysis | |
| CN116503673B (en) | Arrhythmia identification and detection method and system based on electrocardiogram | |
| Bozkurt et al. | Comparison of AR parametric methods with subspace-based methods for EMG signal classification using stand-alone and merged neural network models | |
| Karimifard et al. | A robust method for diagnosis of morphological arrhythmias based on Hermitian model of higher-order statistics | |
| CN115486849A (en) | Electrocardiosignal quality evaluation method, device and equipment | |
| Azami et al. | Automatic signal segmentation based on singular spectrum analysis and imperialist competitive algorithm | |
| Kumari et al. | Optimization of multi-layer perceptron neural network using genetic algorithm for arrhythmia classification | |
| CN115363598B (en) | Electrocardiosignal processing method and system | |
| Kumari et al. | Classification of cardiac arrhythmia using hybrid genetic algorithm optimisation for multi-layer perceptron neural network | |
| Vladimir et al. | Genetic programming application for features selection in task of arterial hypertension classification | |
| Saeed et al. | A Slantlet based statistical features extraction for classification of normal, arrhythmia, and congestive heart failure in electrocardiogram | |
| Aligholipour et al. | Silent paroxysmal atrial fibrillation detection by neural networks based on ECG records | |
| Kumar | Heart disease detection using radial basis function classifier | |
| Yaman et al. | EMG signal classification using Decision trees and neural networks |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |