CN113974647B - System and method for reconstructing sudden cardiac death risk factor - Google Patents
System and method for reconstructing sudden cardiac death risk factor Download PDFInfo
- Publication number
- CN113974647B CN113974647B CN202111251594.XA CN202111251594A CN113974647B CN 113974647 B CN113974647 B CN 113974647B CN 202111251594 A CN202111251594 A CN 202111251594A CN 113974647 B CN113974647 B CN 113974647B
- Authority
- CN
- China
- Prior art keywords
- sudden cardiac
- cardiac death
- sudden
- model
- wave
- 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.)
- Active
Links
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
-
- 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
- A61B5/24—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
- A61B5/355—Detecting T-waves
-
- 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
- A61B5/366—Detecting abnormal QRS complex, e.g. widening
-
- 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
- A61B5/7203—Signal processing specially adapted for physiological signals or for diagnostic purposes for noise prevention, reduction or removal
-
- 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
- A61B5/725—Details of waveform analysis using specific filters therefor, e.g. Kalman or adaptive filters
-
- 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
- A61B5/7253—Details of waveform analysis characterised by using transforms
- A61B5/726—Details of waveform analysis characterised by using transforms using Wavelet transforms
-
- 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
-
- 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
-
- 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
- A61B5/7271—Specific aspects of physiological measurement 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
- A61B5/7271—Specific aspects of physiological measurement analysis
- A61B5/7275—Determining trends in physiological measurement data; Predicting development of a medical condition based on physiological measurements, e.g. determining a risk factor
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T10/00—Road transport of goods or passengers
- Y02T10/10—Internal combustion engine [ICE] based vehicles
- Y02T10/40—Engine management systems
Landscapes
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Animal Behavior & Ethology (AREA)
- Veterinary Medicine (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Heart & Thoracic Surgery (AREA)
- Medical Informatics (AREA)
- Molecular Biology (AREA)
- Surgery (AREA)
- Cardiology (AREA)
- General Health & Medical Sciences (AREA)
- Public Health (AREA)
- Pathology (AREA)
- Artificial Intelligence (AREA)
- Signal Processing (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physiology (AREA)
- Psychiatry (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 provides a system and a method for reconstructing sudden cardiac death risk factors by multi-feature quantization and model parameter optimization of a nonlinear support vector machine, wherein the system comprises the steps of preprocessing data of sudden cardiac death electrocardiosignal data sets and normal sinus rhythm electrocardiosignal data sets; carrying out electrocardiographic waveform detection on the processed electrocardiographic data set; extracting sudden cardiac death risk factors; performing characteristic quantization scaling treatment on the extracted initial characteristics; using a nonlinear support vector machine as a verification model of the sudden cardiac death risk factor, and determining an error punishment parameter C and a nuclear parameter gamma through model parameter optimization; obtaining a prediction model of sudden cardiac death through the formulated sudden cardiac death risk factors and the optimized model parameters; the effects of reconstructing and verifying the sudden cardiac death risk factors are achieved, and the method has good guiding significance for researching sudden cardiac death.
Description
Technical Field
The invention belongs to the technical field of machine learning and electrocardiosignal splitting processing, and particularly relates to a system and a method for reconstructing a sudden cardiac death risk factor by multi-feature quantization and model parameter optimization of a nonlinear support vector machine.
Background
Sudden cardiac death ((Sudden Cardiac Death, SCD)) refers to unpredictable sudden death that occurs immediately or within 24 hours. Sudden cardiac death patients can feel obvious symptoms such as dyspnea, chest pain, palpitation and the like in a common morbidity period of about 1 hour, and then cardiac arrest (Sudden Cardiac Arrest, SCA) occurs, and if the sudden cardiac death patients cannot be effectively rescued within a few minutes, death can be caused. Because sudden cardiac death is sudden, even if people with no history or mild symptoms of cardiovascular diseases at ordinary times, the sudden cardiac death is induced by bad life style such as smoking, drinking and the like and dangerous causes such as obesity, hypertension, diabetes and the like, and the subconscious loss and sudden cardiac death are caused. Medical imaging and electrocardiography are the main means of risk assessment for sudden cardiac death in clinic. Intrinsic activity and change of the heart can be intuitively observed through medical images, but heart information cannot be flexibly acquired in real time due to the limitation of observation time and equipment, and an electrocardiogram is widely applied to heart health monitoring by noninvasive real-time convenience.
The current search for the cause of sudden cardiac death is mainly from a medical point of view, and specific wave bands in an electrocardiogram are observed in a targeted manner based on physiology and anatomy, but the limitations of the above methods are limited by human knowledge, so that we sometimes do not pay attention to some hidden pathogenic factors. Under the background of the combination of the current doctors and the industry, how to utilize the powerful data processing capability of a computer and the related data processing method in the artificial intelligence field to discover potential hidden factors of sudden cardiac death or discover the influence of each factor on sudden cardiac death, thereby improving the prediction capability of sudden cardiac death is a problem to be solved urgently.
Although sudden cardiac death factors are found to be insufficient for definitive diagnosis or treatment of sudden cardiac death, screening thereof has significant reference value for relevant studies.
Disclosure of Invention
Aiming at the defects and shortcomings in the prior art, the invention aims to provide a system and a method for reconstructing sudden cardiac death risk factors by multi-feature quantization and model parameter optimization of a nonlinear support vector machine, wherein the system comprises the steps of carrying out data preprocessing on a sudden cardiac death electrocardiosignal data set and a normal sinus rhythm electrocardiosignal data set; carrying out electrocardiographic waveform detection on the processed electrocardiographic data set; extracting sudden cardiac death risk factors; performing characteristic quantization scaling treatment on the extracted initial characteristics; using a nonlinear support vector machine as a verification model of the sudden cardiac death risk factor, and determining an error punishment parameter C and a nuclear parameter gamma through model parameter optimization; obtaining a prediction model of sudden cardiac death through the formulated sudden cardiac death risk factors and the optimized model parameters; the effects of reconstructing and verifying the sudden cardiac death risk factors are achieved, and the method has good guiding significance for researching sudden cardiac death.
The invention can adopt the following technical scheme:
a system for reconstructing sudden cardiac death risk factors by multi-feature quantization and model parameter optimization of a nonlinear support vector machine is characterized by comprising:
the preprocessing module is used for preprocessing data of the cardiac sudden death electrocardiosignal data set and the normal sinus rhythm electrocardiosignal data set;
the waveform detection module is used for carrying out electrocardiographic waveform detection on the preprocessed electrocardiographic data set and obtaining electrocardiographic waveforms for sudden death risk factor extraction;
the initial characteristic acquisition module is used for extracting the sudden cardiac death risk factors to be confirmed according to the electrocardio waveforms to obtain electrocardio initial characteristics;
the feature set construction module is used for processing the electrocardio initial features to obtain a training set;
the characteristic optimizing module is used for obtaining a prediction model of sudden cardiac death according to the training set through the formulated sudden cardiac death risk factors and the optimized model parameters, and verifying the recombined sudden cardiac death risk factors through model feedback; and taking the nonlinear support vector machine as a verification model of the sudden cardiac death risk factor, carrying out model parameter optimization based on a Gaussian kernel function, and determining an error penalty parameter C and a kernel parameter gamma.
Further, the preprocessing module includes: the device comprises a data cutting sub-module, a frequency band decomposing sub-module, a noise filtering sub-module and a signal reconstructing sub-module.
Further, the waveform detection module detects the QRS complex, compares the waveform integration area size under a sliding window with a fixed size in a retrieval interval based on the RR interval, and detects the tail end and the peak of the T wave.
Further, the initial feature acquisition module pair includes: the potential factors of the electrocardiosignal sequence RR interval, the QRS time limit, the QTc interval, the T wave crest interval, the Tp-Te/QT index, the T wave amplitude and the heart rhythm variability are extracted.
Further, the feature set construction module calculates an average value, a standard deviation and an approximate entropy of the initial feature for constructing a training set.
Further, in the feature set construction module, an operation of performing feature scaling processing on the initial feature is further included after the training set is constructed: firstly, determining the difference of the dimension between the features, and then selecting the minimum maximum value scaling or standard scaling to process each feature to obtain a feature value with smaller difference in the same dimension data.
Further, the characteristic optimizing module uses an improved grid searching method and a quantum particle swarm optimization algorithm as parameter optimizing algorithms respectively, compares the influence of the two parameter optimizing algorithms on the performance of the sudden cardiac death prediction model, and determines an error punishment parameter C and a nuclear parameter gamma; the sensitivity Se, the specificity Sp, the positive prediction rate Ppv and the accuracy Acc are used as indexes of model evaluation.
Further, classifying the electrocardiographic data of 20-30 min, 30-40 min, 40-50 min and 60-70 min before the sudden cardiac death occurs as an electrocardiographic signal data set of sudden cardiac death; the characteristic optimizing module outputs four models respectively corresponding to four time periods.
The method for reconstructing the sudden cardiac death risk factor by multi-feature quantization and model parameter optimization of the nonlinear support vector machine is characterized by comprising the following steps:
step S1: performing data preprocessing on an electrocardiosignal data set of sudden cardiac death and a normal sinus rhythm electrocardiosignal data set;
step S2: carrying out electrocardiographic waveform detection on the electrocardiographic data set processed in the step S1, and obtaining electrocardiographic waveforms for sudden death risk factor extraction; detecting a QRS wave group based on an R wave crest detection method for constructing a self-adaptive threshold and a QRS wave group detection algorithm of a multi-decision rule, comparing the size of a waveform integration area under a sliding window with a fixed size in a retrieval interval based on RR intervals, and detecting the tail end of a T wave and the peak of the T wave;
step S3: extracting cardiac sudden death risk factors to be confirmed to obtain electrocardiographic initial features, and constructing a feature set suitable for model training for respective average value, standard deviation and approximate entropy of the initial features;
step S4: performing feature scaling processing on the initial features extracted in the step S3;
step S5: taking a nonlinear support vector machine as a verification model of the sudden cardiac death risk factor, carrying out model parameter optimization based on a Gaussian kernel function, and determining an error penalty parameter C and a kernel parameter gamma;
step S6: and obtaining a prediction model of sudden cardiac death through the formulated sudden cardiac death risk factors and the optimized model parameters, and verifying the recombined sudden cardiac death risk factors through model feedback.
Further, the step S1 specifically includes:
step S11: classifying cardiac sudden death electrocardio data according to four time periods of 20-30 min, 30-40 min, 40-50 min and 60-70 min before the occurrence of a cardiac sudden death event, and then cutting cardiac sudden death signals of each time period into data with the time length of 1 min;
step S12: for normal sinus rhythm data, randomly taking a length of 10min for each note, and cutting the note into a plurality of lengths of 1 min;
step S13: processing the electrocardiosignals by using discrete wavelet transformation and a Mallat algorithm to obtain multi-layer electrocardiosignal decomposition signals with different frequency bands;
step S14: performing threshold processing on the decomposed signal by using a threshold value and a threshold function, and filtering electrocardiosignal signals;
step S15: reconstructing the multi-layer decomposition signals with noise filtered to obtain filtered electrocardiosignals;
the step S2 specifically comprises the following steps:
step S21: passing the filtered electrocardiosignal through a low-pass filter and a high-pass filter, wherein the two filters form a band-pass filter;
step S22: calculating the point-by-point of each sample point, taking the absolute value, amplifying the slope information of the R wave, and finally carrying out window integration on the signal;
step S23: correcting the R wave determined by the R wave detection rule, and determining the positions of the Q wave crest and the S wave crest based on the RR interval;
step S24: integrating the waveform in the searching interval of the T wave end by using a sliding window with fixed length, comparing the integration results to determine the T wave end, and then searching forward according to the T wave end to determine the position of the T wave crest value;
in step S4, determining the difference of the dimension between the features, and processing each feature by selecting minimum maximum value scaling or standard scaling to obtain a feature value with smaller difference in the same dimension data;
the step S5 specifically comprises the following steps:
step S51: dividing four data sets corresponding to four time periods into a training set and a testing set according to a certain proportion, and sending the training set into a support vector machine;
step S52: the improved grid search method and the quantum particle swarm optimization algorithm are used as parameter optimizing algorithms, the influence of the two parameter optimizing algorithms on the performance of the sudden cardiac death prediction model is compared, and an error punishment parameter C and a nuclear parameter gamma are determined;
in step S6, four models are output, corresponding to four time periods, respectively, and the sensitivity Se, the specificity Sp, the positive predictive rate Ppv, and the accuracy Acc are used as indexes for model evaluation.
Compared with the prior art, the method and the device for predicting the cardiac sudden death feature of the heart by using the characteristic set have the advantages that the difference between the normal sinus rhythm and the cardiac signal before the cardiac sudden death is quantified, the characteristic set is used for representing the prediction information implied by the cardiac signal which looks normal before the cardiac sudden death, a methodology and a specific machine learning construction scheme are provided for exploring the cardiac sudden death risk factor, and the neglect of the artificial subjective factors of the cardiac sudden death risk factor possibly existing due to the limitation of the prior knowledge is made up. Under the condition that the prediction accuracy is basically ensured to be higher, the prediction time is 20 minutes before sudden cardiac death occurs, the method automatically establishes and optimizes the model, improves the prediction time to 70 minutes before sudden cardiac death occurs on the basis of reconstructing and verifying the sudden death risk factors, and ensures the reliability of the extracted sudden death risk factors on the basis of model training.
Drawings
The invention is described in further detail below with reference to the attached drawings and detailed description:
FIG. 1 is a schematic diagram of the system architecture and workflow of an embodiment of the present invention.
FIG. 2 is a schematic diagram of a model parameter optimization training process according to an embodiment of the present invention.
Detailed Description
In order to make the features and advantages of the present invention more comprehensible, embodiments accompanied with figures are described in detail below:
it should be noted that the following detailed description is exemplary and is intended to provide further explanation of the present application. Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. It is noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of example embodiments in accordance with the present application. As used herein, the singular is also intended to include the plural unless the context clearly indicates otherwise, and furthermore, it is to be understood that the terms "comprises" and/or "comprising" when used in this specification are taken to specify the presence of stated features, steps, operations, devices, components, and/or combinations thereof.
As shown in fig. 1, the scheme for reconstructing sudden cardiac death risk factors by multi-feature quantization and model parameter optimization of the nonlinear support vector machine provided by the invention mainly comprises the following steps: data preprocessing, waveform detection, initial feature extraction, feature set construction, feature scaling, model parameter optimizing and optimizing model, final verification effect feedback reconstruction risk factors and the like;
the data preprocessing specifically comprises the following steps:
electrocardiogram data are acquired from an sudden cardiac death dynamic electrocardiogram database (Sudden Cardiac Death Holter Database) and an MIT-BIH normal sinus rhythm database (MIT-BIH Normal Sinus Rhythm Database), wherein the sudden cardiac death dynamic electrocardiogram database has 23 complete Holter records with the number of 30-52. Each record contains 2 leads and has a sampling rate of 250Hz and the duration of the record varies from several hours to more than 20 hours. The time at which the cardiac arrest event occurred is not noted due to the electrocardiographic records numbered 40, 42, 49; the recorded 35 electrocardiosignals have serious noise interference, and the poor waveform detection effect affects the model prediction performance; the electrocardiogram recorded 38 before sudden cardiac death occurred shows ventricular tachycardia for a long period of time, has obvious waveform anomalies and has little predictive value, so the embodiment does not use the 5-stroke electrocardiogram recorded above. The MIT-BIH normal sinus rhythm database has 18 records, each record also contains 2 leads, the sampling rate is 128Hz, and the duration of the records is about 24 hours.
Classifying the sudden cardiac death electrocardio data according to four time periods of 20-30 min, 30-40 min, 40-50 min and 60-70 min before the sudden cardiac death event. The electrocardiographic data of each time period is then cut into 1min lengths, and samples with obvious ventricular arrhythmias or severe noise interference are removed. For normal sinus rhythm data, a length of 10min was randomly taken per note and cut into several lengths of 1 min.
In the embodiment, based on discrete wavelet transform filtering, a db8 wavelet function is selected according to morphological characteristics of an electrocardiosignal, and after 8 decomposition layers are selected according to sampling frequency of an original signal and a noise distribution frequency band, the original electrocardiosignal is subjected to wavelet decomposition by using a Mallat algorithm. And selecting a threshold value and a threshold function to perform threshold processing on the frequency band to be filtered after decomposition, and finally performing wavelet inverse decomposition and reconstructing and recovering the filtered electrocardiosignal.
The waveform detection specifically comprises:
the QRS wave group and the T wave which are commonly concerned by sudden cardiac death at present are extracted, and potential factors such as J wave, P wave and the like are not ignored, so that the possibility that waveforms which are not commonly concerned are taken as waveform detection sources exists.
The QRS complex detection algorithm firstly carries out pretreatment on the electrocardiosignals subjected to the discrete wavelet transformation denoising, highlights the amplitude of R waves and reduces the amplitudes of other waveforms. And then determining the position of the R wave crest according to the R wave detection rule, and then adjusting and correcting the position of the R wave crest in a range. And finally, determining self-adaptive Q wave and S wave detection windows according to the average RR intervals, and determining the positions of the Q wave crest and the S wave crest by the derivative of the sample points in the detection windows.
The pretreatment is carried out before the R wave detection, so that the waveform form of the R wave becomes obvious. The filtered electrocardiosignal is passed through a low-pass filter and a high-pass filter, the two filters form a band-pass filter, the interference of other waveforms such as P wave, T wave and the like is reduced, and the QRS wave group is reservedA frequency band in which energy is concentrated; and then, deriving each sample point by point, taking an absolute value, amplifying slope information of the R wave, and finally, carrying out window integration on the signal to make the signal smooth. After determining the peak position of the R wave, obtaining the RR interval average value of the electrocardiosignal
For the kth R wave peak position R k A Q wave peak detection window will be defined. Initializing Q wave guide number prevDiffQ, sequentially calculating derivative DiffQ of each sample point in Q wave detection window from right to left, whenWhen this point is considered the Q wave peak. Defining the S-wave crest detection window as +.>Initializing S waveguide number prevDiffS, sequentially calculating derivative DiffS of each sample point in the S wave detection window from left to right, and when +.>When this point is considered to be the S-wave peak.
The specific steps of T wave detection are as follows:
assuming that the sampling frequency of the electrocardiosignal s (k) is fs, the sampling frequency is represented by the ith R wave crest R i And the (i+1) th R wave peak R i+1 Between the detection of the T wave end and the T wave peak, R i And R is i+1 Is defined as RR i . Defining the T-wave end search interval as [ k ] a ,k b ]，k a And k b Are all located at R i And R is i+1 Between them.
For k=k a ,k a +1,…,k b Calculating the average value of the electrocardiosignal amplitude values corresponding to p sample points before and after each k valueAnd k integral area A under a sliding window of length w k ：
The size w of the sliding window should not exceed the width of the T-wave, otherwise a band higher than the T-wave amplitude may be brought into the range of integration, causing a T-wave end detection error. However, the width of the T wave is not known in practice, and w is chosen in the patent as the empirical value 32. The smoothing window parameter p is used for smoothing the signal amplitude of k points, reducing noise influence, and p is far smaller than w, otherwise, the calculated value is calculatedThe signal amplitude, which cannot effectively represent the k point, is chosen here as p=4.
Find [ k ] a ,k b ]Points k' and k "corresponding to the maximum area and the minimum area in the range:
when k' corresponds to area A k′ Area A corresponding to k' k″ The method meets the following conditions:
the T wave is considered to be a two-phase T wave, the relatively larger point of k 'and k' is taken as the T wave end Tend, otherwise |A k′ I and Ak ″ The k point corresponding to the larger value in the l is taken as the end Tend of the T wave. Wherein the parameter lambda is to ensure that a possible two-phase T can be identifiedT wave in wave form. The two-phase T wave has a positive T wave peak and a negative T wave peak, and the peak values of the two wave peaks are not greatly different, thus obtaining A k′ And A k″ Can be within a factor of λ, the present embodiment sets λ to 6.
At [ k ] a ,Tend]And searching the point with the largest difference value with the Tend amplitude in the range as a T wave crest Teak.
The initial feature extraction and construction feature set specifically comprises:
the filtered electrocardiosignals are represented by s (x), and the s (x) has n cardiac cycles and the sampling frequency is fs. In addition to the heart rate variability index rMSSD, the initial features of RR interval, QRS time limit, QTc interval, tp-Te/QT index, T wave amplitude and other factors of different samples are different, and cannot directly form a feature set for model training. In the embodiment, a feature set suitable for model training is constructed by calculating the average value, standard deviation and approximate entropy of each initial feature, and elements in the feature set and the initial features corresponding to the elements are finally determined.
The feature scaling is specifically:
these features selected in this embodiment are substantially similar to normal distributions, wherein the feature values of the mean tamp_mean of the feature T wave amplitude are distributed between [0,3], and the feature values of the standard deviation qrs_std of the feature QRS time period are distributed between [0,0.1], and the dimensions are different, requiring unified dimension or normalization processing. And the standard scaling is carried out on the features, and the variation range of each feature is in the same order of magnitude after the normalization processing, so that the learning of the support vector machine is facilitated. When the feature distribution is similar to a normal distribution, feature scaling using normalization may be prioritized.
The model parameter optimizing and optimizing model specifically comprises the following steps:
the feature set after feature scaling is divided into four data sets DS1, DS2, DS3 and DS4, which contain the same normal sinus rhythm electrocardiographic features but different sudden cardiac death electrocardiographic features. The data sets DS1, DS2, DS3 and DS4 contain the electrocardiographic features 20-30 min, 30-40 min, 40-50 min and 60-70 min before sudden cardiac death, respectively. The cardiac sudden death electrocardio characteristics and the normal sinus rhythm electrocardio characteristics in the four data sets are randomly distributed into a training set and a testing set according to a certain proportion, and the record numbers of the training set and the testing set of the prediction model in the corresponding electrocardio database are ensured to be completely separated.
And selecting a nonlinear support vector machine as a basic model for predicting sudden cardiac death, and selecting a Gaussian kernel function as a kernel function of the support vector machine. The support vector machine has two parameters: error penalty parameter C and kernel parameter γ. The improved grid search method and the quantum particle swarm optimization algorithm are selected as parameter optimization algorithms, and the influence of the two parameter optimization algorithms on the performance of the sudden cardiac death prediction model is compared.
And (3) repeating independent experiments on the support vector machine by using an improved grid search method, wherein the training set and the testing set which are divided by each experiment are different. Setting search ranges of parameters C and gamma, wherein the parameters C and gamma form a grid, and each point in the grid corresponds to a combination of the parameters C and gamma. The training set is subdivided into a new training set and a verification set by using a K-fold cross-validation method. C corresponding to the model with highest average classification accuracy * And gamma * The value is used as the optimal parameter combination of preliminary search, and the parameter C is set again to be in [0.1C with a certain step length * ,10C * ]Sequentially taking values in the range, and setting the parameter gamma to be in [0.1 gamma ] with a certain step length * ,10γ * ]Sequentially taking values in the range, repeating the steps to obtain a new optimal parameter combination of fine search by using C and gamma corresponding to the model with highest classification accuracy, and verifying the optimal parameter combination of preliminary search and two support vector machines corresponding to the optimal parameter combination of fine search by using a test set.
From the accuracy of the model corresponding to the combination of the two optimal parameters of the preliminary search and the fine search in the improved grid search method on the test set, it can be seen that the accuracy of the prediction model can be improved by performing the fine search again on the basis of the preliminary search. In this embodiment, C and γ corresponding to the model with the highest average classification accuracy on the verification set are used as the optimal parameter combinations, but the situation that the accuracy corresponding to a plurality of C and γ is the highest may occur in the grid, at this time, a group with the smallest values of C and γ is selected as the optimal parameter combination by default, but the performance of the support vector machine corresponding to the parameter combination on the test set is not necessarily the best. Thus, the overall accuracy of the grid search method, which is improved, will be better performing results on the test set in both the preliminary search and the fine search.
Repeating independent experiments are carried out on the quantum particle swarm optimization algorithm, the training set and the testing set used in each experiment are in one-to-one correspondence with the training set and the testing set of the independent repeated experiments in the improved grid search method, and the classification performance of the support vector machine model based on the two parameter optimization algorithms on the testing set is transversely compared. A particle group X= { X_1, X_2, …, X_N } composed of N particles is defined, the current position of the particle i at the moment t is X_i (t), the individual optimal position is Prest_i (t), the global optimal positions of all particles are Gbest (t), and the global optimal position of the particle group in the whole optimizing process is Gbest. The number of particles is set, and the number of optimized parameters determines the dimension of the particles. And defining a fitness function for evaluating whether the current position of the particle is an optimal position, wherein the fitness function is a support vector machine based on a Gaussian kernel, and the smaller the fitness value is, the better the fitness of the particle is, and the closer the fitness of the particle is to the position of an optimal parameter.
The parameter searching process of the quantum particle swarm optimization algorithm has strong randomness, so that the consumed time is only 1/10 of that of the grid searching method. In the case of classification accuracy, sensitivity may be another relatively important indicator of identifying sudden cardiac death effects. The cardiac sudden death prediction period selected in the embodiment is 20-70 min before the occurrence of cardiac sudden death, so that the embodiment is more prone to selecting a grid search method with higher sensitivity as a parameter optimizing algorithm of a support vector machine under the condition of almost identifying accuracy. Of course, if the prediction period is within 10min, the vector particle swarm optimization algorithm or other less time consuming parameter optimization algorithm should be prioritized, or the search step in the grid search method should be increased by some amount.
The verification effect feedback reconstruction risk factor specifically comprises the following steps:
the following four indexes are adopted to evaluate the performance of the sudden death risk factor reconstruction verification model:
sensitivity (Se): the ratio of the number of SCD samples which are predicted to be correct by the model to the number of total SCD samples is represented, the identification capacity of the model to the sudden cardiac death samples is reflected, and the definition formula is as follows:
specificity (Sp): the ratio of the number of NSR samples predicted to be correct by the model to the number of total NSR samples is represented, the identification capacity of the model to normal sinus rhythm samples is reflected, and the definition formula is as follows:
positive predictive rate (Positive Predictive Value, ppv): the ratio of the number of SCD samples predicted correctly by the model to the number of SCD samples predicted is shown, the identification capability of the model to the sudden cardiac death samples is also shown, and the definition formula is as follows:
accuracy (Accuracy, acc): the ratio of the number of samples correctly predicted as SCD class and NSR class to the total number of samples is represented, the integral prediction capacity of the model is reflected, and the definition formula is as follows:
where True Positive (TP) indicates that the model predicts the correct number of SCD class samples, true Negative (TN) indicates that the correct number of NSR class samples is predicted, false Positive (FP) indicates that the NSR class is predicted to be the number of SCD class samples, and False Negative (FN) indicates that the SCD class is predicted to be the number of NSR class samples. The sudden cardiac death and the normal sinus rhythm samples of the data set are balanced in number, so that accuracy is an evaluation index which most intuitively reflects the overall prediction performance of the model. This embodiment focuses more on the ability to verify sudden death risk factor reconstruction than on normal sinus rhythm samples, so sensitivity is another important evaluation indicator. Specificity and positive predictive rate are also common indicators for evaluating classifier performance. The verification effect of the model directly reflects the prediction capability of sudden death risk factors on cardiac sudden death, the model directly acts as feedback on the sudden death risk factor extraction and reconstruction stage, and the possibility that the extracted factors are effective can be improved by reconstruction verification of the potential sudden death risk factors extracted by manual extraction or intelligent algorithm while a set of sudden death risk factor verification methodology is provided for us.
The above scheme provided in this embodiment may be stored in a computer readable storage medium in a coded form, implemented in a computer program, and input basic parameter information required for calculation through computer hardware, and output a calculation result.
It will be apparent to those skilled in the art that embodiments of the present invention may be provided as a method, apparatus, or computer program product. Accordingly, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The present invention is described with reference to flowchart illustrations of methods, apparatus (means), and computer program products according to embodiments of the invention. It will be understood that each flow of the flowchart, and combinations of flows in the flowchart, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows.
Finally, it should be noted that: the above embodiments are only for illustrating the technical aspects of the present invention and not for limiting the same, and although the present invention has been described in detail with reference to the above embodiments, it should be understood by those of ordinary skill in the art that: modifications and equivalents may be made to the specific embodiments of the invention without departing from the spirit and scope of the invention, which is intended to be covered by the claims.
The patent is not limited to the best mode, any person can obtain other various types of systems and methods for multi-feature quantization of the nonlinear support vector machine and optimizing and reconstructing the sudden cardiac death risk factor by model parameters under the teaching of the patent, and all equivalent changes and modifications made according to the scope of the patent are covered by the patent.
Claims (8)
1. A system for reconstructing sudden cardiac death risk factors by multi-feature quantization and model parameter optimization of a nonlinear support vector machine is characterized by comprising:
the preprocessing module is used for preprocessing data of the cardiac sudden death electrocardiosignal data set and the normal sinus rhythm electrocardiosignal data set;
the waveform detection module is used for carrying out electrocardiographic waveform detection on the preprocessed electrocardiographic data set and obtaining electrocardiographic waveforms for sudden death risk factor extraction;
the initial characteristic acquisition module is used for extracting the sudden cardiac death risk factors to be confirmed according to the electrocardio waveforms to obtain electrocardio initial characteristics;
the feature set construction module is used for processing the electrocardio initial features to obtain a training set;
the characteristic optimizing module is used for obtaining a prediction model of sudden cardiac death according to the training set through the formulated sudden cardiac death risk factors and the optimized model parameters, and verifying the recombined sudden cardiac death risk factors through model feedback; taking a nonlinear support vector machine as a verification model of the sudden cardiac death risk factor, carrying out model parameter optimization based on a Gaussian kernel function, and determining an error penalty parameter C and a kernel parameter；
The characteristic optimizing module uses an improved grid searching method and a quantum particle swarm optimization algorithm as parameter optimizing algorithms respectively, compares the influence of the two parameter optimizing algorithms on the performance of the sudden cardiac death prediction model, and determines an error punishment parameter C and a nuclear parameterThe method comprises the steps of carrying out a first treatment on the surface of the Taking sensitivity Se, specificity Sp, positive prediction rate Ppv and accuracy Acc as indexes of model evaluation;
if the selected sudden cardiac death prediction period is 20-70 min before sudden cardiac death occurs, selecting an improved grid search method with high sensitivity as a parameter optimizing algorithm of a support vector machine under the condition that the difference of recognition accuracy is within a preset range; if the selected cardiac sudden death prediction period is within 10 minutes before cardiac sudden death occurs, a quantum particle swarm optimization algorithm with less time consumption is selected as a parameter optimization algorithm of a support vector machine;
classifying electrocardiographic data of 20-30 min, 30-40 min, 40-50 min and 60-70 min before occurrence of sudden cardiac death as an electrocardiographic signal data set of sudden cardiac death; the characteristic optimizing module outputs four models respectively corresponding to four time periods.
2. The system for reconstructing sudden cardiac death risk factor by multi-feature quantization and model parameter optimization of a nonlinear support vector machine according to claim 1, wherein: the preprocessing module comprises: the device comprises a data cutting sub-module, a frequency band decomposing sub-module, a noise filtering sub-module and a signal reconstructing sub-module.
3. The system for reconstructing sudden cardiac death risk factor by multi-feature quantization and model parameter optimization of a nonlinear support vector machine according to claim 1, wherein: the waveform detection module detects the QRS complex, compares the waveform integration area size under a sliding window with a fixed size in a retrieval interval based on the RR interval, and detects the tail end and the peak of the T wave.
4. The system for reconstructing sudden cardiac death risk factor by multi-feature quantization and model parameter optimization of a nonlinear support vector machine according to claim 1, wherein: the initial feature acquisition module pair includes: the potential factors of the electrocardiosignal sequence RR interval, the QRS time limit, the QTc interval, the T wave crest interval, the Tp-Te/QT index, the T wave amplitude and the heart rhythm variability are extracted.
5. The system for reconstructing sudden cardiac death risk factor by multi-feature quantization and model parameter optimization of a nonlinear support vector machine according to claim 1, wherein: the feature set construction module calculates an average value, a standard deviation and an approximate entropy of the initial features, and is used for constructing a training set.
6. The system for reconstructing sudden cardiac death risk factor by multi-feature quantization and model parameter optimization of a nonlinear support vector machine according to claim 5, wherein: the feature set construction module further comprises an operation of performing feature scaling processing on the initial features after constructing the training set: firstly, determining the difference of the dimensions among the features, and then, selecting the minimum maximum value scaling or standard scaling to process each feature to obtain feature values in the same dimension and the same order of magnitude.
7. The method for reconstructing the sudden cardiac death risk factor by multi-feature quantization and model parameter optimization of the nonlinear support vector machine is characterized by comprising the following steps:
step S1: performing data preprocessing on an electrocardiosignal data set of sudden cardiac death and a normal sinus rhythm electrocardiosignal data set;
step S11: classifying cardiac sudden death electrocardio data according to four time periods of 20-30 min, 30-40 min, 40-50 min and 60-70 min before the occurrence of a cardiac sudden death event, and then cutting cardiac sudden death signals of each time period into data with the time length of 1 min;
step S12: for normal sinus rhythm data, randomly taking a length of 10min for each note, and cutting the note into a plurality of lengths of 1 min;
step S13: processing the electrocardiosignals by using discrete wavelet transformation and a Mallat algorithm to obtain multi-layer electrocardiosignal decomposition signals with different frequency bands;
step S14: performing threshold processing on the decomposed signal by using a threshold value and a threshold function, and filtering electrocardiosignal signals;
step S15: reconstructing the multi-layer decomposition signals with noise filtered to obtain filtered electrocardiosignals;
step S2: carrying out electrocardiographic waveform detection on the electrocardiographic data set processed in the step S1, and obtaining electrocardiographic waveforms for sudden death risk factor extraction; detecting a QRS wave group based on an R wave crest detection method for constructing a self-adaptive threshold and a QRS wave group detection algorithm of a multi-decision rule, comparing the size of a waveform integration area under a sliding window with a fixed size in a retrieval interval based on RR intervals, and detecting the tail end of a T wave and the peak of the T wave;
step S3: extracting cardiac sudden death risk factors to be confirmed to obtain electrocardiographic initial features, and constructing a feature set suitable for model training for respective average value, standard deviation and approximate entropy of the initial features;
step S4: performing feature scaling processing on the initial features extracted in the step S3;
step S5: taking a nonlinear support vector machine as a verification model of the sudden cardiac death risk factor, carrying out model parameter optimization based on a Gaussian kernel function, and determining an error penalty parameter C and a kernel parameter；
Step S51: dividing four data sets corresponding to four time periods into a training set and a testing set according to a certain proportion, and sending the training set into a support vector machine;
step S52: the improved grid search method and quantum particle swarm optimization algorithm are used as parameter optimizing algorithms, the influence of the two parameter optimizing algorithms on the performance of the sudden cardiac death prediction model is compared, and an error punishment parameter C and a nuclear parameter are determined；
If the selected sudden cardiac death prediction period is 20-70 min before sudden cardiac death occurs, selecting an improved grid search method with high sensitivity as a parameter optimizing algorithm of a support vector machine under the condition that the difference of recognition accuracy is within a preset range; if the selected cardiac sudden death prediction period is within 10 minutes before cardiac sudden death occurs, a quantum particle swarm optimization algorithm with less time consumption is selected as a parameter optimization algorithm of a support vector machine;
step S6: obtaining a prediction model of sudden cardiac death through the formulated sudden cardiac death risk factors and the optimized model parameters, and verifying the recombined sudden cardiac death risk factors through model feedback;
in step S6, four models are output, corresponding to four time periods, respectively, and the sensitivity Se, the specificity Sp, the positive predictive rate Ppv, and the accuracy Acc are used as indexes for model evaluation.
8. The method for reconstructing sudden cardiac death risk factor by multi-feature quantization and model parameter optimization of a nonlinear support vector machine according to claim 7, wherein the method comprises the steps of:
the step S2 specifically comprises the following steps:
step S21: passing the filtered electrocardiosignal through a low-pass filter and a high-pass filter, wherein the two filters form a band-pass filter;
step S22: calculating the point-by-point of each sample point, taking the absolute value, amplifying the slope information of the R wave, and finally carrying out window integration on the signal;
step S23: correcting the R wave determined by the R wave detection rule, and determining the positions of the Q wave crest and the S wave crest based on the RR interval;
step S24: integrating the waveform in the searching interval of the T wave end by using a sliding window with fixed length, comparing the integration results to determine the T wave end, and then searching forward according to the T wave end to determine the position of the T wave crest value;
in step S4, the difference of the dimensions between the features is determined, and each feature is processed by selecting the minimum maximum scaling or the standard scaling, so as to obtain feature values in the same dimension and in the same order of magnitude.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111251594.XA CN113974647B (en) | 2021-10-26 | 2021-10-26 | System and method for reconstructing sudden cardiac death risk factor |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111251594.XA CN113974647B (en) | 2021-10-26 | 2021-10-26 | System and method for reconstructing sudden cardiac death risk factor |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113974647A CN113974647A (en) | 2022-01-28 |
| CN113974647B true CN113974647B (en) | 2023-08-04 |
Family
ID=79742028
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111251594.XA Active CN113974647B (en) | 2021-10-26 | 2021-10-26 | System and method for reconstructing sudden cardiac death risk factor |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113974647B (en) |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104107042A (en) * | 2014-07-10 | 2014-10-22 | 杭州电子科技大学 | Electromyographic signal gait recognition method based on particle swarm optimization and support vector machine |
| CN106344006A (en) * | 2016-11-03 | 2017-01-25 | 太原理工大学 | J wave detection method based on pole symmetrical mode decomposition and support vector machine |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8529448B2 (en) * | 2009-12-31 | 2013-09-10 | Cerner Innovation, Inc. | Computerized systems and methods for stability—theoretic prediction and prevention of falls |
| EP3616605A1 (en) * | 2018-08-30 | 2020-03-04 | Tata Consultancy Services Limited | Non-invasive detection of coronary heart disease from short single-lead ecg |
-
2021
- 2021-10-26 CN CN202111251594.XA patent/CN113974647B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104107042A (en) * | 2014-07-10 | 2014-10-22 | 杭州电子科技大学 | Electromyographic signal gait recognition method based on particle swarm optimization and support vector machine |
| CN106344006A (en) * | 2016-11-03 | 2017-01-25 | 太原理工大学 | J wave detection method based on pole symmetrical mode decomposition and support vector machine |
Non-Patent Citations (6)
| Title |
|---|
| 基于动态心电的支持向量机模型对心肌梗死后心脏性死亡的预警研究;尹蕾蕾;宋涛;张英涛;王秀杰;承恒达;曲秀芬;;中国心脏起搏与心电生理杂志(第05期);第408-412页 * |
| 基于心电信号循环平稳特征的心脏性猝死识别研究;张爱华;师文韬;;中国医疗器械杂志(第05期);第322-326页 * |
| 基于心电特征参数的心肌梗死疾病辅助诊断模型的建立;张清丽;苏士美;尹咪咪;张建华;刘莹;;郑州大学学报(医学版);第52卷(第02期);第151-153页 * |
| 基于支持向量机的水质浊度补偿研究;李文;王兴浩;何云霄;罗学科;;计算机测量与控制;第28卷(06);第140-143页 * |
| 基于自适应向量机检测睡眠呼吸暂停综合征的最优特征组合筛选;王新康 等;《中国医药导报》;第16卷(第12期);165-168 * |
| 短时非线性方法用于心率变异性分析;刘红夺儿;朱逸;湛萍;王志刚;彭屹;;中国生物医学工程学报(第02期);第229-234页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113974647A (en) | 2022-01-28 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US8204583B2 (en) | System for seizure monitoring and detection | |
| US9380954B2 (en) | Method for physiological signal analysis and its system and computer program product storing physiological signal analysis program | |
| Diker et al. | A diagnostic model for identification of myocardial infarction from electrocardiography signals | |
| US10357169B2 (en) | Methods for determining whether patient monitor alarms are true or false based on a multi resolution wavelet transform and inter-leads variability | |
| Kelwade et al. | Prediction of cardiac arrhythmia using artificial neural network | |
| Swapna et al. | Automated detection of diabetes using higher order spectral features extracted from heart rate signals | |
| Kshirsagar et al. | Classification of ECG-signals using artificial neural networks | |
| Raj et al. | Automated recognition of cardiac arrhythmias using sparse decomposition over composite dictionary | |
| CN111832537B (en) | Abnormal electrocardiosignal identification method and abnormal electrocardiosignal identification device | |
| CN114052744A (en) | Electrocardiosignal classification method based on pulse neural network | |
| Mahesh et al. | ECG arrhythmia classification based on logistic model tree | |
| Krak et al. | Electrocardiogram classification using wavelet transformations | |
| Vimal et al. | Random forest classifier based ECG arrhythmia classification | |
| CN113974647B (en) | System and method for reconstructing sudden cardiac death risk factor | |
| Amhia et al. | Designing an optimum and reduced order filter for efficient ECG QRS peak detection and classification of arrhythmia data | |
| Afonso et al. | Filter bank-based ECG beat classification | |
| Sanamdikar et al. | Analysis of several characteristics of ECG signal for cardiac arrhythmia detection | |
| Murthy et al. | Design and implementation of hybrid techniques and DA-based reconfigurable FIR filter design for noise removal in EEG signals on FPGA | |
| Talatov et al. | Methodology for processing and analysis of diagnostic indicators electrocardiogram based on Labview | |
| Jayagopi et al. | On the classification of arrhythmia using supplementary features from tetrolet transforms. | |
| CN117582219B (en) | Exercise-based hypoxia prediction method and device for continuous physiological data based on six-minute walking test | |
| Tadejko et al. | Hybrid wavelet-mathematical morphology feature extraction for heartbeat classification | |
| Bouaziz et al. | Automatic detection method of R-wave positions in electrocardiographic signals | |
| Tobón et al. | Improved heart rate variability measurement based on modulation spectral processing of noisy electrocardiogram signals | |
| Shankar et al. | Epilepsy Detection Using Embedded Machine Learning |
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 |