CN115115038A - Model construction method based on single lead electrocardiosignal and gender identification method - Google Patents
Model construction method based on single lead electrocardiosignal and gender identification method Download PDFInfo
- Publication number
- CN115115038A CN115115038A CN202211050595.2A CN202211050595A CN115115038A CN 115115038 A CN115115038 A CN 115115038A CN 202211050595 A CN202211050595 A CN 202211050595A CN 115115038 A CN115115038 A CN 115115038A
- Authority
- CN
- China
- Prior art keywords
- electrocardiosignal
- model
- lead
- data
- value
- 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
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
-
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
Landscapes
- Health & Medical Sciences (AREA)
- Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Public Health (AREA)
- Veterinary Medicine (AREA)
- Theoretical Computer Science (AREA)
- Mathematical Physics (AREA)
- Pathology (AREA)
- Cardiology (AREA)
- Heart & Thoracic Surgery (AREA)
- Medical Informatics (AREA)
- Evolutionary Computation (AREA)
- Surgery (AREA)
- Animal Behavior & Ethology (AREA)
- Physiology (AREA)
- Signal Processing (AREA)
- Psychiatry (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Software Systems (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Fuzzy Systems (AREA)
- Measurement And Recording Of Electrical Phenomena And Electrical Characteristics Of The Living Body (AREA)
Abstract
The invention discloses a model construction method based on a single-lead electrocardiosignal and a gender identification method, and belongs to the technical field of electrocardiosignal identification. Utilizing 12-lead electrocardiosignals to carry out gender identification research, wherein information redundancy exists, and the resource consumption and the time consumption are calculated; the existing single-lead gender identification method is easily interfered by noise signals and has limited accuracy. The invention provides a model construction method and a gender identification method based on a single lead electrocardiosignal for the first time, which comprises the following steps: preprocessing electrocardiosignal data, evaluating signal quality, constructing a deep neural network model and training the model, and predicting the model through the constructed model. Compared with a 12-lead gender identification method, the method saves more resources and consumes less time, and compared with the existing single-lead gender identification machine learning method, the method has better robustness and higher accuracy.
Description
Technical Field
The invention relates to the technical field of electrocardiosignal identification, in particular to a model construction method and a gender identification method based on a single-lead electrocardiosignal.
Background
Today, gender identification is used in many areas such as forensics, testing, and demographic data collection. One of the most important applications is the application in the medical field, particularly the application in the aspect of man-machine interaction of unknown sex, and the different sexes have obvious physiological and psychological differences; in addition, the sex identification result can reflect the change of hormone level to a certain extent, and accurate sex identification can provide more accurate treatment scheme and action suggestion for the user.
An Electrocardiogram (ECG) is a biomedical signal that reflects detailed information about heart activity in the form of electrical signals. Studies have shown that all individuals have their own uniqueness in their cardiac electrical signals, indicating the feasibility of gender identification based on electrocardiograms.
The sex recognition research on the electrocardiogram can be roughly divided into two types of single-lead electrocardiogram and 12-lead electrocardiogram according to the electrocardiogram type, and the sex recognition of 12-lead electrocardiogram signals is a deep learning method. First, the conditions required for 12-lead signal acquisition are more stringent than for single lead, requiring 10 electrode pads to be attached to each part of the body in a scattered manner. Second, 12-lead data analysis consumes more computing resources and takes longer. Meanwhile, research has shown that lead I contains information capable of identifying gender, and 12-lead data is used for identifying gender, so that information redundancy is caused. At present, for gender identification of a single-lead electrocardiosignal, feature extraction (such as HRV) is usually performed by a pattern recognition method, and then gender classification is performed by using the existing machine learning method according to the extracted features.
Chinese patent application, application No. CN202011584667.2, published 2021, 04/06, discloses an electrocardiographic signal identification and classification method, comprising: constructing a 12-lead electrocardiosignal Embedding module based on a ResNet deep neural network; resampling an input sample, taking each lead as a different channel, and inputting the different channel into a one-dimensional ResNet network; training the constructed Embedding module by using a training set, wherein N electrocardiosignals are shared; performing word Embedding of N-dimensional vectors on all electrocardiosignals by using a trained Embedding module, and finally generating a vector with the length of N bits for each electrocardiosignal, wherein each bit represents the probability of belonging to a certain electrocardiosignal; extracting basic characteristics and morphological characteristics of the electrocardiosignals; combining the N-dimensional vectors obtained after the electrocardiosignals pass through the Embedding module with the extracted features as input, outputting N electrocardiosignals as N binary classes, and establishing N binary models based on LightGBM; and identifying and classifying the electrocardiosignals by using the established two classification models. The method can automatically extract effective characteristics from the electrocardiosignals to identify the electrocardio types, improves the identification efficiency and accuracy, uses the 12-lead electrocardiosignals to predict and identify, and has long calculation time and more calculation resource consumption.
Disclosure of Invention
1. Technical problem to be solved
Utilizing 12-lead electrocardiosignals to carry out gender identification research, wherein information redundancy exists, and the time consumption of consumed resources is calculated; the existing single-lead gender identification method is easily interfered by noise signals and has limited accuracy. The invention provides a model construction method and a gender identification method based on a single-lead electrocardiosignal for the first time, the method saves more resources and consumes less time compared with a 12-lead gender identification method, and compared with the existing single-lead gender identification machine learning method, the method has better robustness and higher accuracy.
2. Technical scheme
The purpose of the invention is realized by the following technical scheme.
A model construction method based on a single lead electrocardiosignal comprises the following steps:
preprocessing electrocardiosignal data: collecting electrocardiosignal data, using I-lead electrocardiosignals, unifying the sampling rate of the electrocardiosignal data into the same numerical value, randomly intercepting original data by using a window with the length of d, if the length of the original data is less than d, supplementing 0 after the data to ensure that the length reaches d, d is the signal length, and carrying out the treatment on all preprocessed electrocardiosignal X i Marking is carried out, the marking numerical value isWhere 0 indicates that the user is female and 1 indicates that the user is male.
Constructing a deep neural network model: constructing a deep neural network F, inputting an electrocardiosignal X, and outputting a gender prediction probability corresponding to the electrocardiosignalThe value range of the probability value y is 0 to 1, and the notation y = F (X); the deep neural network is formed by connecting K one-dimensional convolution layers and 1 full-connection layer in sequence.
Model training: the parameters of the deep neural network model F to be trained are all variables defined during model construction, and the variables are the weight of edges connected between any two layers of the network; randomly initializing the parameters to make them satisfy a normal distribution with a mean of 0 and a variance of 1; defining a trained objective function Loss and measuring a real label l i And a prediction probability y i The difference degree between the two parameters optimizes the parameters of the deep neural network model; calculating data of a batch of samples which are transmitted to the nodes of the output layer in the forward direction in each iteration, calculating to obtain a gradient, then performing backward transmission, and updating network model parameters in the backward transmission process; and finally obtaining stable network model parameters through multiple rounds of iteration, storing the final network model parameters, and finishing the model construction after the training process is finished.
Further, the deep neural network model adopts Sigmoid function as activation function:
where x is the inactive output value. The Sigmoid function is monotonous and continuous, the output range is limited, so that data is not easy to diverge in the transmission process, and the output range is (0,1), so that the Sigmoid function can be used for activating a model output layer and outputting an expression probability. And the Sigmoid function is derivable everywhere within the domain of definition, the derivation is easy.
Furthermore, a cross entropy loss function CrossEntrol is selected for model training. Prediction result y of cross entropy comparison model i And the true tag l of the data i And as the prediction is more accurate, the value of the cross entropy is smaller, and if the prediction is completely correct, the value of the cross entropy is 0. Cross entropy as a loss function in deep neural networks,/ i Representing the distribution of the real marks, y i Then the cross entropy loss function may measure l for the predicted label distribution of the trained model i And y i The similarity of (c). Cross entropy as a loss function also has the advantage that the problem of the learning rate reduction of the mean square error loss function can be avoided when the gradient is decreased by using Sigmoid function, because the learning rate can be controlled by the output error.
If l is i =1,CrossEntropy(l i ,y i )=-log(y i );
If l is i =0,CrossEntropy(l i ,y i )=-log(1-y i );
The optimization problem is solved.
Furthermore, in the model training, the parameters of the deep neural network model are optimized by using a stochastic gradient descent method. Gradient descent is a mathematical method in solving the minimum of the loss function, and the purpose of gradient descent is to minimize the loss function, i.e., to minimize the loss function. The step-by-step iterative solution can be performed by a gradient descent method to obtain the minimized loss function and the model parameter value. Two gradient descent methods, namely a random gradient descent method and a batch gradient descent method, are developed based on a basic gradient descent method. The stochastic gradient descent algorithm randomly selects only one sample at a time to update the model parameters, which is random, that is, we approximate all of our samples with one example of the samples, and is not globally optimal because the computed gradient is not exactly one. But such methods converge faster than bulk gradients and are therefore also more widely used.
Furthermore, after the step of preprocessing the electrocardiosignal data, the method further comprises the following steps of signal quality evaluation:
detecting the position of the qrs wave of the electrocardiosignal, and cutting out signal segments S with the length of 0.5S at two sides according to the detected position i ;
Averaging the cut signals according to dimensions to obtain an average waveform M;
wherein M is j Is the jth voltage value of the average waveform, n is the number of segments cut out from the signal X, S ij The j voltage value of the ith signal segment;
Where cov is the covariance and σ is the standard deviation.
And averaging the correlation coefficients of the segments and the average waveform to obtain the correlation coefficient rho of the current electrocardiosignal X.
Estimating the signal quality from the coefficients: a threshold value of 0.65 is established for the coefficient, and the signal quality is considered to be poor when the threshold value is less than the threshold value, and the signal is directly discarded; if the signal quality is higher than the threshold value, the signal quality is considered to be acceptable, and the corresponding X is i Can be used for subsequent gender identification.
A gender identification method based on a single lead electrocardiosignal comprises the following steps:
using the model of the single-lead electrocardiosignal to carry out model prediction on the input test electrocardiosignal dataWhere d is the signal length, calculating y = f (x) to obtain the predicted probability valueSetting a threshold thresh, if the predicted probability value y is greater than thresh, determining that the electrocardiosignal data corresponds to a male, otherwise, determining that the electrocardiosignal data corresponds to a female.
Further, the threshold thresh is 0.5.
3. Advantageous effects
Compared with the prior art, the invention has the advantages that:
by the deep learning method based on the gender identification on the single lead signal, computing resources can be saved, time consumption is short, the anti-interference capability is high, robustness is good, and accuracy is high.
Drawings
FIG. 1 is a schematic flow diagram of the present invention;
FIG. 2 is a schematic diagram of a model parameter updating process.
Detailed Description
The invention is described in detail below with reference to the drawings and specific examples.
Example 1
The embodiment discloses a model construction method based on a single lead electrocardiosignal, which comprises the following steps in combination with fig. 1:
preprocessing electrocardiosignal data, evaluating signal quality, constructing a deep neural network model and training the model, and predicting the model through the constructed model.
Step 1: preprocessing of electrocardiographic signal data
Step 1.1: the method collects electrocardiosignal data from various sources, particularly, only adopts electrocardiosignals using I leads, and firstly, the sampling rates of the electrocardiosignals are unified into the same value, wherein 500Hz is taken as an example. If the signal sampling rate is not 500Hz, the resampling algorithm adopting linear interpolation is adjusted to be 500 Hz.
Step 1.2: and randomly intercepting the original data by using a window with the length of d, and if the length of the original data is less than d, supplementing 0 behind the data to enable the length to reach d. The size of d can be selected according to actual requirements, 10 seconds are generally recommended to be selected, and the model noise identification accuracy is reduced to a certain extent due to too large or too small d. Recording the preprocessed electrocardiosignal of the ith strip asWhere d is the signal length.
For all the preprocessed electrocardiosignals X i Marking is carried out, the marking numerical value is. Where 0 indicates that the user is female and 1 indicates that the user is male.
Step 2: signal quality assessment
In order to reduce the influence of noise signals on model prediction, a signal quality evaluation module is added in the method, and the module is divided into the following parts:
step 2.1: detecting the position of the qrs-detect of the electrocardiosignal qrs wave by using a gqrs-detect method in a Python toolkit wfdbs, and cutting out signal segments S with the lengths of 0.5S at two sides according to the detected position i ;
Step 2.2: averaging the cut signals according to dimensions to obtain an average waveform M;
wherein M is j Is the jth voltage value of the average waveform, n is the number of segments cut out from the signal X, S ij Is the jth voltage value of the ith signal segment.
Where cov is the covariance and σ is the standard deviation.
And averaging the correlation coefficients of the segments and the average waveform to obtain the correlation coefficient rho of the current electrocardiosignal X.
Where n is the number of fragments cut out of the signal X.
Estimating the signal quality from the coefficients: a threshold value of 0.65 is established for the coefficient, and the signal quality is considered to be poor when the threshold value is less than the threshold value, and the signal is directly discarded; if the signal quality is higher than the threshold value, the signal quality is considered to be acceptable, and the corresponding X is i Can be used for subsequent gender identification.
And step 3: deep neural network model construction
Constructing a deep neural network F, inputting an electrocardiosignal X, and outputting a gender prediction probability corresponding to the electrocardiosignalThe probability value y ranges from 0 to 1, let y = f (x).
Step 3.1: the deep neural network constructed by the invention is formed by sequentially connecting K one-dimensional Convolution layers (1-D Connected Layer) and 1 Fully Connected Layer (full Connected Layer), and K is determined by experience and training.
The one-dimensional convolution layer establishes local connection among a plurality of channels on the electrocardiosignal data, and can be used for automatically learning and extracting local characteristics of the electrocardiosignal. In addition, the sequential connection of the K one-dimensional convolution layers can learn electrocardiosignal characteristics of a plurality of levels. For example, if the first layer of one-dimensional convolutional layer is learned as "local features", the second layer of one-dimensional convolutional layer is learned as "local features of local features", which have a wider granularity than that of the first layer, and finally obtain "global features" of the entire segment of electrocardiographic signal data. Building a one-dimensional convolutional layer requires specifying the size of the convolutional kernel (kernel _ size), the number of convolutional kernels (filters), the step size of the convolution (stride), and the like. The selection of these parameters is typically tuned on the actual verification data set.
On the 'overall characteristics' of the whole section of electrocardiosignal, linear combination among the characteristics is constructed to obtain the final output. The linear combination is realized by using a full connection layer, namely, all nodes on the upper layer of the network are connected with all nodes on the lower layer of the network. Finally, a Sigmoid function is adopted as an activation function, and the result obtained by linear combination is normalized to a value y with the value range between 0 and 1 i The output of the network can be regarded as the probability that the input electrocardiographic signals correspond to the two genders.
Where x is the inactive output value.
And 4, step 4: model training
The parameters of the deep neural network model F to be trained are all variables defined when the model is constructed in the step 3, and the variables are the weight of the edge connected between any two layers of the network. We initialize these parameters randomly so that they satisfy a normal distribution with a mean of 0 and a variance of 1.
Defining a trained objective function Loss and measuring a real label l i And a prediction probability y i The cross entropy loss function CrossEntrol is selected according to the difference degree.
If l is i =1,CrossEntropy(l i ,y i )=-log(y i );
If l is i =0,CrossEntropy(l i ,y i )=-log(1-y i );
The optimization problem described above is solved. Parameters of the deep neural network model are optimized by using a Stochastic Gradient Descent (SGD) method, data of a batch of samples which are transmitted to nodes of an output layer in the forward direction are calculated in each iteration, the Gradient is calculated and then transmitted in the reverse direction, and the parameters of the network model are updated in the process of the reverse transmission. Through multiple rounds of iteration, stable model parameters are finally obtained, and the fact that the training process of the deep neural network model is converged is shown in the attached figure 2 in detail.
And storing the final network parameters, and finishing the training process.
And 5: model prediction
A gender identification method based on a single lead electrocardiosignal carries out model prediction by using a model of the single lead electrocardiosignal constructed in the step 1-4. In the model prediction stage, the input test electrocardiosignal data is subjected toWhere d is the signal length, calculating y = F (X) to obtain a prediction probability value. And selecting a proper threshold value thresh according to the requirement of the actual situation, if the predicted probability value y is greater than thresh, judging that the electrocardiosignal data corresponds to a male, and if not, judging that the electrocardiosignal data corresponds to a female. If there is no special requirement, the threshold is set to 0.5 by default.
The invention and its embodiments have been described above schematically, without limitation, and the invention can be embodied in other specific forms without departing from the spirit or essential characteristics thereof. The representation in the drawings is only one of the embodiments of the invention, the actual construction is not limited thereto, and any reference signs in the claims shall not limit the claims concerned. Therefore, if a person skilled in the art receives the teachings of the present invention, without inventive design, a similar structure and an embodiment to the above technical solution should be covered by the protection scope of the present patent. Furthermore, the word "comprising" does not exclude other elements or steps, and the word "a" or "an" preceding an element does not exclude the presence of a plurality of such elements. Several of the elements recited in the product claims may also be implemented by one element in software or hardware. The terms first, second, etc. are used to denote names, but not any particular order.
Claims (8)
1. A model construction method based on a single lead electrocardiosignal comprises the following steps:
preprocessing electrocardiosignal data: collecting electrocardiosignal data, using I-lead electrocardiosignals, unifying the sampling rate of the electrocardiosignal data into the same numerical value, randomly intercepting original data by using a window with the length of d, if the length of the original data is less than d, supplementing 0 after the data to ensure that the length reaches d, d is the signal length, and carrying out the treatment on all preprocessed electrocardiosignal X i Marking is carried out, the marking numerical value isWherein0 indicates that the user is female, and 1 indicates that the user is male;
constructing a deep neural network model: constructing a deep neural network F, inputting an electrocardiosignal X, and outputting a gender prediction probability corresponding to the electrocardiosignalThe value range of the probability value y is 0 to 1, and the notation y = F (x); the deep neural network is formed by sequentially connecting K one-dimensional convolution layers and 1 full-connection layer;
model training: the parameters of the deep neural network model F to be trained are all variables defined during model construction, and the variables are weights connected between any two layers of the network; randomly initializing these parameters such that they satisfy a normal distribution with a mean of 0 and a variance of 1; defining a trained objective function Loss and measuring a real label l i And a prediction probability y i The difference degree between the two parameters optimizes the parameters of the deep neural network model; calculating data of a batch of samples which are transmitted to the nodes of the output layer in the forward direction in each iteration, calculating to obtain a gradient, then performing backward transmission, and updating network model parameters in the backward transmission process; obtaining stable network model parameters finally through multiple iterations, storing the final network model parameters, and finishing the training process;
and completing model construction.
3. The model construction method based on the single-lead electrocardiosignal according to claim 1, characterized in that a cross entropy loss function Cross Entrophy is selected in model training:
if l is i =1,CrossEntropy(l i ,y i )=-log(y i );
If l is i =0,CrossEntropy(l i ,y i )=-log(1-y i );
The optimization problem described above is solved.
4. The method for constructing the model based on the single-lead electrocardiosignal according to claim 1 or 3, wherein in the model training, the parameters of the deep neural network model are optimized by using a stochastic gradient descent method.
5. The model construction method based on the single lead electrocardiosignal as claimed in claim 1, characterized in that after the step of preprocessing the electrocardiosignal data, the method further comprises the following steps of signal quality evaluation:
detecting the position of the qrs wave of the electrocardiosignal, and cutting out signal segments S with the length of 0.5S at two sides according to the detected position i ;
Averaging the cut signals according to dimensions to obtain an average waveform M;
wherein M is j Is the jth voltage value of the average waveform, n is the number of segments cut out from the signal X, S ij The j voltage value of the ith signal segment;
Wherein cov is covariance, and σ is standard deviation;
and then averaging the correlation coefficients of the segments and the average waveform to obtain the correlation coefficient rho of the current electrocardiosignal X:
a threshold value is established for the coefficient, the signal quality is considered to be poor when the threshold value is smaller than the threshold value, and the signal is directly discarded; if the signal quality is higher than the threshold value, the signal quality is considered to be acceptable, and the corresponding X is i Can be used for subsequent gender identification.
6. The model construction method based on the single-lead electrocardiosignal according to claim 5, wherein the threshold value of the correlation coefficient p is set to 0.65.
7. A gender identification method based on a single lead electrocardiosignal comprises the following steps:
the model of the single-lead electrocardiosignal of any one of claims 1 to 5 is used for model prediction, the input test electrocardiosignal data is calculated to obtain a prediction probability value, a threshold value thresh is set, if the prediction probability value is greater than thresh, the electrocardiosignal data is judged to be male, otherwise, the electrocardiosignal data is female.
8. The gender identification method based on the single lead electrocardiosignal according to claim 7, characterized in that: the prediction probability value y ranges from 0 to 1, and the threshold thresh is 0.5.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202211050595.2A CN115115038B (en) | 2022-08-30 | 2022-08-30 | Model construction method based on single lead electrocardiosignal and gender identification method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202211050595.2A CN115115038B (en) | 2022-08-30 | 2022-08-30 | Model construction method based on single lead electrocardiosignal and gender identification method |
Publications (2)
Publication Number | Publication Date |
---|---|
CN115115038A true CN115115038A (en) | 2022-09-27 |
CN115115038B CN115115038B (en) | 2022-11-08 |
Family
ID=83335451
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202211050595.2A Active CN115115038B (en) | 2022-08-30 | 2022-08-30 | Model construction method based on single lead electrocardiosignal and gender identification method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN115115038B (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117338309A (en) * | 2023-08-21 | 2024-01-05 | 合肥心之声健康科技有限公司 | Electrocardiosignal approximation threshold calculation method, identification method and storage medium |
Citations (12)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104063645A (en) * | 2014-07-01 | 2014-09-24 | 清华大学深圳研究生院 | Identity recognition method based on ECG dynamic self-updating samples |
CN108968951A (en) * | 2018-08-15 | 2018-12-11 | 武汉中旗生物医疗电子有限公司 | Electrocardiogram detecting method, apparatus and system |
CN109977923A (en) * | 2019-04-12 | 2019-07-05 | 江西科技学院 | Driver's sex-screening method and system based on EEG signals |
CN110403601A (en) * | 2019-08-27 | 2019-11-05 | 安徽心之声医疗科技有限公司 | Electrocardiosignal QRS wave group recognition methods based on deep learning |
CN110477909A (en) * | 2019-07-31 | 2019-11-22 | 同济大学 | A kind of gender classification method based on tranquillization state eeg data |
US20210361217A1 (en) * | 2018-01-12 | 2021-11-25 | Mayo Foundation For Medical Education And Research | Ecg-based age and sex estimation |
CN113693611A (en) * | 2021-08-31 | 2021-11-26 | 平安科技(深圳)有限公司 | Machine learning-based electrocardiogram data classification method and device |
CN113855047A (en) * | 2021-10-19 | 2021-12-31 | 南京大学 | FVEP signal enhancement method based on abnormal segment recognition |
CN113951883A (en) * | 2021-11-12 | 2022-01-21 | 上海交通大学 | Gender difference detection method based on electroencephalogram signal emotion recognition |
CN113990499A (en) * | 2021-10-22 | 2022-01-28 | 安徽心之声医疗科技有限公司 | Heart age assessment method and device, electronic equipment and storage medium |
CN114648049A (en) * | 2022-05-20 | 2022-06-21 | 合肥心之声健康科技有限公司 | Method, device and system for constructing and classifying electrocardio image classification model |
CN114886416A (en) * | 2022-05-07 | 2022-08-12 | 重庆邮电大学 | Heart rate and cardiac cycle detection method based on ballistocardiogram signals |
-
2022
- 2022-08-30 CN CN202211050595.2A patent/CN115115038B/en active Active
Patent Citations (12)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104063645A (en) * | 2014-07-01 | 2014-09-24 | 清华大学深圳研究生院 | Identity recognition method based on ECG dynamic self-updating samples |
US20210361217A1 (en) * | 2018-01-12 | 2021-11-25 | Mayo Foundation For Medical Education And Research | Ecg-based age and sex estimation |
CN108968951A (en) * | 2018-08-15 | 2018-12-11 | 武汉中旗生物医疗电子有限公司 | Electrocardiogram detecting method, apparatus and system |
CN109977923A (en) * | 2019-04-12 | 2019-07-05 | 江西科技学院 | Driver's sex-screening method and system based on EEG signals |
CN110477909A (en) * | 2019-07-31 | 2019-11-22 | 同济大学 | A kind of gender classification method based on tranquillization state eeg data |
CN110403601A (en) * | 2019-08-27 | 2019-11-05 | 安徽心之声医疗科技有限公司 | Electrocardiosignal QRS wave group recognition methods based on deep learning |
CN113693611A (en) * | 2021-08-31 | 2021-11-26 | 平安科技(深圳)有限公司 | Machine learning-based electrocardiogram data classification method and device |
CN113855047A (en) * | 2021-10-19 | 2021-12-31 | 南京大学 | FVEP signal enhancement method based on abnormal segment recognition |
CN113990499A (en) * | 2021-10-22 | 2022-01-28 | 安徽心之声医疗科技有限公司 | Heart age assessment method and device, electronic equipment and storage medium |
CN113951883A (en) * | 2021-11-12 | 2022-01-21 | 上海交通大学 | Gender difference detection method based on electroencephalogram signal emotion recognition |
CN114886416A (en) * | 2022-05-07 | 2022-08-12 | 重庆邮电大学 | Heart rate and cardiac cycle detection method based on ballistocardiogram signals |
CN114648049A (en) * | 2022-05-20 | 2022-06-21 | 合肥心之声健康科技有限公司 | Method, device and system for constructing and classifying electrocardio image classification model |
Non-Patent Citations (2)
Title |
---|
刘守华 等: "基于深度学习的临床心电图分类算法", 《计算机与现代化》 * |
李琦: "正常人右胸导联心电图性别差异研究", 《实用心电学杂志》 * |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117338309A (en) * | 2023-08-21 | 2024-01-05 | 合肥心之声健康科技有限公司 | Electrocardiosignal approximation threshold calculation method, identification method and storage medium |
CN117338309B (en) * | 2023-08-21 | 2024-03-15 | 合肥心之声健康科技有限公司 | Identity recognition method and storage medium |
Also Published As
Publication number | Publication date |
---|---|
CN115115038B (en) | 2022-11-08 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US11562222B2 (en) | Systems and methods of identity analysis of electrocardiograms | |
CN109620205B (en) | Electrocardiogram data classification method and device, computer equipment and storage medium | |
CN110232678B (en) | Image uncertainty prediction method, device, equipment and storage medium | |
CN110522444B (en) | Electrocardiosignal identification and classification method based on Kernel-CNN | |
CN111345816B (en) | Multi-lead QRS complex detection method, device, equipment and storage medium | |
CN110269605B (en) | Electrocardiosignal noise identification method based on deep neural network | |
Kim et al. | Recurrent neural networks with missing information imputation for medical examination data prediction | |
CN112233798B (en) | Interpretable disease risk analysis system based on pathological mode and attention mechanism | |
CN112426160A (en) | Electrocardiosignal type identification method and device | |
Al-Ma’aitah et al. | Enhanced computational model for gravitational search optimized echo state neural networks based oral cancer detection | |
CN115115038B (en) | Model construction method based on single lead electrocardiosignal and gender identification method | |
CN111798980B (en) | Complex medical biological signal processing method and device based on deep learning network | |
Chen et al. | Edge2Analysis: a novel AIoT platform for atrial fibrillation recognition and detection | |
CN113274031A (en) | Arrhythmia classification method based on deep convolution residual error network | |
CN111297327A (en) | Sleep analysis method, system, electronic equipment and storage medium | |
CN114154530A (en) | Training method and device for atrial fibrillation detection model of electrocardio timing signals | |
US6941288B2 (en) | Online learning method in a decision system | |
CN110414562B (en) | X-ray film classification method, device, terminal and storage medium | |
CN111652880A (en) | Neural network-based CT coronary artery centerline seed point detection and tracking method, device, equipment and readable storage medium | |
CN112686091A (en) | Two-step arrhythmia classification method based on deep neural network | |
Zhou et al. | Research on intelligent diagnosis algorithm of diseases based on machine learning | |
Das et al. | BayesBeat: A Bayesian deep learning approach for atrial fibrillation detection from noisy photoplethysmography data | |
Ramu et al. | Cardiovascular Disease Prediction Using Machine Learning Classifiers | |
Mohamed et al. | Atrial fibrillation predictor with reject option using belief functions theory | |
US20240206821A1 (en) | Apparatus and a method for predicting a physiological indicator |
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 |