CN110200624A - Based on convolutional neural networks-Recognition with Recurrent Neural Network-support vector machines mixed model disease identification algorithm - Google Patents
Based on convolutional neural networks-Recognition with Recurrent Neural Network-support vector machines mixed model disease identification algorithm Download PDFInfo
- Publication number
- CN110200624A CN110200624A CN201910591947.7A CN201910591947A CN110200624A CN 110200624 A CN110200624 A CN 110200624A CN 201910591947 A CN201910591947 A CN 201910591947A CN 110200624 A CN110200624 A CN 110200624A
- Authority
- CN
- China
- Prior art keywords
- recognition
- data
- input
- neural network
- support vector
- 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.)
- Pending
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
-
- 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
- A61B5/369—Electroencephalography [EEG]
-
- 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
- A61B5/7246—Details of waveform analysis using correlation, e.g. template matching or determination of similarity
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
- A61B5/7235—Details of waveform analysis
- A61B5/7264—Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems
- A61B5/7267—Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems involving training the classification device
Landscapes
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Public Health (AREA)
- Surgery (AREA)
- Veterinary Medicine (AREA)
- Artificial Intelligence (AREA)
- General Health & Medical Sciences (AREA)
- Biophysics (AREA)
- Pathology (AREA)
- Biomedical Technology (AREA)
- Heart & Thoracic Surgery (AREA)
- Medical Informatics (AREA)
- Molecular Biology (AREA)
- Animal Behavior & Ethology (AREA)
- Psychiatry (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physiology (AREA)
- Signal Processing (AREA)
- Evolutionary Computation (AREA)
- Fuzzy Systems (AREA)
- Mathematical Physics (AREA)
- Psychology (AREA)
- Measurement And Recording Of Electrical Phenomena And Electrical Characteristics Of The Living Body (AREA)
- Image Analysis (AREA)
Abstract
The invention patent devises in non-invasive diagnosis system based on convolutional neural networks-Recognition with Recurrent Neural Network-support vector machines mixed model recognizer, being made an uproar by data waits preprocess methods to obtain data set, convolutional neural networks and Recognition with Recurrent Neural Network is used to carry out feature extraction to sample as end-to-end feature extractor in this field for the first time, time and the space characteristics for extracting data respectively, are finally classified using characteristic of the support vector machines to extraction.The algorithm finally obtains high-accuracy, high specific and high sensitivity, can be used widely in Non-invasive detection field.
Description
[technical field]
The invention patent is directed to disease Non-invasive detection field, in particular to epilepsy screening field.
[background technique]
Although detection sensitivity with higher and the accuracy of existing disorder in screening method, but they are substantially relied on
Expensive equipment and complicated operation, or have some irreversible damages to body.Therefore, it is necessary to a kind of low cost, simply
Effective non-invasive procedures method carries out screening to disease.
Brain electricity contains a large amount of physiology and pathological information, can directly measure on human body, be suitable for clinic and answer
With diagnosis basis can be provided for certain cerebral diseases, or even become the effective treatment means of certain cerebral diseases.In recent years, to recognizing
Know that the research of function is increasingly valued by people, effective analysis, evaluation cognitive function, the detection to cognitive disorder disease
It has great significance with treatment.Epilepsy is a kind of intracerebral neuron paradoxical discharge, leads to the slow of part or whole brain disorder
Property disease, electroencephalogram contains cerebral function information abundant, has very high reference value to epileptic condition diagnosis.It is examined in tradition
During disconnected, doctor needs to collect patient one day or more days eeg datas, and a large amount of eeg data makes medical worker's labor
Fatigue resistance increases, and detection efficiency reduces, and medical worker may be by interference caused by subjective factors, and there are the different disadvantages of check criteria
End.Therefore, particularly important are become to the intelligent diagnostics of epileptic condition.Domestic and foreign scholars make epileptic condition diagnosis at present
It researchs and analyses, but there are still class categories few, the low problem of classification accuracy.
For above-mentioned phenomenon, the applicant devises high-precision convolutional neural networks-Recognition with Recurrent Neural Network-supporting vector
The disease identification algorithm of machine mixed model.This algorithm is for the follow-up data processing of non-invasive detection system and identification process.Experiment
Show that eeg data effectively can be categorized into healthy phase, epileptic attack phase, and classification accuracy by method used herein
It has a distinct increment.
[summary of the invention]
The invention patent is in view of the above-mentioned drawbacks of the prior art, devise convolutional neural networks-Recognition with Recurrent Neural Network-
The disease identification algorithm of support vector machines mixed model.The invention patent is for the first time by machine learning and deep learning connected applications
Non-invasive diagnosis field is detected in brain, system can realize the classification of Disease and healthy population well, and carry out high-precision
Disease detection.
For the efficient diagnosis for realizing disease, the invention patent is proposed based on convolutional neural networks-Recognition with Recurrent Neural Network-
Hold the algorithm for pattern recognition of vector machine mixed model.The characteristic extraction procedure of this algorithm is end-to-end system, Recognition with Recurrent Neural Network
Recognition with Recurrent Neural Network RNN is selected, it is characterized in that the described method comprises the following steps, as shown in Figure 1:
Step 1: eeg data being acquired using portable brain electric signal acquisition method；
Step 2: the original eeg data input convolutional neural networks CNN of acquisition carries out the extraction of space characteristics.
Step 3: the original eeg data input Recognition with Recurrent Neural Network RNN of acquisition carries out the extraction of temporal characteristics.
Step 4: space characteristics and temporal characteristics being inputed into support vector machines together and carry out final classification, are obtained most
Whole diagnostic result.
The step 1 the following steps are included:
Step 1.1: data acquisition: eeg data being acquired using portable brain electric signal acquisition method；
Step 1.2: data prediction: eeg data filtering, be filtered using small echo, to EEG signals carry out decompose and
Reconstruct, obtains the information of time domain.
The step 2 the following steps are included:
Step 2.1: entering data by two layers of convolutional layer, two layers of maximum pond layer, one layer of feature flattening layer and one
The CNN network that the full articulamentum of layer is constituted；
Step 2.2: being used as space characteristics collection F1 [f for the feature extraction of input feature vector flattening layer is come out11,f12…
f1n]。
The step 3 the following steps are included:
Step 3.1: entering data into three layers of RNN network；
Step 3.2: being used as temporal characteristics collection F2 [f for the feature extraction of input feature vector flattening layer is come out21,f22…
f2n]。
The step 4 the following steps are included:
Step 4.1: inputting support vector machines after space characteristics F1 and temporal characteristics F2 is merged；
Step 4.2: the diagnostic result of final data classification, output illness or health is carried out by SVM.
The present invention extracts further feature using deep learning parallel, then conventional machines study is used to carry out most as strong classifier
Whole medical diagnosis on disease.The defect that original diagnosis algorithm is easy based on small sample over-fitting is not only overcome using the present invention, and
Accuracy rate, sensitivity and specificity are improved to 94% or more simultaneously.
[Detailed description of the invention]
Fig. 1 algorithm flow chart
Fig. 2 algorithm realization procedure chart
Fig. 3 algorithm training process figure
[specific embodiment]
With reference to the accompanying drawing, the implementation process for method that the present invention will be described in detail.It should be emphasized that following the description is only
Illustratively, the range and its application being not intended to be limiting of the invention.
Eeg data is acquired using portable brain electric signal acquisition method in this patent, recycles data analysing method pair
Sample data carries out identification classification.CNN-RNN-SVM algorithm is used herein, first with CNN and RNN to temporal characteristics and space characteristics
It extracts, classifies together to SVM after then merging feature, obtain the result of high-accuracy, sensitivity and specificity.
Fig. 2 is algorithm implementation figure, and the present invention includes the following steps:
Step 1: carrying out brain wave acquisition using portable brain electric signal acquisition method and obtain sample data, data carry out data
Pretreatment, obtains initial history set of data samples D；
Step 2: using the subset in initial history set of data samples D as training set, inputting convolutional neural networks CNN and carry out
Trained parameter inside preservation model after training；
Step 3: using the subset in initial history set of data samples D as training set, inputting Recognition with Recurrent Neural Network RNN and carry out
Trained parameter inside preservation model after training；
Step 4: training sample data being re-entered after CNN and RNN after extracting further feature and input SVM progress mould
The training of type inner parameter, and preservation model inner parameter；
Step 5: sample data unbred in historical data is inputted into CNN and RNN and difference as test set respectively
Its further feature is extracted, space characteristics and temporal characteristics are obtained.Space characteristics and temporal characteristics are inputed into supporting vector together
Machine SVM carries out final classification, obtains last diagnostic result.
Signal is acquired first, and eeg data is inputted in non-invasive detection system；
Secondly, execute step 1.2 pair acquisition data be filtered, be filtered using small echo, to EEG signals into
Row decomposes and reconstruct, obtains the information of time domain.
Then data input convolutional neural networks and Recognition with Recurrent Neural Network.Unlike conventional machines learning method, volume
Product neural network and Recognition with Recurrent Neural Network are a systems end to end, can automatically extract to feature, not need time-consuming consumption
The manual feature extraction process of power.Neural network is adjusted inner parameter by backpropagation and gradient decline, so that most
Termination fruit can achieve optimal.
The training of CNN mainly comprises the steps that
Step 2.1: by the CNN network that training data inputs two layers of convolutional network and two layers of fully-connected network is constituted；
Step 2.2: network is declined by backpropagation and gradient, is adjusted according to label value to inner parameter, is made mould
Type is preferably fitted and outputs and inputs.
Step 2.3: the CNN model after training is saved.
The training of RNN mainly comprises the steps that
Step 3.1: training data is inputted to three layers of RNN network；
Step 3.2: network is declined by backpropagation and gradient, is adjusted according to label value to inner parameter, is made mould
Type is preferably fitted and outputs and inputs；
Step 3.3: the RNN model after training is saved.
It is since CNN and RNN feature extractor has trained, the sample of training set is again defeated when being trained to SVM
Enter two networks, and feature extraction is come out and is classified again to SVM.
The training of SVM mainly comprises the steps that
Step 4.1: the CNN model that trained data input is saved will propose the feature of input feature vector flattening layer
It takes out as space characteristics collection F1 [f11,f12…f1n]；
Step 4.2: the RNN model that trained data input is saved will propose the feature of input feature vector flattening layer
It takes out as temporal characteristics collection F2 [f21,f22…f2n]；
Step 4.3: input SVM carries out the training of SVM parameter after space characteristics F1 and temporal characteristics F2 is merged.
Finally, test data input model is verified its accuracy rate, sensitivity and specificity when model training is good.Most
Whole model can be used for actual medical diagnosis on disease after saving.
Step 5.1: the CNN model that the input of untrained data is saved, it will be i.e. by the feature of input feature vector flattening layer
It extracts as space characteristics collection F1 [f11,f12…f1n]；
Step 5.2: the RNN model that the input of untrained data is saved, it will be i.e. by the feature of input feature vector flattening layer
It extracts as temporal characteristics collection F2 [f21,f22…f2n]；
Step 5.3: input SVM carries out final prediction after space characteristics F1 and temporal characteristics F2 is merged.Algorithm training
Figure is as shown in Figure 3.
The assessment result of final mask is as shown in table 1, and accuracy rate, sensitivity, specificity and F1 score are above 94%.F1
Score is a kind of index for being used to measure two disaggregated model accuracy in statistics.It has combined the accurate of disaggregated model
Rate and recall rate.Therefore, various indexs are integrated, the algorithm in this patent possesses preferable practical prospect and generalization ability.
1 model evaluation result of table
It is led it should be noted that above embodiments are only to illustrate the invention patent mixed model algorithm in brain electricity non-invasive diagnosis
The explanation of domain application, rather than the restriction to the invention patent.Those skilled in the art should understand that can be to originally setting
The technical solution of meter is modified or replaced equivalently, and without departing from the spirit and scope of the technical solution of the present invention, should all be contained
Lid is in the scope of the claims of the present invention.
Claims (8)
1. the present invention is to be based on convolutional neural networks-Recognition with Recurrent Neural Network-support vector machines mixed model in non-invasive detection system
Algorithm for pattern recognition, feature, which is set, the described method comprises the following steps:
Step 1: brain electricity being acquired using portable brain electric signal acquisition method and data prediction, obtain initial history number
According to sample set D；
Step 2: initial history set of data samples D input convolutional neural networks CNN is trained rear preservation model parameter.
Step 3: initial history set of data samples D input Recognition with Recurrent Neural Network RNN is trained rear preservation model parameter.
Step 4: training set data being re-entered after CNN and RNN to input SVM after extracting further feature and is trained；
Step 5: unbred sample data being inputted into CNN and RNN respectively and extracts its further feature respectively, obtains space spy
It seeks peace temporal characteristics.Space characteristics and temporal characteristics are inputed into support vector machines together and carry out final classification, are obtained final
Diagnostic result.
2. according to claim 1 based on convolutional neural networks-Recognition with Recurrent Neural Network-support vector machines mixed model disease
The algorithm for pattern recognition of non-invasive diagnosis carries out pattern-recognition it is characterized in that applying in non-invasive diagnosis.
3. according to claim 1 based on convolutional neural networks-Recognition with Recurrent Neural Network-support vector machines mixed model disease
The algorithm for pattern recognition of non-invasive diagnosis, it is characterized in that the step include the following steps 1 the following steps are included:
Step 1.1: data acquisition data acquisition: being carried out to brain electricity using portable brain electric signal acquisition method；
Step 1.2: data prediction: the response data of sensor constitutes initial history set of data samples D by wavelet filtering.
4. according to claim 1 based on convolutional neural networks-Recognition with Recurrent Neural Network-support vector machines mixed model disease
The algorithm for pattern recognition of non-invasive diagnosis, it is characterized in that the step includes with Noise Elimination from Wavelet Transform in the step 1.2, to brain
Electric signal is decomposed and is reconstructed, and time-domain information is obtained.
5. according to claim 1 based on convolutional neural networks-Recognition with Recurrent Neural Network-support vector machines mixed model disease
The algorithm for pattern recognition of non-invasive diagnosis, it is characterized in that the step 2 the following steps are included:
Step 2.1: the CNN network that training data inputs two layers of convolutional network and two layers of fully-connected network composition is trained；
Step 2.2: the CNN model after training is saved.
6. according to claim 1 based on convolutional neural networks-Recognition with Recurrent Neural Network-support vector machines mixed model disease
Algorithm for pattern recognition, it is characterized in that the step 3 the following steps are included:
Step 3.1: the RNN network that training data inputs three layers is trained；
Step 3.2: the RNN model after training is saved.
7. according to claim 1 based on convolutional neural networks-Recognition with Recurrent Neural Network-support vector machines mixed model disease
Non-invasive diagnosis algorithm for pattern recognition, it is characterized in that the step 4 the following steps are included:
Step 4.1: the CNN model that trained data input is saved will go out the feature extraction of input feature vector flattening layer
As space characteristics collection F1 [f11, f12 ... f1n]；
Step 4.2: the RNN model that trained data input is saved will go out the feature extraction of input feature vector flattening layer
As temporal characteristics collection F2 [f21, f22 ... f2n]；
Step 4.3: input SVM carries out the training of parameter after space characteristics F1 and temporal characteristics F2 is merged.
8. according to claim 1 based on convolutional neural networks-Recognition with Recurrent Neural Network-support vector machines mixed model disease
Non-invasive diagnosis algorithm for pattern recognition, it is characterized in that the step 5 the following steps are included:
Step 5.1: the CNN model that the input of untrained data is saved, it will be i.e. by the feature extraction of input feature vector flattening layer
It is used as space characteristics collection F1 [f11, f12 ... f1n] out；
Step 5.2: the RNN model that the input of untrained data is saved, it will be i.e. by the feature extraction of input feature vector flattening layer
It is used as temporal characteristics collection F2 [f21, f22 ... f2n] out；
Step 5.3: input SVM carries out final prediction after space characteristics F1 and temporal characteristics F2 is merged.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910591947.7A CN110200624A (en) | 2019-07-02 | 2019-07-02 | Based on convolutional neural networks-Recognition with Recurrent Neural Network-support vector machines mixed model disease identification algorithm |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910591947.7A CN110200624A (en) | 2019-07-02 | 2019-07-02 | Based on convolutional neural networks-Recognition with Recurrent Neural Network-support vector machines mixed model disease identification algorithm |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110200624A true CN110200624A (en) | 2019-09-06 |
Family
ID=67795905
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910591947.7A Pending CN110200624A (en) | 2019-07-02 | 2019-07-02 | Based on convolutional neural networks-Recognition with Recurrent Neural Network-support vector machines mixed model disease identification algorithm |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110200624A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111933290A (en) * | 2020-08-14 | 2020-11-13 | 苏州赫亚斯顿智能科技有限公司 | Method and device for establishing artificial reproduction pregnancy prediction by machine learning model |
| CN112992356A (en) * | 2021-03-30 | 2021-06-18 | 太原理工大学 | Heart failure prediction method and device based on convolutional layer feature rearrangement and SVM |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106901723A (en) * | 2017-04-20 | 2017-06-30 | 济南浪潮高新科技投资发展有限公司 | A kind of electrocardiographic abnormality automatic diagnosis method |
| CN106909784A (en) * | 2017-02-24 | 2017-06-30 | 天津大学 | Epileptic electroencephalogram (eeg) recognition methods based on two-dimentional time-frequency image depth convolutional neural networks |
| CN109614885A (en) * | 2018-11-21 | 2019-04-12 | 齐鲁工业大学 | A kind of EEG signals Fast Classification recognition methods based on LSTM |
| US10304208B1 (en) * | 2018-02-12 | 2019-05-28 | Avodah Labs, Inc. | Automated gesture identification using neural networks |
-
2019
- 2019-07-02 CN CN201910591947.7A patent/CN110200624A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106909784A (en) * | 2017-02-24 | 2017-06-30 | 天津大学 | Epileptic electroencephalogram (eeg) recognition methods based on two-dimentional time-frequency image depth convolutional neural networks |
| CN106901723A (en) * | 2017-04-20 | 2017-06-30 | 济南浪潮高新科技投资发展有限公司 | A kind of electrocardiographic abnormality automatic diagnosis method |
| US10304208B1 (en) * | 2018-02-12 | 2019-05-28 | Avodah Labs, Inc. | Automated gesture identification using neural networks |
| CN109614885A (en) * | 2018-11-21 | 2019-04-12 | 齐鲁工业大学 | A kind of EEG signals Fast Classification recognition methods based on LSTM |
Non-Patent Citations (1)
| Title |
|---|
| MOHAMAD SHAHBAZI1 等: "A GENERALIZABLE MODEL FOR SEIZURE PREDICTION BASED ON DEEP LEARNING", 《 2018 IEEE GLOBAL CONFERENCE ON SIGNAL AND INFORMATION PROCESSING》 * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111933290A (en) * | 2020-08-14 | 2020-11-13 | 苏州赫亚斯顿智能科技有限公司 | Method and device for establishing artificial reproduction pregnancy prediction by machine learning model |
| CN111933290B (en) * | 2020-08-14 | 2023-10-10 | 北京赫雅智能科技有限公司 | Method and device for predicting artificial reproduction conception by machine learning model |
| CN112992356A (en) * | 2021-03-30 | 2021-06-18 | 太原理工大学 | Heart failure prediction method and device based on convolutional layer feature rearrangement and SVM |
| CN112992356B (en) * | 2021-03-30 | 2022-04-26 | 太原理工大学 | Heart failure prediction method and device based on convolutional layer feature rearrangement and SVM |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Aslan et al. | Automatic Detection of Schizophrenia by Applying Deep Learning over Spectrogram Images of EEG Signals. | |
| Peng et al. | Automatic epileptic seizure detection via Stein kernel-based sparse representation | |
| Poulos, M. Rangoussi, N. Alexandris, A. Evangelou | On the use of EEG features towards person identification via neural networks | |
| CN113786204A (en) | Epilepsia intracranial electroencephalogram early warning method based on deep convolution attention network | |
| Gruszczyńska et al. | Epilepsy identification based on EEG signal using RQA method | |
| CN111449644A (en) | Bioelectricity signal classification method based on time-frequency transformation and data enhancement technology | |
| Lopes et al. | Automatic electroencephalogram artifact removal using deep convolutional neural networks | |
| Yildiz et al. | Classification and analysis of epileptic EEG recordings using convolutional neural network and class activation mapping | |
| Djamal et al. | Identification of post-stroke EEG signal using wavelet and convolutional neural networks | |
| CN110321783A (en) | A kind of MEG spike detection method and system based on 1D convolutional neural networks | |
| Tawhid et al. | Automatic and efficient framework for identifying multiple neurological disorders from EEG signals | |
| CN114532993B (en) | Automatic detection method for electroencephalogram high-frequency oscillation signals of epileptic patients | |
| CN112674782B (en) | Device and method for detecting epileptic-like electrical activity of epileptic during inter-seizure period | |
| Zeng et al. | GRP-DNet: A gray recurrence plot-based densely connected convolutional network for classification of epileptiform EEG | |
| Prasanth et al. | Deep learning for interictal epileptiform spike detection from scalp EEG frequency sub bands | |
| CN113723557B (en) | Depression brain electricity classifying system based on multiband space-time convolution network | |
| Mustafa et al. | The analysis of eeg spectrogram image for brainwave balancing application using ann | |
| CN110289097A (en) | A kind of Pattern Recognition Diagnosis system stacking model based on Xgboost neural network | |
| CN110200624A (en) | Based on convolutional neural networks-Recognition with Recurrent Neural Network-support vector machines mixed model disease identification algorithm | |
| CN114595725B (en) | Electroencephalogram signal classification method based on addition network and supervised contrast learning | |
| CN107045624B (en) | Electroencephalogram signal preprocessing and classifying method based on maximum weighted cluster | |
| CN111067513A (en) | Sleep quality detection key brain area judgment method based on characteristic weight self-learning | |
| CN114176610A (en) | Workload assessment method for diagnosis of mild cognitive dysfunction patient | |
| Wang et al. | One and two dimensional convolutional neural networks for seizure detection using eeg signals | |
| Pan et al. | Epileptic seizure detection with hybrid time-frequency EEG input: a deep learning approach |
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 | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20190906 |
|
| WD01 | Invention patent application deemed withdrawn after publication |