CN108962391A - Epileptics prediction technique early period based on wavelet packet character and random forest - Google Patents

Epileptics prediction technique early period based on wavelet packet character and random forest Download PDF

Info

Publication number
CN108962391A
CN108962391A CN201810409965.4A CN201810409965A CN108962391A CN 108962391 A CN108962391 A CN 108962391A CN 201810409965 A CN201810409965 A CN 201810409965A CN 108962391 A CN108962391 A CN 108962391A
Authority
CN
China
Prior art keywords
entropy
wavelet packet
feature
epileptics
random forest
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
Application number
CN201810409965.4A
Other languages
Chinese (zh)
Inventor
曹九稳
王玉星
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Hangzhou Dianzi University
Original Assignee
Hangzhou Dianzi University
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Hangzhou Dianzi University filed Critical Hangzhou Dianzi University
Priority to CN201810409965.4A priority Critical patent/CN108962391A/en
Publication of CN108962391A publication Critical patent/CN108962391A/en
Pending legal-status Critical Current

Links

Classifications

    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
    • G16H50/50—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for simulation or modelling of medical disorders
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • G06F18/243—Classification techniques relating to the number of classes
    • G06F18/24323—Tree-organised classifiers

Landscapes

  • Engineering & Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Public Health (AREA)
  • Medical Informatics (AREA)
  • Health & Medical Sciences (AREA)
  • Theoretical Computer Science (AREA)
  • General Engineering & Computer Science (AREA)
  • Databases & Information Systems (AREA)
  • Evolutionary Computation (AREA)
  • Physics & Mathematics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • General Physics & Mathematics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Biomedical Technology (AREA)
  • Evolutionary Biology (AREA)
  • Artificial Intelligence (AREA)
  • Pathology (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Epidemiology (AREA)
  • General Health & Medical Sciences (AREA)
  • Primary Health Care (AREA)
  • Data Exchanges In Wide-Area Networks (AREA)

Abstract

Epileptics prediction technique early period based on wavelet packet character and random forest that the invention discloses a kind of.The present invention includes the following steps: step 1, original signal WAVELET PACKET DECOMPOSITION；Step 2 extracts energy accounting, Wavelet Packet Entropy feature to wavelet packet coefficient；Step 3 carries out Classification and Identification with random forests algorithm.The present invention combines wavelet packet character and random forests algorithm, analyzes the EEG signals of epileptics different times, can more accurately identify the EEG signals of morbidity early period, and the principle of entire method understands that forecasting efficiency is high.Also there is certain reference value to the research of other diseases.

Description

Epileptics prediction technique early period based on wavelet packet character and random forest
Technical field
The invention belongs to field of signal processing, and it is pre- early period to be related to a kind of epileptics based on wavelet packet character and random forest Survey method.
Background technique
Epilepsy be by cerebral neuron group's high level of synchronization paradoxical discharge caused by one group of disease, often can repeatedly, suddenly Breaking-out, seriously affects the life and work of patient.Epileptic accounts about the 0.5%-2.5% of global total population at present, he In can either be performed the operation by the drug method of excision epileptogenic focus of some patientss treated, other part is suffered from There are no suitable methods to be treated by person.For this some patients, best bet is exactly in epileptic seizure the last period Between predict epilepsy and will break out, then take some prevention & protection measures to keep away by doctor or patient oneself in a short time Exempt from secondary injury.
Traditional epileptics prediction technique is that the EEG signals of epileptic patient are divided into early period, interphase and three classes stage of attack, Statistics feature is extracted to three classes EEG signals, is classified with neural network, there are following 2 disadvantages for this method:
1. the classification of pair epileptics early period is simple, it cannot achieve the purpose that epilepsy disease forecasting well.
2. statistics feature can only reflect the distribution situation of signal totality, reflect it is not very sensitive to the variation of signal, only Have when significant change occurs for signal, the distribution of feature can just change therewith.
Summary of the invention
The present invention is directed to the deficiency of traditional epileptic prediction scheme, proposes a kind of based on wavelet packet character and random forest Epileptics prediction technique early period.Technical solution of the present invention mainly includes the following steps:
Step 1, original signal WAVELET PACKET DECOMPOSITION；
Step 2 extracts energy accounting, Wavelet Packet Entropy feature to wavelet packet coefficient；
Step 3 carries out Classification and Identification with random forests algorithm；
The specific implementation of the step 1 including the following steps:
Original EEG signals are divided into 5 classifications by 1-1., and starting morbidity to morbidity end is stage of attack (ictal)； The data of four hours are as interictal (inter) before and after onset；20 minutes before the onset are first predictions to morbidity is started Phase (pre-ictalA)；Breaking-out preceding 40 first 20 minutes to breaking-out are second time span of forecasts (pre-ictalB)；It breaks out previous small When to breaking-out preceding 40 minutes be third time span of forecast (pre-ictalC), in this way by EEG signals carry out division can make predict Time is more accurate, also to carry out following processing to the signal of every one kind before feature extraction:
The 1-2. present invention handles signal using frame, and each frame has 23 channels, when a length of 4s signal.To the every of each frame The signal in one channel carries out (db4) WAVELET PACKET DECOMPOSITION, and Decomposition order is 6 layers.
Step 1 needs to pay attention to: in view of the influence between frame shifting, having the frame of 2s to be overlapped in the 4s frame length chosen in 1-2.
Step 2 the realization process includes:
2-1. is because of the case where emphasis that we study is four species rhythm wave (α, β, θ, δ) in EEG signals, according to small echo Packet decomposes rule, and the 6th layer of preceding 15 frequency subbands contain these four rhythm and pace of moving things waves, so extracting for this 15 frequency subbands Following several features: Shannon entropy (shannon entropy), logarithmic entropy (log entropy), threshold value entropy (threshold Entropy), determine entropy (sure entropy), norm entropy (norm entropy)；
The specific acquiring method of Wavelet Packet Entropy is as follows:
Before being illustrated, entropy (E) must be following cost function:
With E (s)=0
Shannon entropy (shannon entropy):
Logarithmic entropy (log entropy):
Threshold value entropy (threshold entropy):
Threshold value is p, works as siWhen > p, E (si)=1, otherwise E (si)=0.
Determine entropy (sure entropy):
Norm entropy (norm entropy):
K is the order of norm.
This step will generate 15*5=75 Wavelet Packet Entropy feature.
2-2. seeks energy accounting feature to the 6th layer of frequency sub-band signals, can generate 64 subband energy ratio features.
E6=e (si)/sum(e(si))*100
e(si) be some subband energy.
2-3. can obtain 64+15*5=139 dimensional feature vector according to step 2-1 and 2-2, each channel, including The Wavelet Packet Entropy feature of the energy accounting feature of 64 subbands and 75 subbands.139*23=so is had for each sample The feature vector of 3197 dimensions.
Step 2 needs to pay attention to:
(1) in 2-3, since the energy accounting of signal has the characteristics that relatively, numerically, same layer wavelet packet frequency The summation of the energy accounting of band is 100, and for the integrality of energy, energy accounting is all frequency subbands of layer 6.
(2) in 2-4, each sample contains 23 channels, each channel is characterized in 139 dimensions, be here by The feature in each channel is grouped together into the feature of 3197 dimensions.
Step 3 the realization process includes:
N is set by the tree of random forest, carries out model training with the sample of random forests algorithm selection 80%；It is remaining 20% sample is used for model measurement.
Wherein, N is preferably 50；Because when N increases, although classification accuracy increases, the complexity of model Degree also increases, and comprehensively considers complexity and accuracy rate, and the quantity N selection 50 of tree is most suitable.Input when model training is energy Accounting is measured, Shannon entropy, threshold value entropy, determines entropy and several features of norm entropy at logarithmic entropy；
The present invention has the beneficial effect that:
New method proposed in this invention: the prediction side of epileptics early period based on wavelet packet character and random forest Method extracts wavelet packet character on the basis of WAVELET PACKET DECOMPOSITION, makes and classify of random forest, it is simple to the processing of EEG signals and Effectively, classify on identical sample basis, the classification accuracy of support vector machines (SVM) is 79.4%, k arest neighbors (K-NN) classification accuracy is 84%, and the classification accuracy of discriminant analysis is 78.7%, and the classification of random forest (RF) is accurate Rate is 86.4%, it can be seen that the classification accuracy of random forest is higher compared with traditional sorting algorithm.In general, Entirely the principle of prediction technique understands, high-efficient, 20min, 20min- before the onset of capable of accurately identifying in a short time The EEG signals of 40min, 40min-60min, when there is new sample to be entered model, model judges out which the sample belongs to A classification, and issue corresponding alarm.
The present invention proposes a kind of prediction technique of epileptics, wavelet packet character and random forests algorithm is combined, to epilepsy The EEG signals of sick different times are analyzed, and can more accurately identify the EEG signals of morbidity early period, and entire method Principle understand that forecasting efficiency is high.Also there is certain reference value to the research of other diseases.
Detailed description of the invention
The division mode of Fig. 1 initial data of the present invention；
The extraction of Fig. 2 wavelet packet character of the present invention and classification process figure；
The structure of Fig. 3 random forest of the present invention.
Specific embodiment
It elaborates with reference to the accompanying drawings and detailed description to the present invention.
As shown in Figure 1-3, epileptics prediction technique early period based on wavelet packet character and random forest, general is directed to hair The realization step of the Epileptic Prediction of state recognition, has detailed introduction, i.e., skill of the invention in summary of the invention before making Art scheme mainly includes the following steps:
Step 1, original signal WAVELET PACKET DECOMPOSITION.
Step 2 extracts energy accounting, Wavelet Packet Entropy feature to wavelet packet coefficient.
Step 3 carries out Classification and Identification with random forests algorithm.
The specific implementation of the step 1 including the following steps:
Original EEG signals are divided into 5 classifications by 1-1., and starting morbidity to morbidity end is stage of attack (ictal)； The data of four hours are as interictal (inter) before and after onset；20 minutes before the onset are first predictions to morbidity is started Phase (pre-ictalA)；Breaking-out preceding 40 first 20 minutes to breaking-out are second time span of forecasts (pre-ictalB)；It breaks out previous small When to breaking-out preceding 40 minutes be third time span of forecast (pre-ictalC), in this way by EEG signals carry out division can make predict Time is more accurate, also to carry out following processing to the signal of every one kind before feature extraction:
The present invention handles signal using frame, and each frame has 23 channels, when a length of 4s signal.To each of each frame The signal in channel carries out (db4) WAVELET PACKET DECOMPOSITION, and Decomposition order is 6 layers.
Step 1 needs to pay attention to: in view of the influence between frame shifting, having the frame of 2s to be overlapped in mono- frame of 4s chosen in 1-2.
Step 2 the realization process includes:
2-1. is because of the case where emphasis that we study is four species rhythm wave (α, β, θ, δ) in EEG signals, according to small echo Packet decomposes rule, and the 6th layer of preceding 15 frequency subbands contain these four rhythm and pace of moving things waves, so extracting for this 15 frequency subbands Following several features: Shannon entropy (shannon entropy), logarithmic entropy (log entropy), threshold value entropy (threshold Entropy), determine entropy (sure entropy), norm entropy (norm entropy), the specific acquiring method of Wavelet Packet Entropy is such as Under:
Before being illustrated, entropy (E) must be following cost function:
With E (s)=0
Shannon entropy (shannon entropy):
Logarithmic entropy (log entropy):
Threshold value entropy (threshold entropy):
Threshold value is p, as s (i) > p, E (si)=1, otherwise E (si)=0.
Determine entropy (sure entropy):
P is threshold value.
Norm entropy (norm entropy):
K is the order of norm.
This step can generate 15*5=75 Wavelet Packet Entropy feature.
2-2. seeks energy accounting feature to the 6th layer of frequency sub-band signals, can generate 64 subband energy ratio features.
E6=e (si)/sum(e(si))*100
e(si) be some subband energy.
2-3. can obtain 64+15*5=139 dimensional feature vector according to two step of front, each channel, including 64 sons The Wavelet Packet Entropy feature of the energy accounting feature of band and 75 subbands.139*23=3197 dimension so is had for each sample Feature vector.
Step 2 needs to pay attention to:
(3) in 2-3, since the energy accounting of signal has the characteristics that relatively, numerically, same layer wavelet packet frequency The summation of the energy accounting of band is 100, and for the integrality of energy, energy accounting is all frequency subbands of layer 6.
(4) in 2-4, each sample contains 23 channels, each channel is characterized in 139 dimensions, be here by The feature in each channel is grouped together into the feature of 3197 dimensions.
Step 3 the realization process includes:
3. being classified with Random Forest model.The tree of random forest is set as N, wherein the sample of selection 80% carries out model The sample of training, residue 20% is used for model measurement.
Wherein, N is preferably 50；Input when model training is energy accounting, Shannon entropy, logarithmic entropy, threshold value entropy, determination Entropy and several features of norm entropy；
In order to reach better epileptic seizure prediction effect, below by from practical application when parameter selection and design aspect Expansion is introduced, the reference to be used for other application as the invention:
In 1-2, the data of each frame 4s in order to make information between every two frame of data there is connection, and do not allow information It is excessive to repeat, so frame shifting is selected as 2s.
In 3, the tree of random forest is arranged to 50.Because the complexity of random forest can increase with the increase of tree Add, but the accuracy rate classified also will increase, and comprehensively considers complexity and predictablity rate, 50 trees are arranged.
The present invention proposes the prediction technique of epileptics early period a kind of, and wavelet packet character and random forests algorithm are combined, right The EEG signals of epileptics different times are analyzed, and can more accurately identify the EEG signals of morbidity early period, and entire The principle of method understands that forecasting efficiency is high.Also there is certain reference value to the research of other diseases.

Claims (3)

1. epileptics prediction technique early period based on wavelet packet character and random forest, which comprises the steps of:
Step 1, original signal WAVELET PACKET DECOMPOSITION；
Step 2 extracts energy accounting, Wavelet Packet Entropy feature to wavelet packet coefficient；
Step 3 carries out Classification and Identification with random forests algorithm；
The specific implementation of the step 1 includes the following steps:
Original EEG signals are divided into 5 classifications by 1-1., and starting morbidity to morbidity end is stage of attack；Four before and after onset The data of hour are as interictal；20 minutes before the onset are first time span of forecasts to morbidity is started；Arrive hair within breaking-out first 40 minutes Making first 20 minutes is second time span of forecast；Break out previous hour to breaking-out first 40 minutes be third time span of forecast, mentioned in feature Following processing is carried out to the signal of every one kind before taking:
Signal is handled using frame, each frame is that have 23 channels, when a length of 4s signal；To the letter in each channel of each frame Number carry out WAVELET PACKET DECOMPOSITION, Decomposition order be 6 layers；
Step 2 the realization process includes:
2-1. according to WAVELET PACKET DECOMPOSITION rule, the 6th layer of preceding 15 frequency subbands include four species rhythm wave α, β in EEG signals, θ,δ；So being extracted following several features for this 15 frequency subbands: Shannon entropy, threshold value entropy, determines entropy and model at logarithmic entropy Number entropy；
The specific acquiring method of Wavelet Packet Entropy is as follows:
Before being illustrated, entropy (E) must be following cost function:
With E (s)=0
Shannon entropy:
Logarithmic entropy:
Threshold value entropy:
Threshold value is p, works as siWhen > p, E (si)=1, otherwise E (si)=0；
Determine entropy:
Norm entropy:
K is the order of norm；
15*5=75 Wavelet Packet Entropy feature will be generated by step 2-1；
2-2. seeks energy accounting feature to the 6th layer of frequency sub-band signals, can generate 64 subband energy ratio features；
E6=e (si)/sum(e(si))*100
e(si) be some subband energy；
2-3. can obtain 64+15*5=139 dimensional feature vector according to step 2-1 and 2-2, each channel, including 64 sons The Wavelet Packet Entropy feature of the energy accounting feature of band and 75 subbands；Therefore 139*23=3197 dimension is had for each sample Feature vector；
Step 3 the realization process includes:
N is set by the tree of random forest, carries out model training with the sample of random forests algorithm selection 80%；Residue 20% Sample be used for model measurement.
2. epileptics prediction technique early period according to claim 1 based on wavelet packet character and random forest, feature It is in the 4s frame length chosen to have the frame of 2s to be overlapped in step 1-2.
3. epileptics prediction technique early period according to claim 1 based on wavelet packet character and random forest, feature It is that N is preferably 50；Input when model training is energy accounting: Shannon entropy, threshold value entropy, determines entropy and norm at logarithmic entropy Several features of entropy.
CN201810409965.4A 2018-05-02 2018-05-02 Epileptics prediction technique early period based on wavelet packet character and random forest Pending CN108962391A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201810409965.4A CN108962391A (en) 2018-05-02 2018-05-02 Epileptics prediction technique early period based on wavelet packet character and random forest

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201810409965.4A CN108962391A (en) 2018-05-02 2018-05-02 Epileptics prediction technique early period based on wavelet packet character and random forest

Publications (1)

Publication Number Publication Date
CN108962391A true CN108962391A (en) 2018-12-07

Family

ID=64498860

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201810409965.4A Pending CN108962391A (en) 2018-05-02 2018-05-02 Epileptics prediction technique early period based on wavelet packet character and random forest

Country Status (1)

Country Link
CN (1) CN108962391A (en)

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109497997A (en) * 2018-12-10 2019-03-22 杭州妞诺科技有限公司 Based on majority according to the seizure detection and early warning system of acquisition
CN109903852A (en) * 2019-01-18 2019-06-18 杭州电子科技大学 Based on the customized intelligent Epileptic Prediction of PCA-LDA
CN109992866A (en) * 2019-03-25 2019-07-09 新奥数能科技有限公司 Training method, device, readable medium and the electronic equipment of load forecasting model
CN110811609A (en) * 2019-10-28 2020-02-21 杭州电子科技大学 Intelligent epileptic spike detection method based on fusion of adaptive template matching and machine learning algorithm

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101352337A (en) * 2007-12-25 2009-01-28 天津大学 Method for capturing signals and extracting characteristics of stand imagination action brain wave
CN101751563A (en) * 2010-01-21 2010-06-23 上海大学 EEG signal identification method based on LabWindows/CVI and Matlab hybrid programming
US7860561B1 (en) * 2004-06-04 2010-12-28 Cleveland Medical Devices Inc. Method of quantifying a subject's wake or sleep state and system for measuring
CN106919956A (en) * 2017-03-09 2017-07-04 温州大学 Brain wave age forecasting system based on random forest
CN107274086A (en) * 2017-06-09 2017-10-20 东莞市毅豪电子科技有限公司 A kind of gridding governance information approach based on hidden Markov model

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US7860561B1 (en) * 2004-06-04 2010-12-28 Cleveland Medical Devices Inc. Method of quantifying a subject's wake or sleep state and system for measuring
CN101352337A (en) * 2007-12-25 2009-01-28 天津大学 Method for capturing signals and extracting characteristics of stand imagination action brain wave
CN101751563A (en) * 2010-01-21 2010-06-23 上海大学 EEG signal identification method based on LabWindows/CVI and Matlab hybrid programming
CN106919956A (en) * 2017-03-09 2017-07-04 温州大学 Brain wave age forecasting system based on random forest
CN107274086A (en) * 2017-06-09 2017-10-20 东莞市毅豪电子科技有限公司 A kind of gridding governance information approach based on hidden Markov model

Non-Patent Citations (3)

* Cited by examiner, † Cited by third party
Title
崔洁: "基于改进多元多尺度熵的癫痫脑电信号自动分类", 《中国优秀硕士学位论文全文数据库（医药卫生科技辑）》 *
朱俊敏 等: "小波域音频信号降噪研究", 《计算机工程与应用》 *
李品祎: "基于便携式采集设备脑电信号的疲劳检测算法研究", 《中国优秀硕士学位论文全文数据库（医药卫生科技辑）》 *

Cited By (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109497997A (en) * 2018-12-10 2019-03-22 杭州妞诺科技有限公司 Based on majority according to the seizure detection and early warning system of acquisition
CN109903852A (en) * 2019-01-18 2019-06-18 杭州电子科技大学 Based on the customized intelligent Epileptic Prediction of PCA-LDA
CN109992866A (en) * 2019-03-25 2019-07-09 新奥数能科技有限公司 Training method, device, readable medium and the electronic equipment of load forecasting model
CN109992866B (en) * 2019-03-25 2022-11-29 新奥数能科技有限公司 Training method and device of load prediction model, readable medium and electronic equipment
CN110811609A (en) * 2019-10-28 2020-02-21 杭州电子科技大学 Intelligent epileptic spike detection method based on fusion of adaptive template matching and machine learning algorithm
CN110811609B (en) * 2019-10-28 2022-11-22 杭州电子科技大学 Epileptic spike intelligent detection device based on self-adaptive template matching and machine learning algorithm fusion

Similar Documents

Publication Publication Date Title
CN108962391A (en) Epileptics prediction technique early period based on wavelet packet character and random forest
CN110693493B (en) Epilepsia electroencephalogram prediction feature extraction method based on convolution and recurrent neural network combined time multiscale
Kumar et al. Classification of seizure and seizure-free EEG signals using local binary patterns
Mert et al. Emotion recognition based on time–frequency distribution of EEG signals using multivariate synchrosqueezing transform
Zandi et al. Automated real-time epileptic seizure detection in scalp EEG recordings using an algorithm based on wavelet packet transform
Song et al. Automatic recognition of epileptic EEG patterns via extreme learning machine and multiresolution feature extraction
Martis et al. Epileptic EEG classification using nonlinear parameters on different frequency bands
CN111931576A (en) Epileptic signal classification method based on EWT and improved CSP
Gupta et al. Classification of focal EEG signals using FBSE based flexible time-frequency coverage wavelet transform
Ilias et al. Multimodal detection of epilepsy with deep neural networks
CN112200016A (en) Electroencephalogram signal emotion recognition based on ensemble learning method AdaBoost
Hassanpour et al. Time–frequency based newborn EEG seizure detection using low and high frequency signatures
Kumar et al. Classification of seizure and seizure-free EEG signals using multi-level local patterns
Bavkar et al. Optimal EEG channels selection for alcoholism screening using EMD domain statistical features and harmony search algorithm
Ammar et al. Seizure detection with single-channel EEG using Extreme Learning Machine
Abbaszadeh et al. Probabilistic prediction of Epileptic Seizures using SVM
Zhou et al. Epileptic seizure detection via logarithmic normalized functional values of singular values
Acharjee et al. Multiclass epileptic seizure classification using time-frequency analysis of EEG signals
Ebrahimpour et al. Epileptic seizure detection using a neural network ensemble method and wavelet transform
CN112800928B (en) Epileptic seizure prediction method of global self-attention residual error network integrating channel and spectrum characteristics
Qatmh et al. Detection of epileptic seizure using discrete wavelet transform on gamma band and artificial neural network
Choubey et al. Classification of healthy, inter-ictal and seizure signal using various classification techniques.
Karim et al. Identification of seizure from single channel EEG using Support Vector Machine & Hilbert Vibration Decomposition
Mandal et al. Majority voting-based hybrid feature selection in machine learning paradigm for epilepsy detection using EEG
Tsipouras Uterine EMG Signals Spectral Analysis for Pre-Term Birth Prediction.

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
RJ01 Rejection of invention patent application after publication

Application publication date: 20181207

RJ01 Rejection of invention patent application after publication