阅读说明：本技术 一种基于睡眠脑网络的抑郁症患者数据分类方法及装置 (Depression patient data classification method and device based on sleep brain network ) 是由 罗语溪 汪婷婷 连佳铠 张仰婷 于 2021-03-05 设计创作，主要内容包括：本发明公开了一种基于睡眠脑网络的抑郁症患者数据分类方法及装置,所述方法包括：获取待检测对象的睡眠脑网络的脑电数据,所述脑电数据包括若干导脑电图通道数据；采用WPLI算法对所述若干导脑电图通道数据进行相位同步计算,得到相位特征集；采用预设的筛选算法对所述相位特征集进行筛选,得到筛选特征集；对所述筛选特征集进行二分类检测,得到检测结果。本发明可以对多导联的脑电信号进行同步计算,减少信号干扰,从而有效评估抑郁症患者,大大检测的准确率。(The invention discloses a method and a device for classifying depression patient data based on a sleep brain network, wherein the method comprises the following steps: acquiring electroencephalogram data of a sleeping brain network of a to-be-detected object, wherein the electroencephalogram data comprises a plurality of electroencephalogram channel data; performing phase synchronization calculation on the plurality of electroencephalogram channel data by adopting a WPLI algorithm to obtain a phase characteristic set; screening the phase characteristic set by adopting a preset screening algorithm to obtain a screening characteristic set; and performing two-classification detection on the screening characteristic set to obtain a detection result. The invention can synchronously calculate the multi-lead EEG signals and reduce signal interference, thereby effectively evaluating depression patients and greatly improving the detection accuracy.)

1. A method for classifying data of a depressed patient based on a sleep brain network, the method comprising:

acquiring electroencephalogram data of a sleeping brain network of a to-be-detected object, wherein the electroencephalogram data comprises a plurality of electroencephalogram channel data;

performing phase synchronization calculation on the plurality of electroencephalogram channel data by adopting a WPLI algorithm to obtain a phase characteristic set;

screening the phase characteristic set by adopting a preset screening algorithm to obtain a screening characteristic set;

and performing two-classification detection on the screening characteristic set to obtain a classification result.

2. The sleep brain network-based depression patient data classification method according to claim 1, wherein the performing phase synchronization calculation on the plurality of electroencephalogram channel data by using WPLI algorithm to obtain a set of phase characteristics includes:

calculating channel phase differences between every two pieces of the electroencephalogram channel data by adopting a WPLI algorithm to obtain a plurality of phase difference values;

respectively calculating sine values corresponding to the plurality of phase difference values to obtain a plurality of sine values;

and generating a phase characteristic set by the plurality of sine value sets.

3. The sleep brain network-based depression patient data classification method according to claim 2, characterized in that the WPLI algorithm is calculated as follows:

wherein WPLI_i，j，τBetween 0 and 1, E { } is a desired value operator, Δ ω_i，j，τIs the phase difference between nodes i and j, Δ ω_i，j，τIs calculated as follows:

Δω_i，j，τ＝ω_i(τ)-ω_j(τ)。

4. the sleep brain network-based data classification method for the patients with depression according to claim 1, wherein the screening of the phase feature set by using a preset screening algorithm to obtain a screening feature set comprises:

evaluating each phase characteristic value of the phase characteristic set by adopting a preset screening algorithm to obtain a plurality of evaluation characteristic values;

extracting N evaluation characteristic values from the plurality of evaluation characteristic values according to a preset percentage, and generating a screening characteristic set from the N evaluation characteristic value sets, wherein N is a positive integer greater than or equal to 1.

5. The sleep brain network-based data classification method for the depression patients according to claim 1, wherein the classifying detecting the screening feature set for two classes to obtain the classification result comprises:

and inputting the screening feature set into a preset support vector machine, and performing two-classification detection on the screening feature set by using a preset linear function, a preset polynomial function and a preset Gaussian function by using the preset support vector machine to obtain a diseased classification result or a normal classification result.

6. The sleep brain network-based depressive patient data classification method according to claim 4, further comprising:

randomly extracting N-1 evaluation characteristic values from the N evaluation characteristic values, and combining the N-1 evaluation characteristic values into a characteristic training set;

carrying out classification verification on the non-extracted evaluation characteristic values by adopting the characteristic training set to obtain verification accurate values;

respectively calculating verification accurate values corresponding to each evaluation characteristic value to obtain N verification accurate values;

and calculating the average value of the N verification accurate values to obtain a verification average value.

7. The method for classifying data of depression patients based on sleep brain network as claimed in claim 1, wherein the acquiring of electroencephalogram data of sleep brain network of the subject to be detected comprises:

acquiring sleep electroencephalogram data of a sleep brain network of a to-be-detected object, wherein the sleep electroencephalogram data comprises M (M) electroencephalogram channel data of a patient and M electroencephalogram channel data of a normal user, and M is a positive integer greater than or equal to 1;

extracting abnormal data from each electroencephalogram channel data, and deleting the abnormal data in each electroencephalogram channel data to obtain a plurality of clear channel data;

and filtering the data of the plurality of leading clear channels by adopting a preset filter to obtain the electroencephalogram data of the sleeping brain network of the object to be detected.

8. A depressed patient data classification apparatus based on a sleep brain network, the apparatus comprising:

the acquisition module is used for acquiring electroencephalogram data of a sleeping brain network of a to-be-detected object, wherein the electroencephalogram data comprises a plurality of electroencephalogram channel data;

the phase calculation module is used for performing phase synchronization calculation on the plurality of electroencephalogram channel data by adopting a WPLI algorithm to obtain a phase characteristic set;

the screening module is used for screening the phase characteristic set by adopting a preset screening algorithm to obtain a screening characteristic set;

and the detection module is used for carrying out two-classification detection on the screening characteristic set to obtain a classification result.

9. An electronic device, comprising: memory, processor and computer program stored on the memory and executable on the processor, characterized in that the processor when executing the program implements a sleep brain network based data classification method for depression patients according to any one of claims 1 to 7.

10. A computer-readable storage medium having stored thereon computer-executable instructions for causing a computer to perform the sleep brain network based depression patient data classification method according to any one of claims 1 to 7.

Technical Field

The invention relates to the technical field of medical assistance, in particular to a depression patient data classification method and device based on a sleep brain network.

Background

The most common depressive disorder in depression (depression) is characterized primarily clinically by a marked and persistent depression in the mood. The sufferers are not happy, unhappy and pleasurable and have reduced interest if the sufferers are light, and the sufferers are pessimistic and desperate if the sufferers are serious, and the sufferers spend the same day as the year, so the symptoms not only influence the daily life of the sufferers, but also increase the life pressure of the relatives and friends of the sufferers.

In order to determine the diseased condition of a patient early, two detection methods are commonly used at present, the first method is to record the behavior and symptoms of the patient within a period of time and then perform detection judgment by combining the current environment of the patient. The second method is to collect the brain electrical signal of the patient, determine the working state of the brain of the patient according to the brain electrical signal, and then carry out detection and judgment according to the working state.

However, the conventional detection method has the following problems: the detection and judgment are carried out through behaviors and environments, and symptoms in a period of time need to be recorded, so that the detection period is long and the detection efficiency is low; the electroencephalogram signals are adopted for detection, the data volume to be processed is large, the redundancy is large, the data processing is difficult, the processing time is long, and the detection accuracy is low.

Disclosure of Invention

The invention provides a depression patient data classification method and device based on a sleep brain network.

A first aspect of an embodiment of the present invention provides a method for classifying data of a depression patient based on a sleep brain network, the method including:

acquiring electroencephalogram data of a sleeping brain network of a to-be-detected object, wherein the electroencephalogram data comprises a plurality of electroencephalogram channel data:

performing phase synchronization calculation on the plurality of electroencephalogram channel data by adopting a WPLI algorithm to obtain a phase characteristic set;

screening the phase characteristic set by adopting a preset screening algorithm to obtain a screening characteristic set;

and performing two-classification detection on the screening characteristic set to obtain a classification result.

In a possible implementation manner of the first aspect, the performing phase synchronization calculation on the plurality of electroencephalogram channel data by using a WPLI algorithm to obtain a set of phase characteristics includes:

calculating channel phase differences between every two pieces of the electroencephalogram channel data by adopting a WPLI algorithm to obtain a plurality of phase difference values;

respectively calculating sine values corresponding to the plurality of phase difference values to obtain a plurality of sine values;

and generating a phase characteristic set by the plurality of sine value sets.

In one possible implementation manner of the first aspect, the calculation formula of the WPLI algorithm is as follows:

wherein WPLI_i，j，τBetween 0 and 1, E {. is an expected value operator,. DELTA.ω_i，j，τIs the phase difference between nodes i and j, Δ ω_i，j，τIs calculated as follows:

Δω_i，j，τ＝ω_i(τ)-ω_j(τ)。

in a possible implementation manner of the first aspect, the screening the phase feature set by using a preset screening algorithm to obtain a screening feature set includes:

evaluating each phase characteristic value of the phase characteristic set by adopting a preset screening algorithm to obtain a plurality of evaluation characteristic values;

extracting N evaluation characteristic values from the plurality of evaluation characteristic values according to a preset percentage, and generating a screening characteristic set from the N evaluation characteristic value sets, wherein N is a positive integer greater than or equal to 1.

In a possible implementation manner of the first aspect, the performing classification detection on the screening feature set to obtain a classification result includes:

and inputting the screening feature set into a preset support vector machine, and performing two-classification detection on the screening feature set by using a preset linear function, a preset polynomial function and a preset Gaussian function by using the preset support vector machine to obtain a diseased classification result or a normal classification result.

In a possible implementation manner of the first aspect, the method further includes:

randomly extracting N-1 evaluation characteristic values from the N evaluation characteristic values, and combining the N-1 evaluation characteristic values into a characteristic training set;

carrying out classification verification on the non-extracted evaluation characteristic values by adopting the characteristic training set to obtain verification accurate values;

respectively calculating verification accurate values corresponding to each evaluation characteristic value to obtain N verification accurate values;

and calculating the average value of the N verification accurate values to obtain a verification average value.

In a possible implementation manner of the first aspect, the acquiring electroencephalogram data of a sleep brain network of a subject to be detected includes:

acquiring sleep electroencephalogram data of a sleep brain network of a to-be-detected object, wherein the sleep electroencephalogram data comprises M (M) electroencephalogram channel data of a patient and M electroencephalogram channel data of a normal user, and M is a positive integer greater than or equal to 1;

extracting abnormal data from each electroencephalogram channel data, and deleting the abnormal data in each electroencephalogram channel data to obtain a plurality of clear channel data;

and filtering the data of the plurality of leading clear channels by adopting a preset filter to obtain the electroencephalogram data of the sleeping brain network of the object to be detected.

A second aspect of an embodiment of the present invention provides a depression patient data classification apparatus based on a sleep brain network, the apparatus including:

the acquisition module is used for acquiring electroencephalogram data of a sleeping brain network of a to-be-detected object, wherein the electroencephalogram data comprises a plurality of electroencephalogram channel data;

the phase calculation module is used for performing phase synchronization calculation on the plurality of electroencephalogram channel data by adopting a WPLI algorithm to obtain a phase characteristic set;

the screening module is used for screening the phase characteristic set by adopting a preset screening algorithm to obtain a screening characteristic set;

and the detection module is used for carrying out two-classification detection on the screening characteristic set to obtain a classification result.

Compared with the prior art, the method and the device for classifying the data of the depression patient based on the sleep brain network have the advantages that: the invention can adopt WPLI algorithm to represent the functional connectivity of brain network, and then determine whether the detected object is depression patient or not through the characteristic value, thereby greatly reducing signal interference and improving detection accuracy, meanwhile, the whole process is simple and rapid, and physiological burden and psychological burden of the detected object during detection can be reduced to a certain extent by collecting electroencephalogram signals of the sleep network of the detected object, thereby the detected data can better accord with the actual condition of the detected object, and further the detection accuracy is improved.

Drawings

Fig. 1 is a schematic flowchart of a method for classifying data of a depressed patient based on a sleep brain network according to an embodiment of the present invention;

FIG. 2 is a sleep stage classification diagram of a sleep brain network according to an embodiment of the present invention;

FIG. 3 is a functional connectivity diagram of brain channels provided by an embodiment of the present invention;

fig. 4 is a schematic structural diagram of a depression patient data classification device based on a sleep brain network according to an embodiment of the present invention.

Detailed Description

The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.

The current commonly used detection method has the following problems: the detection and judgment are carried out through behaviors and environments, and symptoms in a period of time need to be recorded, so that the detection period is long and the detection efficiency is low; the electroencephalogram signals are adopted for detection, the data volume to be processed is large, the redundancy is large, the data processing is difficult, the processing time is long, and the detection accuracy is low.

In order to solve the above problems, a method for classifying data of a depression patient based on a sleep brain network provided by the embodiments of the present application will be described and illustrated in detail by the following specific examples.

Referring to fig. 1, a flow chart of a method for classifying data of a depressed patient based on a sleep brain network according to an embodiment of the present invention is shown.

By way of example, the method for classifying data of depression patients based on sleep brain network may include:

s11, acquiring electroencephalogram data of the sleep brain network of the object to be detected, wherein the electroencephalogram data comprise a plurality of electroencephalogram channel data.

In this embodiment, the electroencephalogram data may include multi-lead electroencephalogram channel data of the user, which is electroencephalogram waves of the user while sleeping. The electroencephalogram channel data during sleeping are easy to acquire, the daily life of a user cannot be influenced, the user can conveniently perform detection, and the detection efficiency can be improved.

Referring to fig. 2, a sleep stage classification diagram of a sleep brain network according to an embodiment of the present invention is shown. In actual practice, sleep may be divided into different epochs. Specifically, the sleep period may be divided into a Wake period (wakeful period), a REM period (rapid eye movement sleep period), and an NREM period (non-rapid eye movement sleep period), wherein the NREM period may include N1 period, N2 period, and N3 period.

After electroencephalogram channel data are obtained, the electroencephalogram channel data are divided into a Wake period, a REM period, an N1 period, an N2 period and an N3 period respectively, and then the electroencephalogram data of each period are detected respectively, so that the processing amount of the electroencephalogram data is reduced, and the processing efficiency of the data is improved.

When acquiring electroencephalogram data of different periods, data and data are prone to interfere with each other, and redundancy and overlapping of data are large, in order to reduce redundant data, in this embodiment, the step S11 may include the following sub-steps:

the substep S111 is to obtain a plurality of electroencephalogram channel data of the sleep brain network of the object to be detected, wherein the electroencephalogram channel data comprise M electroencephalogram channel data of the patient and M electroencephalogram channel data of the normal user, and M is a positive integer greater than or equal to 1.

Referring to fig. 3, a functional connectivity diagram of brain channels provided by an embodiment of the present invention is shown. In order to improve the accuracy of detection, in this embodiment, the data of the user's electroencephalogram channels may be acquired, and then the patient detection may be performed through the data of the electroencephalogram channels. Meanwhile, in order to distinguish the sick user from the normal user, in actual operation, electroencephalogram channel data of the patient and electroencephalogram channel data of the user can be acquired respectively.

Specifically, 16-lead electroencephalogram channel data for the patient and 16-lead electroencephalogram channel data for the user may be separated. For convenience of processing, the 16 electroencephalogram channel data may be labeled as FP1, FP2, F3, F4, F7, F8, C3, C4, P3, P4, O1, O2, T3, T4, T5, T6, respectively.

And a substep S112, extracting abnormal data from each electroencephalogram channel data, and deleting the abnormal data in each electroencephalogram channel data to obtain a plurality of clear guidance channel data.

To reduce overlapping data and interfering data, anomalous data may be deleted from the electroencephalogram channel data, thereby reducing the amount of redundancy of data.

For example, an artifact waveform or an overlap waveform in an electroencephalogram waveform is detected using a 30-second electroencephalogram waveform as one electroencephalogram channel data, and the artifact waveform or the overlap waveform is deleted from the electroencephalogram waveform.

And a substep S113, performing filtering processing on the plurality of leading clear channel data by adopting a preset filter to obtain preset electroencephalogram data.

Because the frequency ranges of the brain waves delta, theta, alpha and beta are 0.5-4 Hz, 4-8 Hz, 8-12 Hz and 12-32Hz respectively, after abnormal data in the brain waves are processed, the preset zero phase shift Butterworth band-pass filter (0.5-60Hz) can be adopted to filter the brain waves to obtain corresponding brain wave data.

The interference signals in the data can be further eliminated by filtering through the filter, so that the detection accuracy can be improved.

And S12, performing phase synchronization calculation on the plurality of electroencephalogram channel data by adopting a WPLI algorithm to obtain a phase feature set.

After obtaining the data of the plurality of electroencephalogram conducting channels, phase synchronization calculation can be carried out on the data of the plurality of electroencephalogram conducting channels to obtain a phase characteristic set, and the functional connectivity of the sleep brain network of the user can be determined through the phase characteristic set.

In this embodiment, the WPLI algorithm is a weighted phase lag exponential algorithm. The WPLI algorithm can weight signals for the imaginary part of the cross-power spectrum, improves the anti-interference capability to noise sources and the sensitivity to phase synchronous change, and can effectively and accurately evaluate the functional connectivity of the tristimania sleep network.

To improve the computational efficiency, step S12 may include the following sub-steps, as an example:

and a substep S121 of calculating a channel phase difference between every two of the plurality of electroencephalogram channel data by adopting a WPLI algorithm to obtain a plurality of phase difference values.

For example, the electroencephalogram channel data includes 16 channels, and the channel phase difference between the first electroencephalogram channel data and the second to sixteenth electroencephalogram channel data may be calculated first, then the channel phase difference between the second electroencephalogram channel data and the third to sixteenth electroencephalogram channel data may be calculated, and so on until the channel phase difference between the fifteenth electroencephalogram channel data and the sixteenth electroencephalogram channel data is calculated.

Specifically, the calculation formula of the WPLI algorithm is as follows:

wherein WPLI_i，j，τBetween 0 and 1, E {. is an expected value operator,. DELTA.ω_i，j，τIs the phase difference between nodes i and j, Δ ω_i，j，τIs calculated as follows:

Δω_i，j，τ＝ω_i(τ)-ω_j(τ)。

and a substep S122 of respectively calculating sine values corresponding to the plurality of phase difference values to obtain a plurality of sine values.

After the plurality of phase difference values are obtained through calculation, the sine value of each phase difference value can be respectively calculated, and a plurality of sine values are obtained.

And a substep S123 of generating a phase characteristic set from the plurality of sine value sets.

The plurality of sine values can be used as characteristic values of electroencephalogram channel data, and then the plurality of sine values are collected to generate a phase characteristic set. The set of phase features is a set of several sine values.

And S13, screening the phase characteristic set by adopting a preset screening algorithm to obtain a screening characteristic set.

The phase characteristic set is screened by adopting a preset screening algorithm, all characteristic values in the phase characteristic set can be classified and sorted, unnecessary characteristic values are eliminated, the data quantity required by detection is reduced, and the processing efficiency is improved.

In order to further improve the screening efficiency, step S13 may include the following sub-steps, as an example:

and the substep S131, evaluating each phase characteristic value of the phase characteristic set by adopting a preset screening algorithm to obtain a plurality of evaluation characteristic values.

Specifically, the preset screening algorithm may include a Fisher score algorithm and a Pearson Correlation algorithm, and the Fisher score algorithm may be adopted to score each phase feature value of the phase feature set to obtain an evaluation feature value. The evaluation feature value may be a fractional value, for example, 5 points, 10 points. Each phase characteristic value of the phase characteristic set can be classified by adopting a Pearson Correlation algorithm, the phase characteristic value can be distinguished as a diseased characteristic value, the phase characteristic value is a normal characteristic value, and after the diseased characteristic value and the normal characteristic value are distinguished, the phase characteristic value is labeled, so that subsequent distinguishing and screening are facilitated.

After scoring and classifying, the phase feature values within each phase feature set may be sorted by the magnitude of their score values. Specifically, the sorting may be performed from high to low, or from low to high, and may be specifically adjusted according to actual needs.

And a substep S132 of extracting N evaluation characteristic values from the plurality of evaluation characteristic values according to a preset percentage, and generating a screening characteristic set from the N evaluation characteristic value sets, wherein N is a positive integer greater than or equal to 1.

After the plurality of evaluation characteristic values are sorted according to the score values, the evaluation characteristic values with preset percentage can be extracted, and then the N evaluation characteristic value sets are generated into a screening characteristic set.

For example, 100 evaluation characteristic values can be extracted, 10% of the evaluation characteristic values before the numerical value can be extracted, and then the screening characteristic set is generated by using the 10% evaluation characteristic value set.

And S14, carrying out two-classification detection on the screening feature set to obtain a classification result.

After the screening feature set is generated, the screening feature set can be subjected to two-classification detection, and the screening feature set is classified into a diseased class or a normal class, so that a diseased classification result or a normal classification result is obtained, and finally the medical staff can use the results as reference.

To further improve the accuracy of the detection, step S14 may include the following sub-steps, as an example:

and a substep S141 of inputting the screening feature set into a preset support vector machine, and performing two-classification detection on the screening feature set by using a preset linear function, a preset polynomial function and a preset Gaussian function by the preset support vector machine to obtain a diseased classification result or a normal classification result.

In particular, the support vector machine may include three classifiers, each of which may correspond to a function. And then inputting the screening feature set into three classifiers, wherein the three classifiers can respectively carry out classification detection on the screening feature set and carry out labeling according to classification results to obtain diseased classification results or normal classification results.

To determine the accuracy of the classification, the method may further include, as an example:

s15, randomly extracting N-1 evaluation characteristic values from the N evaluation characteristic values, and combining the N-1 evaluation characteristic values into a characteristic training set.

And S16, carrying out classification verification on the non-extracted evaluation characteristic values by adopting the characteristic training set to obtain verification accurate values.

And S17, respectively calculating the verification accurate value corresponding to each evaluation characteristic value to obtain N verification accurate values.

And S18, calculating the average value of the N verification accurate values to obtain a verification average value.

In this embodiment, the verification may be performed for as many times as there are evaluation feature values, for example, N times as many evaluation feature values are performed.

Specifically, N-1 evaluation characteristic values are randomly extracted from the N evaluation characteristic values to serve as a training set, the remaining 1 evaluation characteristic value serves as a test set, and the training set is adopted to carry out verification detection on the test set to obtain a verification score value. And then, repeatedly executing the steps for N times to obtain a verification score value corresponding to each evaluation characteristic value, averaging the verification score values obtained by each verification to obtain a final verification result value, and comparing the verification result value with a preset value.

In this embodiment, an embodiment of the present invention provides a method for classifying data of a depression patient based on a sleep brain network, which has the following beneficial effects: the invention can adopt WPLI algorithm to represent the functional connectivity of brain network, and then determine whether the detected object is depression patient or not through the characteristic value, thereby greatly reducing signal interference and improving detection accuracy, meanwhile, the whole process is simple and rapid, and physiological burden and psychological burden of the detected object during detection can be reduced to a certain extent by collecting electroencephalogram signals of the sleep network of the detected object, thereby the detected data can better accord with the actual condition of the detected object, and further the detection accuracy is improved.

The embodiment of the invention also provides a depression patient data classification device based on the sleep brain network, and referring to fig. 4, a schematic structural diagram of the depression patient data classification device based on the sleep brain network provided by the embodiment of the invention is shown.

Wherein, as an example, the depression patient data classification device based on the sleep brain network may include:

the acquiring module 401 is configured to acquire electroencephalogram data of a sleeping brain network of a subject to be detected, where the electroencephalogram data includes a plurality of electroencephalogram channel data;

the phase calculation module 402 is configured to perform phase synchronization calculation on the plurality of electroencephalogram channel data by using a WPLI algorithm to obtain a phase feature set;

the screening module 403 is configured to screen the phase feature set by using a preset screening algorithm to obtain a screening feature set;

and the detection module 404 is configured to perform two-class detection on the screening feature set to obtain a classification result.

Further, the phase calculation module is further configured to:

calculating channel phase differences between every two pieces of the electroencephalogram channel data by adopting a WPLI algorithm to obtain a plurality of phase difference values;

respectively calculating sine values corresponding to the plurality of phase difference values to obtain a plurality of sine values;

and generating a phase characteristic set by the plurality of sine value sets.

Further, the calculation formula of the WPLI algorithm is as follows:

wherein WPLI_i，j，τBetween 0 and 1, E {. is an expected value operator,. DELTA.ω_i，j，τIs the phase difference between nodes i and j, Δ ω_i，j，τIs calculated as follows:

Δω_i，j，τ＝ω_i(τ)-ω_j(τ)。

further, the screening module is further configured to:

evaluating each phase characteristic value of the phase characteristic set by adopting a Fisher score and Pearson Correlation algorithm to obtain a plurality of evaluation characteristic values;

extracting N evaluation characteristic values from the plurality of evaluation characteristic values according to a preset percentage, and generating a screening characteristic set from the N evaluation characteristic value sets, wherein N is a positive integer greater than or equal to 1.

Further, the detection module is also used for

And inputting the screening feature set into a preset support vector machine, and performing two-classification detection on the screening feature set by using a preset linear function, a preset polynomial function and a preset Gaussian function by using the preset support vector machine to obtain a diseased classification result or a normal classification result.

Further, the apparatus further comprises:

the extraction module is used for randomly extracting N-1 evaluation characteristic values from the N evaluation characteristic values and combining the N-1 evaluation characteristic values into a characteristic training set;

the verification module is used for carrying out classification verification on the non-extracted evaluation characteristic values by adopting the characteristic training set to obtain verification accurate values;

the accurate value calculating module is used for calculating the verification accurate value corresponding to each evaluation characteristic value respectively to obtain N verification accurate values;

and the average value calculating module is used for calculating the average value of the N verification accurate values to obtain a verification average value.

Further, the obtaining module is further configured to:

acquiring sleep electroencephalogram data of a sleep brain network of a to-be-detected object, wherein the sleep electroencephalogram data comprises M (M) electroencephalogram channel data of a patient and M electroencephalogram channel data of a normal user, and M is a positive integer greater than or equal to 1;

extracting abnormal data from each electroencephalogram channel data, and deleting the abnormal data in each electroencephalogram channel data to obtain a plurality of clear channel data;

and filtering the data of the plurality of leading clear channels by adopting a preset filter to obtain the electroencephalogram data of the sleeping brain network of the object to be detected.

Further, an embodiment of the present application further provides an electronic device, including: a memory, a processor and a computer program stored on the memory and executable on the processor, the processor when executing the program implementing the method for classifying data of a depressed patient based on a sleep brain network as described in the above embodiments.

Further, the present application also provides a computer-readable storage medium storing computer-executable instructions for causing a computer to execute the method for classifying data of depression patients based on sleep brain network according to the above embodiment.

While the foregoing is directed to the preferred embodiment of the present invention, it will be understood by those skilled in the art that various changes and modifications may be made without departing from the spirit and scope of the invention.