An rTMS efficacy prediction method for insomnia patients based on electroencephalogram microstate

Abstract

The present application relates to a kind of insomnia patient rTMS curative effect prediction method based on electroencephalogram microstate, comprising: step 1: obtaining data set, wherein the data set includes the baseline electroencephalogram microstate index of multiple insomnia patients and corresponding rTMS treatment effect category label;Step 2: establish curative effect prediction model, utilize data set to train curative effect prediction model;Step 3: according to the curative effect prediction model of training completion, the treatment effect of rTMS of insomnia patient to be predicted is predicted.The insomnia patient rTMS curative effect prediction method based on electroencephalogram microstate proposed in the present application, by carrying out rTMS treatment insomnia experiment to construct data set, further establish reliable curative effect prediction model, utilize the curative effect prediction model can accurately predict the treatment effect of rTMS to individual, can provide individualized treatment decision information and reasonable medical advice for insomnia patient.

Description

Technical Field

[0001] This invention belongs to the field of repetitive transcranial magnetic stimulation (rTMS) technology, specifically relating to a method for predicting the efficacy of rTMS in insomnia patients based on brainwave microstates. Background Technology

[0002] Insomnia not only negatively impacts daytime cognitive functions such as alertness, memory, attention, and executive function, but also increases the risk of cardiovascular disease, dementia, and various mental disorders.

[0003] Transcranial magnetic stimulation (TMS) is a painless, non-invasive neuromodulation technique. Repetitive transcranial magnetic stimulation (rTMS) is a mode of TMS that can be used to improve sleep problems in insomnia patients. However, a small percentage of people do not experience significant improvement in sleep quality after rTMS treatment, indicating that rTMS treatment for insomnia has the drawback of significant individual variability in efficacy.

[0004] Current technology cannot accurately predict the therapeutic effect of rTMS on an individual, and cannot provide insomnia patients with personalized treatment decision-making information and reasonable medical advice. Summary of the Invention

[0005] To address the aforementioned problems in existing technologies, this invention provides a method for predicting the therapeutic effect of rTMS in insomnia patients based on brainwave microstates. The technical problem to be solved by this invention is achieved through the following technical solution:

[0006] This invention provides a method for predicting the efficacy of rTMS in insomnia patients based on brainwave microstates, comprising:

[0007] Step 1: Obtain the dataset, which includes baseline EEG microstate indices of multiple insomnia patients and corresponding rTMS treatment effect category labels;

[0008] Step 2: Establish an efficacy prediction model and train the efficacy prediction model using the dataset;

[0009] Step 3: Based on the trained efficacy prediction model, predict the treatment effect of rTMS for patients with insomnia.

[0010] In one embodiment of the present invention, step 1 includes:

[0011] Step 1.1: Collect EEG signals from multiple insomnia patients before rTMS treatment, and preprocess the EEG signals to eliminate artifact interference;

[0012] Step 1.2: Calculate the GFP curve for the EEG signal of each insomnia patient;

[0013] Step 1.3: Perform two clustering processes on the GFP curve using the improved k-means clustering method to obtain the microstate time series corresponding to each insomnia patient;

[0014] Step 1.4: Obtain the baseline EEG microstate indices for each insomnia patient based on the microstate time series corresponding to each patient;

[0015] Step 1.5: Collect the pre-treatment PSQI score and post-treatment PSQI score of each insomnia patient before and after rTMS treatment. Based on the pre-treatment PSQI score and the post-treatment PSQI score, obtain the percentage improvement in PSQI score for each insomnia patient.

[0016] Step 1.6: Based on the median percentage improvement in the PSQI scale score, divide all insomnia patients into two groups and assign a corresponding rTMS treatment effect category label to the insomnia patients in each group;

[0017] Step 1.7: Obtain the dataset based on the baseline EEG microstate indices of each insomnia patient and the corresponding rTMS treatment effect category label.

[0018] In one embodiment of the present invention, in step 1.1,

[0019] The EEG signal is the resting EEG signal of the insomnia patient in a closed state for 5 minutes before rTMS treatment;

[0020] The preprocessing includes filtering, rereference, removal of useless electrodes, interpolation of bad leads, removal of bad segments, and independent component analysis.

[0021] In one embodiment of the present invention, in step 1.2, GFP is calculated according to the following formula:

[0022]

[0023] In the formula, K represents the number of electrodes in the EEG data, and V i (t) represents the voltage of the i-th electrode at time t, V mean (t) represents the average value of the instantaneous voltage between the electrodes.

[0024] In one embodiment of the present invention, step 1.3 includes:

[0025] Step 1.31: Using the improved k-means clustering method, cluster the topological map of the peak of the GFP curve corresponding to each insomnia patient to obtain the individual-level micro-state topological map;

[0026] Step 1.32: Using the improved k-means clustering method, cluster the individual-level microstate topography maps of all insomnia patients to obtain group-level microstate topography maps;

[0027] Step 1.33: Based on the maximum spatial correlation between the topographic map of the peak of the GFP curve corresponding to each insomnia patient and the group-level microstate topographic map, obtain the microstate category corresponding to the peak of the GFP curve;

[0028] Step 1.34: Based on the microstate categories of all peaks in the GFP curve, obtain the microstate time series corresponding to each insomnia patient.

[0029] In one embodiment of the present invention, in step 1.4, the baseline EEG microstate indices include:

[0030] In the microstate time series, the duration of each microstate category, the average duration of all microstate categories, the number of occurrences of each microstate category, the average number of occurrences of all microstate categories, the time percentage of each microstate category, the average GFP corresponding to each microstate category, the percentage of transition from one microstate category to another observed in the microstate time series, the expected percentage of transition from one microstate category to another, and the difference between the percentage of transition from one microstate category to another and the expected percentage of transition.

[0031] In one embodiment of the present invention, in step 1.5, the percentage improvement in the PSQI scale score is calculated according to the following formula:

[0032]

[0033] In the formula, the score reduction rate represents the percentage improvement in the PSQI score, PSOI (before treatment) represents the PSQI score before treatment, and PSOI (after treatment) represents the PSQI score after treatment.

[0034] In one embodiment of the present invention, in step 2, the efficacy prediction model is a logistic regression classifier;

[0035] Before training the efficacy prediction model using the dataset, feature normalization and PCA dimensionality reduction are performed on each training sample in the dataset.

[0036] In one embodiment of the present invention, step 3 includes:

[0037] Step 3.1: Collect the electroencephalogram (EEG) signals of the patient to be predicted for insomnia before rTMS treatment, and preprocess the EEG signals to remove artifact interference;

[0038] Step 3.2: Calculate the corresponding GFP curve based on the EEG signals of the insomnia patient to be predicted after removing artifact interference;

[0039] Step 3.3: Obtain the corresponding microstate time series based on the GFP curve, and obtain the baseline EEG microstate indices of the insomnia patient to be predicted based on the microstate time series;

[0040] Step 3.4: After performing feature normalization and PCA dimensionality reduction on the baseline EEG microstate indices of the insomnia patient to be predicted, the data are input into the trained efficacy prediction model to obtain the rTMS treatment effect prediction category result corresponding to the insomnia patient to be predicted.

[0041] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0042] The present invention provides a method for predicting the efficacy of rTMS in insomnia patients based on EEG microstates. By conducting rTMS treatment experiments on insomnia to construct a dataset, a reliable efficacy prediction model is further established. This efficacy prediction model can accurately predict the treatment effect of rTMS on individuals, and can provide insomnia patients with personalized treatment decision information and reasonable medical advice.

[0043] The above description is merely an overview of the technical solution of the present invention. In order to better understand the technical means of the present invention and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of the present invention more apparent and understandable, preferred embodiments are described in detail below with reference to the accompanying drawings. Attached Figure Description

[0044] Figure 1 This is a flowchart of an rTMS efficacy prediction method for insomnia patients based on EEG microstates, provided in an embodiment of the present invention.

[0045] Figure 2 This is a flowchart of a micro-state time series acquisition method provided by an embodiment of the present invention;

[0046] Figure 3 This is a schematic diagram illustrating the acquisition of a micro-state time series according to an embodiment of the present invention;

[0047] Figure 4 This is a flowchart illustrating the training process of a therapeutic efficacy prediction model provided in an embodiment of the present invention.

[0048] Figure 5 This is an ROC curve of a therapeutic efficacy prediction model provided in an embodiment of the present invention; Detailed Implementation

[0049] To further illustrate the technical means and effects adopted by the present invention to achieve the intended purpose, the following describes in detail, with reference to the accompanying drawings and specific embodiments, a method for predicting the efficacy of rTMS in insomnia patients based on EEG microstates proposed according to the present invention.

[0050] The foregoing and other technical contents, features, and effects of the present invention will be clearly presented in the following detailed description of specific embodiments in conjunction with the accompanying drawings. Through the description of the specific embodiments, a more in-depth and concrete understanding can be gained of the technical means and effects adopted by the present invention to achieve its intended purpose. However, the accompanying drawings are for reference and illustration only and are not intended to limit the technical solutions of the present invention.

[0051] Example 1

[0052] Please see Figure 1 , Figure 1 This is a flowchart of a method for predicting the efficacy of rTMS in insomnia patients based on EEG microstates, provided in an embodiment of the present invention. As shown in the figure, the method for predicting the efficacy of rTMS in insomnia patients based on EEG microstates in this embodiment includes:

[0053] Step 1: Obtain the dataset;

[0054] The dataset includes baseline EEG microstate indices for multiple insomnia patients and corresponding rTMS treatment effect category labels.

[0055] In one optional implementation, step 1 includes:

[0056] Step 1.1: Collect EEG signals from multiple insomnia patients before rTMS treatment and preprocess the EEG signals to eliminate artifact interference;

[0057] Before collecting EEG signals from multiple insomnia patients prior to rTMS treatment, participants meeting the inclusion and exclusion criteria need to be recruited. In this example, 86 insomnia patients were recruited to participate in the rTMS intervention experiment. All participants met the DSM-V diagnostic criteria for insomnia, and MRI scans confirmed that none of them had any structural brain abnormalities. None of the participants reported any history of neurological problems or serious illnesses. Before the experiment began, the precautions were clearly explained to each participant, and all participants signed written consent forms.

[0058] In this embodiment, the EEG signal is the resting-state EEG signal of the insomnia patient in a closed-eye state for 5 minutes before rTMS treatment.

[0059] Optionally, the subjects wore a 64-channel EEG electrode cap, with electrodes placed according to the international 10-20 system, and EEG ointment applied to keep the impedance of each electrode below 10kΩ to ensure high-quality EEG signals. The sampling rate was 1024Hz, and resting-state EEG signals were collected for 5 minutes with the subjects' eyes closed in a soundproof, quiet room between 8:00 and 9:00 am.

[0060] In this embodiment, the EEGLAB software is used to preprocess the acquired EEG signals. After the acquired EEG signals are imported, the EEGLAB software is used to locate the electrodes and perform preprocessing steps such as filtering, rereference, removal of useless electrodes, interpolation of bad leads, removal of bad segments, and independent component analysis.

[0061] Optionally, the acquired EEG signals are subjected to 49-51Hz band-stop filtering to remove power frequency interference; the acquired EEG signals are subjected to 1-40Hz band-pass filtering to remove high-frequency artifacts; and the acquired EEG signals are subjected to independent component analysis to remove eye movement and electromyography components from the EEG.

[0062] Step 1.2: Calculate the GFP curve for the EEG signal of each insomnia patient;

[0063] In this embodiment, GFP is calculated according to the following formula:

[0064]

[0065] In the formula, K represents the number of electrodes in the EEG data; in this embodiment, K is 64, V i (t) represents the voltage of the i-th electrode at time t, V mean (t) represents the average value of the instantaneous voltage between the electrodes.

[0066] Step 1.3: Perform two clustering processes on the GFP curves using the improved k-means clustering method to obtain the microstate time series corresponding to each insomnia patient;

[0067] Please refer to the above. Figure 2 and Figure 3 , Figure 2 This is a flowchart of a micro-state time series acquisition method provided in an embodiment of the present invention. Figure 3 A schematic diagram illustrating the acquisition of a micro-state time series according to an embodiment of the present invention is shown in the figure. In an optional implementation, step 1.3 includes:

[0068] Step 1.31: Using the improved k-means clustering method, cluster the topological map of the peak of the GFP curve corresponding to each insomnia patient to obtain the individual-level micro-state topological map;

[0069] In this embodiment, the improved k-means clustering method (ignoring spatial polarity) is used, with 4 clusters and 100 repetitions. The improved k-means clustering method ignores spatial polarity for two reasons: First, the polarity of EEG signals is generally unimportant because specific processes of neural activity are usually related to the amplitude and frequency of EEG signals, not their polarity. Second, if polarity were considered as different states, the number of clusters would increase significantly. This would complicate the analysis and hinder the extraction and comparison of microstates in EEG signals.

[0070] Step 1.32: Using the improved k-means clustering method, cluster the individual-level microstate topography maps of all insomnia patients to obtain group-level microstate topography maps;

[0071] In this embodiment, the number of clusters is 4.

[0072] Step 1.33: Based on the maximum spatial correlation between the topographic map of the peak of the GFP curve for each insomnia patient and the microstate topographic map at the group level, obtain the microstate category corresponding to the peak of the GFP curve;

[0073] In this embodiment, each peak of the GFP curve of an insomnia patient is assigned a microstate category based on the maximum spatial correlation between the topographic map of the peak of the GFP curve and the microstate topographic map at the group level.

[0074] Step 1.34: Based on the microstate categories of all peaks in the GFP curve, obtain the microstate time series corresponding to each insomnia patient.

[0075] In this embodiment, the GFP curve of the insomnia patient is divided according to the microstate category at each peak of the GFP curve to obtain the microstate time series corresponding to the insomnia patient. The data point at the midpoint of the two GFP peaks is interpolated as the end of one microstate and the beginning of another microstate.

[0076] Step 1.4: Obtain the baseline EEG microstate indices for each insomnia patient based on the microstate time series corresponding to each patient;

[0077] In this embodiment, the baseline EEG microstate indices include: the duration of each microstate category in the microstate time series, the average duration of all microstate categories, the number of occurrences of each microstate category, the average number of occurrences of all microstate categories, the time percentage of each microstate category, the average GFP corresponding to each microstate category, the percentage of transition from one microstate category to another observed in the microstate time series, the expected percentage of transition from one microstate category to another, and the difference between the percentage of transition from one microstate category to another and the expected percentage of transition.

[0078] Step 1.5: Collect the pre-treatment PSQI scores and post-treatment PSQI scores of each insomnia patient before and after rTMS treatment. Based on the pre-treatment and post-treatment PSQI scores, obtain the percentage improvement in PSQI scores for each insomnia patient.

[0079] In this embodiment, the percentage improvement in PSQI scale score is calculated according to the following formula:

[0080]

[0081] In the formula, the score reduction rate represents the percentage improvement in the PSQI score, PSOI (before treatment) represents the PSQI score before treatment, and PSOI (after treatment) represents the PSQI score after treatment.

[0082] The higher the percentage improvement in the PSQI (Pittsburgh Sleep Quality Index) score, the better the treatment effect of rTMS on the insomnia patient.

[0083] In this embodiment, each insomnia patient received left-side 1Hz DLPFC rTMS treatment for 4 consecutive weeks (5 times / week), with each treatment lasting 30 minutes.

[0084] Step 1.6: Based on the median percentage improvement in PSQI scores, all insomnia patients were divided into two groups, and each group of insomnia patients was assigned a corresponding rTMS treatment effect category label.

[0085] In this embodiment, all insomnia patients were arranged in ascending order of the percentage improvement in their PSQI scores. Based on the median, all insomnia patients were divided into two groups: an excellent group and a sub-excellent group. Insomnia patients in the excellent group were assigned an excellent rTMS treatment effect category label, while insomnia patients in the sub-excellent group were assigned a sub-excellent rTMS treatment effect category label.

[0086] Step 1.7: Obtain the dataset based on the baseline EEG microstate indices of each insomnia patient and the corresponding rTMS treatment effect category label.

[0087] Step 2: Build an efficacy prediction model and train the efficacy prediction model using the dataset;

[0088] In this embodiment, the efficacy prediction model is a logistic regression classifier. Before training the efficacy prediction model using the dataset, feature processing is performed on each training sample in the dataset. First, feature normalization is performed, then PCA dimensionality reduction is performed to remove redundant features, specifically retaining 99% of the interpretable variance, and then the model is trained.

[0089] like Figure 4 The flowchart shown in this embodiment of the invention illustrates the training process of a therapeutic efficacy prediction model. In this embodiment, a hierarchical ten-fold cross-validation strategy is used to divide the dataset (10n) into a training set (9n) and a test set (1n). Specifically, after training the therapeutic efficacy prediction model using the training set, the corresponding test set is input into the trained therapeutic efficacy prediction model using the same feature preprocessing and feature selection strategy for model evaluation. The performance index obtained from each fold of the therapeutic efficacy prediction model is averaged to obtain the average performance index of the therapeutic efficacy prediction model. The ROC curve of the therapeutic efficacy prediction model is shown below. Figure 5 As shown in Table 1, the evaluation results of the efficacy prediction model indicators include accuracy, precision, recall, F1 score, and AUC.

[0090] Table 1 Evaluation results of efficacy prediction model indicators

[0091]

[0092] Step 3: Based on the trained efficacy prediction model, predict the treatment effect of rTMS for patients with insomnia.

[0093] In this embodiment, step 3 includes:

[0094] Step 3.1: Collect EEG signals of patients with insomnia to be predicted before rTMS treatment, and preprocess the EEG signals to remove artifact interference;

[0095] In this embodiment, a 64-channel EEG electrode cap is used to collect EEG signals from patients whose insomnia is to be predicted, and EEGLAB software is used to preprocess the collected EEG signals to eliminate artifact interference.

[0096] Step 3.2: Calculate the corresponding GFP curve based on the EEG signals of the insomnia patient to be predicted after removing artifact interference;

[0097] The GFP value is calculated according to the formula (1) as described above, and the corresponding GFP curve of the patient to be predicted for insomnia is obtained.

[0098] Step 3.3: Obtain the corresponding microstate time series based on the GFP curve, and obtain the baseline EEG microstate indices of the insomnia patients to be predicted based on the microstate time series;

[0099] In this embodiment, based on the maximum spatial correlation between the topographic map of the peak of the GFP curve of the patient to be predicted as insomnia and the group-level microstate topographic map obtained in step 1, the microstate category corresponding to the peak of the GFP curve is obtained. Then, based on the microstate categories of all peaks of the GFP curve, the GFP curve of the patient to be predicted as insomnia is divided to obtain the microstate time series corresponding to the patient to be predicted as insomnia, thereby obtaining the baseline EEG microstate index of the patient to be predicted as insomnia.

[0100] Step 3.4: After performing feature normalization and PCA dimensionality reduction on the baseline EEG microstate indices of the patient to be predicted as insomnia, the data are input into the trained efficacy prediction model to obtain the rTMS treatment effect prediction category result corresponding to the patient to be predicted as insomnia.

[0101] This embodiment presents a method for predicting the efficacy of rTMS in insomnia patients based on EEG microstates. By conducting an experiment on rTMS treatment for insomnia to construct a dataset, a reliable efficacy prediction model is further established. This efficacy prediction model can accurately predict the treatment effect of rTMS on an individual, providing insomnia patients with personalized treatment decision information and reasonable medical advice, thus reducing the risk of time and financial loss for insomnia patients.

[0102] It should be noted that, in this document, the terms "comprising," "including," or any other variations are intended to cover non-exclusive inclusion, such that an article or device that comprises a list of elements includes not only those elements but also other elements not expressly listed. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the article or device that includes said element. Terms such as "connected" or "linked" are not limited to physical or mechanical connections but can include electrical connections, whether direct or indirect.

[0103] The above description, in conjunction with specific preferred embodiments, provides a further detailed explanation of the present invention. It should not be construed that the specific implementation of the present invention is limited to these descriptions. For those skilled in the art, various simple deductions or substitutions can be made without departing from the concept of the present invention, and all such modifications and substitutions should be considered within the scope of protection of the present invention.

Claims

1. A method for predicting the efficacy of rTMS in insomnia patients based on brainwave microstates, characterized in that, include: Step 1: Obtain the dataset, which includes baseline EEG microstate indices of multiple insomnia patients and corresponding rTMS treatment effect category labels; Step 1 includes: Step 1.1: Collect EEG signals from multiple insomnia patients before rTMS treatment, and preprocess the EEG signals to eliminate artifact interference; Step 1.2: Calculate the GFP curve for the EEG signal of each insomnia patient; Step 1.3: Perform two clustering processes on the GFP curves using an improved k-means clustering method to obtain the microstate time series corresponding to each insomnia patient; wherein, the improved k-means clustering method ignores spatial polarity during clustering, and step 1.3 includes: Step 1.31: Using the improved k-means clustering method, cluster the topological map of the peak of the GFP curve corresponding to each insomnia patient to obtain the individual-level micro-state topological map; Step 1.32: Using the improved k-means clustering method, cluster the individual-level microstate topography maps of all insomnia patients to obtain group-level microstate topography maps; Step 1.33: Based on the maximum spatial correlation between the topographic topography map at the peak of the GFP curve corresponding to each insomnia patient and the group-level microstate topography map, obtain the microstate category corresponding to the peak of the GFP curve; including: assigning a microstate category to each peak of the GFP curve of the insomnia patient based on the maximum spatial correlation between the topographic topography map at the peak of the GFP curve of the insomnia patient and the group-level microstate topography map, and backfitting the group-level microstate topography map to the individual data of each insomnia patient; Step 1.34: Obtain the microstate time series corresponding to each insomnia patient based on the microstate category of all peaks of the GFP curve; including: dividing the GFP curve of the insomnia patient according to the microstate category at each peak of the GFP curve to obtain the microstate time series corresponding to the insomnia patient, wherein the data point at the midpoint of two GFP peaks is interpolated as the end of one microstate and the beginning of another microstate; Step 1.4: Obtain the baseline EEG microstate indices for each insomnia patient based on the microstate time series corresponding to each patient; Step 1.5: Collect the pre-treatment PSQI score and post-treatment PSQI score of each insomnia patient before and after rTMS treatment. Based on the pre-treatment PSQI score and the post-treatment PSQI score, obtain the percentage improvement in PSQI score for each insomnia patient. Step 1.6: Based on the median percentage improvement in the PSQI scale score, divide all insomnia patients into two groups and assign a corresponding rTMS treatment effect category label to the insomnia patients in each group; Step 1.7: Obtain the dataset based on the baseline EEG microstate indices of each insomnia patient and the corresponding rTMS treatment effect category label for that insomnia patient; Step 2: Establish an efficacy prediction model and train the efficacy prediction model using the dataset; Step 3: Based on the trained efficacy prediction model, predict the treatment effect of rTMS for patients with insomnia.

2. The method for predicting the efficacy of rTMS in insomnia patients based on EEG microstates according to claim 1, characterized in that, In step 1.1, The EEG signal is the resting EEG signal of the insomnia patient in a closed state for 5 minutes before rTMS treatment; The preprocessing includes filtering, rereference, removal of useless electrodes, interpolation of bad leads, removal of bad segments, and independent component analysis.

3. The method for predicting the efficacy of rTMS in insomnia patients based on EEG microstates according to claim 1, characterized in that, In step 1.2, GFP is calculated according to the following formula: In the formula, K represents the number of electrodes in the EEG data, and V i (t) represents the voltage of the i-th electrode at time t, V mean (t) represents the average value of the instantaneous voltage between the electrodes.

4. The method for predicting the efficacy of rTMS in insomnia patients based on EEG microstates according to claim 1, characterized in that, In step 1.4, the baseline EEG microstate indices include: In the microstate time series, the duration of each microstate category, the average duration of all microstate categories, the number of occurrences of each microstate category, the average number of occurrences of all microstate categories, the time percentage of each microstate category, the average GFP corresponding to each microstate category, the percentage of transition from one microstate category to another observed in the microstate time series, the expected percentage of transition from one microstate category to another, and the difference between the percentage of transition from one microstate category to another and the expected percentage of transition.

5. The method for predicting the efficacy of rTMS in insomnia patients based on EEG microstates according to claim 1, characterized in that, In step 1.5, the percentage improvement in the PSQI scale score is calculated according to the following formula: In the formula, the score reduction rate represents the percentage improvement in the PSQI score, PSOI(before_treatment) represents the PSQI score before treatment, and PSOI(after_treatment) represents the PSQI score after treatment.

6. The method for predicting the efficacy of rTMS in insomnia patients based on EEG microstates according to claim 1, characterized in that, In step 2, the efficacy prediction model is a logistic regression classifier; Before training the efficacy prediction model using the dataset, feature normalization and PCA dimensionality reduction are performed on each training sample in the dataset.

7. The method for predicting the efficacy of rTMS in insomnia patients based on EEG microstates according to claim 1, characterized in that, Step 3 includes: Step 3.1: Collect the electroencephalogram (EEG) signals of the patient to be predicted for insomnia before rTMS treatment, and preprocess the EEG signals to remove artifact interference; Step 3.2: Calculate the corresponding GFP curve based on the EEG signals of the insomnia patient to be predicted after removing artifact interference; Step 3.3: Obtain the corresponding microstate time series based on the GFP curve, and obtain the baseline EEG microstate indices of the insomnia patient to be predicted based on the microstate time series; Step 3.4: After performing feature normalization and PCA dimensionality reduction on the baseline EEG microstate indices of the insomnia patient to be predicted, the data are input into the trained efficacy prediction model to obtain the rTMS treatment effect prediction category result corresponding to the insomnia patient to be predicted.

Images