Mental fatigue degree testing method

Abstract

The invention provides a mental fatigue degree testing method, which comprises the following steps of: firstly, performing an experiment on a target, acquiring an experiment result of the target in a normal state and an experiment result of the target in a fatigue state, and extracting electroencephalogram data in the experiment results; then, the brain waves are preprocessed, the preprocessed brain waves are analyzed, then micro-state feature parameters are obtained, the micro-state feature parameters of the target in different states are obtained, and the obtained micro-state feature parameters serve as standards for judging the mental fatigue degree of the target. And the target is monitored in real time, and real-time micro-state characteristic parameters of the target are obtained. The mental fatigue degree can be judged based on the real-time micro-state characteristic parameters according to the fact that the micro-state characteristic parameters obtained through previous experiments can serve as the standard for judging the mental fatigue degree of the target. The mental fatigue degree is judged through the real-time micro-state characteristic parameters, and the work of the employees can be reasonably arranged based on the mental fatigue degree.

Description

Technical Field

[0001] The invention relates to the field of mental fatigue degree testing, and in particular to a method for testing the mental fatigue degree. Background Art

[0002] At present, in the existing technology, the method of judging mental fatigue based on EEG signals often faces the problem of insufficient precision. Due to the complexity and susceptibility to interference of EEG signals themselves, traditional analysis methods have limitations in extracting key features, distinguishing different fatigue states, and eliminating noise interference, which greatly affects the accuracy of the judgment results. Therefore, although EEG signals are an important window to reflect brain activity, their application in mental fatigue assessment still requires more advanced analysis technology and algorithm support to improve the accuracy of judgment. Summary of the invention

[0003] The present invention aims to solve at least one of the technical problems existing in the prior art or related art.

[0004] To this end, a first aspect of the present invention provides a method for testing the degree of mental fatigue.

[0005] In view of this, the first aspect of the present invention provides a method for testing the degree of mental fatigue, including: conducting experiments on a target, obtaining experimental results of the target, and extracting electroencephalogram data in the experimental results; preprocessing the electroencephalogram data to obtain pre-analysis electroencephalogram data; analyzing the pre-analysis electroencephalogram data to obtain microstate characteristic parameters; performing real-time monitoring of the target to obtain real-time microstate characteristic parameters of the target; and judging the degree of mental fatigue based on the real-time microstate characteristic parameters.

[0006] In addition, a method for testing the degree of mental fatigue in the above technical solution provided by the present invention may also have the following additional technical features: In some technical schemes of the present invention, optionally, an experiment is conducted on a target, experimental results of the target are obtained, and electroencephalogram data in the experimental results are extracted, including: obtaining a normal electroencephalogram of the target in a normal state, collecting first behavioral data of the target, and recording a first result of a subjective fatigue questionnaire of the target; inducing the target from a normal state to a fatigue state through a mental fatigue inducing task; obtaining a fatigue electroencephalogram of the target in a fatigue state, collecting second behavioral data of the target, and recording a second result of the subjective fatigue questionnaire of the target; extracting a normal electroencephalogram and a fatigue electroencephalogram as electroencephalogram data.

[0007] In some technical schemes of the present invention, optionally, the EEG data is preprocessed to obtain pre-analysis EEG data, including: channel positioning of the EEG data to obtain first EEG data; removing four active electrodes in the first EEG data to obtain second EEG data; filtering the second EEG data to obtain third EEG data; downsampling the third EEG data to obtain fourth EEG data; baseline correction of the fourth EEG data to obtain fifth EEG data; removing bad segments in the fifth EEG data and replacing bad channels to obtain sixth EEG data; separating and removing eye movement components from the sixth EEG data by an independent component analysis method to obtain seventh EEG data; removing artifacts from the seventh EEG data to obtain eighth EEG data; using the average value of each electrode of the eighth EEG data as a reference electrode to obtain ninth EEG data; and bandpass filtering the ninth EEG data from 2 Hz to 20 Hz to obtain pre-analysis EEG data.

[0008] In some technical schemes of the present invention, optionally, the second EEG data is filtered to obtain the third EEG data, including: removing the 50 Hz power frequency interference in the second EEG data through a FIR digital filter to obtain the initial filtered EEG data; performing a 0.5 Hz to 30 Hz bandpass filtering process on the first processed brain wave to obtain the third EEG data.

[0009] In some technical solutions of the present invention, optionally, the microstate characteristic parameters include: entropy rate, global explained variance, time domain characteristic parameters, transition probability and Hurst exponent; wherein, the time domain characteristic parameters include average duration, time coverage and occurrence frequency, and the transition probability includes the deviation value between the observed transition probability and the expected transition probability.

[0010] In some technical schemes of the present invention, optionally, analyzing the first electroencephalogram to obtain microstate characteristic parameters includes: obtaining the global field power of the first electroencephalogram; based on the global field power, obtaining the first time domain point of the maximum value of the global field power; bringing the first time domain point into the electroencephalogram to obtain a first time series; performing a second preprocessing on the first time series to obtain a microstate time series; and obtaining microstate characteristic parameters based on the microstate time series.

[0011] In some technical solutions of the present invention, optionally, a second preprocessing is performed on the first time series to obtain a microstate time series, including: applying the T-AAHC algorithm to perform a cluster analysis on the first time series to determine the microstate category of the first time series; based on the microstate category, obtaining a microstate template topographic map; based on the microstate template topographic map, obtaining the microstate time series.

[0012] In some technical schemes of the present invention, optionally, based on the microstate time series, microstate characteristic parameters are obtained, including: based on the microstate time series, global explained variance, time domain characteristic parameters and transition probability are obtained; based on the global explained variance, static properties of pre-analysis EEG data are judged; based on the observed transition probability and the expected transition probability, the deviation value between the observed transition probability and the expected transition probability is obtained; based on the deviation value, a transition probability matrix is ​​constructed; based on the transition probability matrix, the short-range correlation of the pre-analysis EEG data is judged.

[0013] In some technical schemes of the present invention, optionally, obtaining microstate characteristic parameters based on the microstate time series also includes: constructing a state transition count matrix based on the number of state transitions in the microstate time series; converting the counts in the state transition count matrix into probabilities to construct a state transition probability matrix; obtaining Shannon entropy based on the state transition probability matrix; obtaining the entropy value of each transition probability of the transition probability matrix based on the transition probability matrix; obtaining conditional entropy based on the entropy value of each transition probability; obtaining the entropy rate based on the conditional Shannon entropy and step size; and judging the dynamic properties of the pre-analysis EEG data based on the entropy rate.

[0014] In some technical schemes of the present invention, optionally, obtaining microstate characteristic parameters based on the microstate time series also includes: performing binary classification on the first time series to obtain a binary time series; accumulating and processing the binary time series to obtain a random walk signal; performing discrete wavelet transform on the random walk signal to decompose and extract wavelet coefficients, and using the wavelet coefficients to obtain the structure function; plotting the value and scale of the structure function value to form a log-log graph; performing linear fitting on the points of the log-log graph to obtain the slope of the fitting line; obtaining the Hurst index based on the slope and the order of the structure function; and judging the long-range correlation of the pre-analysis EEG data based on the Hurst index.

[0015] Additional aspects and advantages of the present invention will be given in part in the following description and in part will be obvious from the following description, or will be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] The above and / or additional aspects and advantages of the present invention will become apparent and easily understood from the description of the embodiments in conjunction with the following drawings, in which: Figure 1 One of the flow charts according to one embodiment of the present invention; Figure 2 is one of the comparison diagrams according to an embodiment of the present invention; Figure 3 is one of the comparison diagrams according to an embodiment of the present invention; Figure 4 is one of the comparison diagrams according to an embodiment of the present invention; Figure 5 is one of the comparison diagrams according to an embodiment of the present invention; Figure 6 is one of the comparison diagrams according to an embodiment of the present invention; Figure 7 is one of the comparison diagrams according to an embodiment of the present invention; Figure 8 is one of the comparison diagrams according to an embodiment of the present invention; Fig. 9 is one of the comparison diagrams according to an embodiment of the present invention; Fig.10 is one of the comparison diagrams according to an embodiment of the present invention; Fig.11 It is one of the comparison diagrams according to an embodiment of the present invention. DETAILED DESCRIPTION

[0017] In order to more clearly understand the above-mentioned purpose, features and advantages of the present invention, the present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be noted that the embodiments of the present application and the features in the embodiments can be combined with each other without conflict.

[0018] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Therefore, the protection scope of the present invention is not limited to the specific embodiments disclosed below.

[0019] Refer to the following Figure 1 A method for testing the degree of mental fatigue according to some embodiments of the present invention is described.

[0020] like Figure 1 As shown, the first aspect of the present invention provides a method for testing the degree of mental fatigue, comprising: Step 102, conducting an experiment on the target, obtaining the experimental results of the target, and extracting the EEG data in the experimental results; Step 104, preprocessing the EEG data to obtain pre-analysis EEG data; Step 106, analyzing the pre-analysis EEG data to obtain microstate characteristic parameters; Step 108, monitor the target in real time to obtain real-time micro-state characteristic parameters of the target; Step 118, judging the degree of mental fatigue based on the real-time micro-state characteristic parameters.

[0021] The present invention provides a method for testing the degree of mental fatigue. First, an experiment is conducted on a target to obtain experimental results of the target in a normal state and an experimental result of the target in a fatigue state, and electroencephalogram data in the experimental results are extracted. Then, the brain waves are preprocessed and analyzed, and microstate characteristic parameters are obtained, and the microstate characteristic parameters of the target in different states are obtained, and the obtained microstate characteristic parameters are used as the standard for judging the degree of mental fatigue of the target. Then, the target is monitored in real time to obtain the real-time microstate characteristic parameters of the target. Because the microstate characteristic parameters obtained according to the previous experiment can be used as the standard for judging the degree of mental fatigue of the target, the degree of mental fatigue can be judged based on the real-time microstate characteristic parameters. The degree of mental fatigue can be accurately analyzed by judging the degree of mental fatigue through the real-time microstate characteristic parameters. This avoids the errors of traditional methods, provides a more accurate fatigue assessment, and helps to avoid the dangers and errors caused by excessive fatigue. In addition, it is possible to make reasonable arrangements for the work of employees based on the degree of mental fatigue, improve actual work efficiency, reduce operating costs, and promote the healthy life of users.

[0022] Specifically, the experimental results include a normal electroencephalogram, a fatigue electroencephalogram, first behavioral data, a first result of a subjective fatigue questionnaire, second behavioral data, and a second result of a subjective fatigue questionnaire.

[0023] Specifically, the microstate characteristic parameters are parameters of the target that can be used as the basis for the mental fatigue parameters, and the real-time microstate characteristic parameters are the microstate characteristic parameters of the target in the current state.

[0024] Furthermore, in some embodiments of the present invention, performing an experiment on a target, obtaining an experimental result of the target, and extracting EEG data from the experimental result include: Obtaining a normal EEG of the target in a normal state, collecting the target's first behavioral data, and recording the first result of the target's subjective fatigue questionnaire; Through mental fatigue inducing tasks, the target is induced from a normal state to a fatigue state; Obtaining the target's fatigue electroencephalogram in a fatigue state, collecting the target's second behavioral data, and recording the target's second result of a subjective fatigue questionnaire; Normal EEG and fatigue EEG were extracted as EEG data.

[0025] In this embodiment, the target's resting EEG, first behavioral data, and the first result of the subjective fatigue questionnaire in a normal state, as well as the target's resting EEG, second behavioral data, and the second result of the subjective fatigue questionnaire in a fatigued state are obtained. Through normal brain waves and fatigue brain waves, the user's brain waves in different states can be confirmed, and through the subjective fatigue questionnaire and behavioral data, it is determined that the user's current state matches the state of the brain waves, which makes it easier to use the subsequently obtained micro-feature state parameters as a benchmark to accurately distinguish between normal and fatigue states, and improve the accuracy of fatigue state recognition.

[0026] It is understandable that the experiment designed to explore the differences between subjects before and after mental fatigue is generally divided into two parts: the working memory task experiment part and the mental fatigue induction experiment part. After the experiment started, the subjects' three-minute resting EEG was collected and the working memory task was performed. After the end, the subjects were prompted to fill in the Karolinska Sleepiness Scale (KSS), see Table 1, to confirm that the experimental subjects were in a normal mental state. After confirming that the experimental subjects were in a normal state, the mental fatigue induction experiment part began, and the second working memory task was started after the subjects were judged to be fatigued. After the task was completed, the resting EEG was collected, and the experimental process ended. In the mental fatigue induction experiment part and the working memory task experiment part, a fixation point appeared in the center of the screen after the start, which lasted for 700ms. During the working memory task, the letters appeared for 2000 milliseconds, and during the mental fatigue induction task, the letters appeared for 500 milliseconds.

[0027] Table 1 Karolinska Sleepiness Scale (KSS) Points Feature description Verbal description 1 Very alert extremelyalert 2 Very alert veryalert 3 vigilance alert 4 A little alert ratheralert 5 Neither alert nor tired neitheralertnorsleepy 6 Somewhat sleepy some signs of sleepiness 7 Sleepy, but alert without difficulty sleepy,butnoefforttokeepalert 8 Drowsiness and difficulty staying alert sleepy,some effort tokeepalert 9 Drowsiness and difficulty staying alert verysleepy,greatefforttokeepalert 10 Very sleepy fightingsleep Furthermore, in some embodiments of the present invention, preprocessing the EEG data to obtain pre-analysis EEG data includes: Perform channel positioning on the electroencephalogram data to obtain first electroencephalogram data; Eliminate four active electrodes in the first electroencephalogram data to obtain second electroencephalogram data; Performing filtering processing on the second electroencephalogram data to obtain third electroencephalogram data; Downsampling the third electroencephalogram data to obtain fourth electroencephalogram data; Performing baseline correction processing on the fourth electroencephalogram data to obtain fifth electroencephalogram data; Remove bad segments from the fifth electroencephalogram data and replace bad channels to obtain the sixth electroencephalogram data; Separating and removing eye movement components, heart components, and muscle artifact components from the sixth electroencephalogram data by an independent component analysis method to obtain the seventh electroencephalogram data; removing artifacts of the seventh electroencephalogram data to obtain the eighth electroencephalogram data; Using the average value of each electrode of the eighth electroencephalogram data as the reference electrode to obtain ninth electroencephalogram data; The ninth electroencephalogram data is subjected to bandpass filtering processing from 2 Hz to 20 Hz to obtain pre-analysis electroencephalogram data.

[0028] In this embodiment, the EEG is subjected to channel positioning, four active electrodes are eliminated, filtering, downsampling, baseline correction, bad segments are eliminated and bad channels are replaced, eye movement components, cardiac components, muscle artifact components and artifacts are removed, and the average value of each electrode is used as the reference electrode. The EEG is preprocessed in the above manner, which significantly improves the purity and reliability of the signal, reduces the computational complexity of subsequent analysis, improves processing efficiency, and enhances analysis accuracy.

[0029] It should be noted that the device for collecting brain waves in the present invention is an EEG signal acquisition device of NeuroScan Company in the United States, which synchronously records the EEG signals during the working memory task of the subject. A 64-conductor electrode cap is used to collect EEG data, and the electrode is active Ag / AgCl. During the experiment, the impedance of each electrode is required to be less than 5KΩ. The placement of the electrode is strictly in accordance with the international standard lead 10-20 system. This placement system is the standard electrode placement method specified by the International Electroencephalography Association. The distance from the midpoint of the frontal pole to the root of the nose and the distance from the occipital point to the outer ridge of the occipital bone are each 10% of the total line length, and the distance between the remaining points is 20% of the total line length. In order to facilitate the subsequent comparison of information exchange between brain regions, the present invention will be divided into 60 electrodes. The amplifier uses SynAmps2, the experimental paradigm is presented by the E-Prime system, and the collected EEG signals are recorded using Curry8 software, with a sampling frequency of 1000Hz. The experimental environment light is suitable and there is no noise.

[0030] Physiological data analysis was performed in Matlab 2023a using the EEG analysis toolbox EEGLAB. The specific steps for preprocessing the collected raw resting EEG signals include electrode positioning; removing bilateral mastoid electrodes; using FIR digital filters to remove 50Hz power frequency interference to improve the signal-to-noise ratio of EEG signals; 0.5-30Hz bandpass filtering; downsampling to 256Hz; interpolating bad leads; removing bad segments; using independent component analysis (ICA) to remove eye movement and other components; whole-brain average reference; removing artifacts, etc. After preprocessing, the processed EEG data was filtered with a 2-20Hz bandpass filter to facilitate subsequent microstate analysis. At this point, the preprocessing process is completed.

[0031] Further, in some embodiments of the present invention, filtering the second electroencephalogram data to obtain third electroencephalogram data includes: Using a FIR digital filter, the 50 Hz power frequency interference in the second EEG data is removed to obtain the primary filtered EEG data; The first processed brain wave is subjected to a bandpass filtering process of 0.5 Hz to 30 Hz to obtain third electroencephalogram data.

[0032] In this embodiment, by performing multiple filtering processes on the second electroencephalogram, the signal-to-noise ratio of the electroencephalogram data is significantly improved, and the stability and clarity of the electroencephalogram data are improved.

[0033] Furthermore, in some embodiments of the present invention, the microstate characteristic parameters include: Entropy rate, global explained variance, temporal characteristic parameters, transition probability, and Hurst exponent; Among them, the time domain characteristic parameters include average duration, time coverage and occurrence frequency, and the transition probability is the transition probability including the deviation value between the observed transition probability and the expected transition probability.

[0034] In this embodiment, entropy rate, global explained variance, time domain characteristic parameter, transition probability and Hurst exponent are selected as parameters for analyzing the degree of mental fatigue, so as to realize multi-dimensional evaluation of the mental fatigue state, which together constitute a comprehensive description of the fatigue state, reduce the error that may be caused by single parameter analysis, and thus improve the accuracy of fatigue degree assessment.

[0035] Further, in some embodiments of the present invention, analyzing the pre-analysis EEG data to obtain microstate characteristic parameters includes: Obtain the global field power of the first EEG; Based on the global field power, obtaining the first time domain point of the maximum value of the global field power; Bring the first time domain point into the electroencephalogram to obtain a first time series; Performing a second preprocessing on the first time series to obtain a microstate time series; Based on the microstate time series, the microstate characteristic parameters are obtained.

[0036] In this embodiment, the global field power of the first electroencephalogram is obtained, and based on the global field power, the first time domain point of the maximum value of the global field power is obtained, the first time domain point is brought into the electroencephalogram, and the first time series is obtained. The first time series is subjected to a second preprocessing to obtain a microstate time series, and the microstate characteristic parameters are obtained based on the microstate time series. By selecting the EEG topographic map at the maximum point as the template topographic map, and then replacing the topographic map around the peak with this topographic map, the researcher is allowed to focus on a specific time period associated with the peak of brain activity, which helps to reduce the complexity of data analysis and improve the pertinence and accuracy of analysis.

[0037] Furthermore, in some embodiments of the present invention, performing a second preprocessing on the first time series to obtain a microstate time series includes: Apply the T-AAHC algorithm to perform cluster analysis on the first time series and determine the microstate category of the first time series; Based on the microstate category, a microstate template topographic map is obtained; Based on the microstate template topographic map, the microstate time series is obtained.

[0038] In this embodiment, the T-AAHC algorithm is applied to perform cluster analysis on the first time series, determine the microstate category of the first time series, obtain the microstate template topographic map based on the microstate category, and obtain the microstate time series based on the microstate template topographic map. By performing cluster analysis and processing on the first time series, the spatial and temporal information of the EEG data can be fully utilized, the accuracy of classification can be improved, the key features of brain activity can be highlighted, and subsequent analysis and interpretation can be more efficient and accurate.

[0039] Furthermore, in some embodiments of the present invention, obtaining microstate characteristic parameters based on the microstate time series includes: Based on the microstate time series, the global explained variance, time domain characteristic parameters and transition probability are obtained; Based on the global explained variance, the static properties of the pre-analyzed EEG data were determined; Based on the observed transition probability and the expected transition probability, a deviation value between the observed transition probability and the expected transition probability is obtained; Based on the deviation value, construct the transition probability matrix; Based on the transition probability matrix, the short-range correlation of the pre-analyzed EEG data was determined.

[0040] In this embodiment, the short-range correlation of the EEG can be determined by the transition probability matrix, reflecting the regulatory ability of the autonomic nervous system and the stability of the fatigue state. By analyzing the changes in these parameters, the impact of mental fatigue on the physiological stability of the individual can be evaluated and the dynamic change process of mental fatigue in a short period of time can be revealed.

[0041] Furthermore, in some embodiments of the present invention, obtaining microstate characteristic parameters based on the microstate time series also includes: Based on the number of state transitions in the microstate time series, a state transition count matrix is ​​constructed; Convert the counts in the state transition count matrix into probabilities and construct a state transition probability matrix; Based on the state transition probability matrix, obtain Shannon entropy; Based on Shannon entropy and step size, obtain the entropy rate; Based on the entropy rate, the dynamic properties of the pre-analyzed EEG data are determined.

[0042] In this embodiment, the dynamic properties are judged by the entropy rate, so that the randomness and regularity of the state transition can be understood, and then an early warning is issued based on the fatigue state to improve the user experience.

[0043] Furthermore, in some embodiments of the present invention, obtaining microstate characteristic parameters based on the microstate time series also includes: Perform binary classification on the first time series to obtain a binary time series; Accumulate and process binary time series to obtain random walk signals; Perform discrete wavelet transform on random walk signal to decompose and extract wavelet coefficients, and use the wavelet coefficients to obtain the structure function; The values ​​of the structure function values ​​are plotted against the scale to form a log-log plot; Perform a linear fit on the points of the log-log plot and obtain the slope of the fitted line; Based on the slope and the order of the structure function, the Hurst exponent is obtained; Based on the Hurst exponent, the long-range correlation of the pre-analyzed EEG data was determined.

[0044] In this embodiment, judging long-range correlation by the Hurst index can reveal the correlation and continuity of brain electrical activity over a long period of time, thereby reflecting the cumulative effect of fatigue state, predicting the development of fatigue state of an individual in the future, and improving the accuracy and reliability of mental fatigue assessment.

[0045] In another embodiment of analyzing the pre-analysis EEG data to obtain microstate characteristic parameters, the EEG microstate analysis is implemented based on cartool software, including the following steps: In the first step, the preprocessed EEG data topography of each subject was analyzed using the T-AAHC algorithm. In T-AAHC analysis, topography with similar topological configurations are clustered into one category. Each category obtained in this step is considered to represent a microstate category. For each resting-state EEG dataset to be analyzed, we first calculated the global field power (GFP) of the EEG time series and extracted the topography at the GFP peak.

[0046] In the second step, since the EEG topography is relatively stable and has a higher signal-to-noise ratio near the local maximum (i.e., "peak") of the GFP time series, the topography located at the GFP peak is used as the "original map" in the T-AAHC algorithm. After completing the above cluster analysis on the data of each subject, the criteria provided in the Cartool software (i.e., Davies & Bouldin, Gamma, Silhouette, Dunn Robust, Point-Biserial, Krzanowski-Lai Index, and Cross-Validation) were used to determine the number of optimal categories at the single-subject level.

[0047] In the third step, each "average category topography" corresponding to the optimal number of categories for each subject in a certain group was extracted and summarized. All the "average category topography" of all subjects in the group after aggregation were clustered using the T-AAHC algorithm, and the optimal number of categories was determined at the group level using the criteria provided by the Cartool software. The analysis at the group level found that the optimal number of categories was 4, and it was believed that the optimal number of categories for each subject was also 4, and the microstate template topography was obtained.

[0048] In the fourth step, the microstate template topography is fitted with the imported EEG data to obtain the microstate time series.

[0049] The fifth step is to calculate the characteristic parameters of the microstate time series, mainly including the global explained variance (GEV), duration, number of occurrences, coverage, and transition probabilities (TP) between microstates.

[0050] By calculating the deviation between the observed transition probability and the expected transition probability, the short-range correlation in the sequence can be reflected. The deviation value will be used to construct a transition probability matrix. Each value in the transition probability matrix represents the average value of the expected observed transition probability (OTP) minus the expected transition probability (ETP). The color-filled area represents the results with significant differences after performing a paired T test on the TP matrix. The transition probability matrix is ​​as follows: Figure 2 shown.

[0051] In addition to analyzing common microstate time domain parameters, this section also analyzes the static properties (GEV) and dynamic properties (entropy rate), short-range (TP) and long-range correlations (Hurst index) of microstate sequences.

[0052] Use the transition probability to construct a transition probability matrix. Each element in the probability matrix represents the probability of transitioning from one state to another. Traverse each element in the transition probability matrix and calculate the entropy value of each transition probability according to the formula.

[0053] ; In the formula is the conditional entropy, which means that when the current microstate is known In the case of Specifically, this is a measure of the average uncertainty of the next microstate change given the current microstate. and is the sum symbol, which means that for all possible current microstates and the next microstate The values ​​of are summed. The current microstate The value is This means that in the entire microstate sequence, the state Frequency of occurrence. In the current microstate The value is In the case of The value is This means that in the state Follow up status The frequency of occurrence. Conditional entropy is used to measure the complexity and uncertainty of the microstate sequence. Dividing the conditional entropy by the step length k, we get the average entropy value (entropy rate) in each step of the transfer. The entropy rate can be used to analyze the dynamic properties of the microstate and understand the randomness and regularity of state transfer.

[0054] Furthermore, the steps to calculate the long-range correlation are as follows: The microstate time series is classified into binary categories to simplify the complexity of the time series and focus the analysis on the trend of the microstate series.

[0055] The random walk signal is obtained by performing cumulative sum operation on the binary sequence. The random walk signal is decomposed into different frequency bands using discrete wavelet transform (DWT) to obtain detail coefficients and approximation coefficients.

[0056] The structure function is calculated using wavelet coefficients to describe the statistical characteristics of the signal at different scales. The detail coefficients of each scale are extracted from the wavelet coefficients, and the square mean of the absolute value of the detail coefficients of each scale is calculated to obtain the structure function.

[0057] The structure function and scale are logarithmized, and a linear fit is performed on the log-log coordinates. The slope is divided by 2 to obtain the Hurst index. The Hurst index is used to measure the long-term dependence and self-similarity of the time series.

[0058] Analysis of behavioral data and KSS subjective scale The behavioral data showed that in the normal state (Acc, 96±2%; RT, 1041.85±154.39ms) and fatigue state (Acc: 93±3%, RT, 1213.78±188.64ms), statistical analysis of accuracy, reaction time and KSS scale results showed that the participants had entered the fatigue state.

[0059] The present invention statistically analyzed the KSS scale scores of each subject before and after the brain fatigue induction experiment and the behavioral data (including accuracy and reaction time) during the working memory task. Behavioral data plays an important supporting role in cognitive neuroscience. When combined with electrophysiological data, it can provide a clearer and more intuitive description of related issues. The paired T test method was used to perform statistical analysis on the KSS scale scores and behavioral data. The α level of the statistical analysis used in this paper was 0.05. The results are shown in Figure 2. Figure 2 As shown in the figure, the KSS scale scores are as follows: Figure 2 As shown in (a), the KSS scale scores in the normal state are significantly lower than those in the fatigue state (t=6.631, p-paired t-test<0.0001, p-corrected<0.0001), which means that after the brain fatigue induction experiment, the subjects subjectively believed that they had entered a state of brain fatigue. The results of the subjects' accuracy and reaction time during the working memory task are shown in Figure 2 As shown in (b) and 2 (c), the subjects' accuracy rate decreased significantly after entering the fatigue state (t=5.627, p-paired t-test<0.0001, p-corrected<0.0145), and the reaction time increased significantly (t=5.529, p-paired t-test<0.0001, p-corrected<0.0006). These results objectively show that the subjects have entered a state of brain fatigue, and brain fatigue will affect the subjects' behavioral performance.

[0060] The data preprocessing was done based on the EEGLAB toolbox. The specific steps for preprocessing the collected raw resting EEG signals include electrode positioning; removing bilateral mastoid electrodes; using FIR digital filters to remove 50Hz power frequency interference to improve the signal-to-noise ratio of EEG signals; 0.5-30Hz bandpass filtering; downsampling to 256Hz; interpolating bad leads; removing bad segments; using independent component analysis (ICA) to remove eye movement and other components; whole-brain average reference; removing artifacts, etc. After preprocessing, the processed EEG data was truncated into 2s segments of resting EEG.

[0061] Studies have shown that healthy people have four microstates in normal and fatigue states. After microstate analysis, four microstate template topographic maps were obtained in normal and fatigue states, respectively, explaining about 65% of the overall variance, such as Figure 3 As shown. Figure 4 The four microstates shown are A, B, C, and D. Microstate A is related to hearing, and the boundary between positive and negative voltages extends diagonally from the left frontal to the right occipital region. Microstate B is related to the visual network, which is opposite to the diagonal of A. Microstate C extends horizontally and is related to the cognitive control network. Microstate D is related to the attention network and sleep, and the positive and negative voltages are located in the frontal central region (ignoring polarity). Under the two experimental conditions, the four types of microstate topography have a high degree of similarity, such as Figure 5 shown.

[0062] Figure 6 is the Global Explained Variance (GEV) value of the subjects in the normal state and the fatigue state. For each category of microstate, the GEV in the normal state and the fatigue state varied between 0.02 and 0.45. Paired t-test showed that after brain fatigue, the GEV of microstate B increased significantly (t=2.822, p-paired t-test=0.009,p-corrected<0.0001). The GEV of microstate C increased significantly (t=3.997, p-paired t-test=0.0005,p-corrected<0.0012). However, the GEV of microstate D showed a downward trend (t=5.351, p-paired t-test<0.0001,p-corrected<0.0001).

[0063] like Figure 7As shown in the figure, entropy rate is often used to measure the temporal properties of a sequence, and it can be used to observe how much uncertainty a microstate sequence has. The entropy rate of the fatigue state is significantly higher than that of the normal state, indicating that the uncertainty is higher in the fatigue state, and the communication between neuronal discharges and brain regions becomes more random and uncoordinated, making the transition sequence of the brain state more complex.

[0064] Figure 8 The average duration of microstates in the normal state and fatigue state of the subjects was subjected to paired T test. For different microstates, the average duration of microstates in the normal state and fatigue state ranged from 72.14 to 131.31ms. The average duration of microstate C increased in the brain fatigue state, showing a statistically significant difference (t=2.315, p-paired t-test=0.0288, p-corrected=0.0006). However, compared with the normal state, microstate D showed a downward trend in the brain fatigue state (t=4.515, p-paired t-test=0.0001, p-corrected<0.0001). * represents the p value obtained by paired T test is less than 0.05, and *** represents the p value obtained by paired T test after FDR correction is less than 0.0001.

[0065] Fig. 9 The time coverage of microstates in the normal state and fatigue state of the subjects was subjected to paired T test. The percentage of total time occupied by each microstate category ranged from 9% to 47%. The time coverage of microstate C in the fatigue state increased, showing a statistically significant difference (t=2.173, p-paired t-test=0.0391, p-corrected=0.0029). However, compared with the normal state, microstate D showed a downward trend in the fatigue state (t=4.095, p-pairedt-test=0.0004,p-corrected<0.0001). Among them, * represents the p value obtained by paired T test is less than 0.05, and ** represents the p value obtained by paired T test after FDR correction is less than 0.05.

[0066] The present invention evaluates the short-range dependency of the microstate sequence of the subjects under the state of brain fatigue based on the Markov chain analysis method. Short-range dependency refers to the correlation or dependency between adjacent microstates in the microstate sequence, which is manifested in the transition probability between certain microstates in a short time scale as a bias in a certain direction or following a certain pattern. By calculating the deviation between the observed transition probability and the expected transition probability, the short-range dependency in the sequence can be reflected. Fig.10As shown in the figure, each value in the transition probability matrix represents the average value of the expected observed transition probability (OTP) minus the expected transition probability (ETP), and the color-filled area represents the results with significant differences after performing a paired T test on the TP matrix.

[0067] Fig.10 The statistical results in show that there are significant differences in the matching process of expected transition probability and observed transition probability between fatigue state and normal state (FDR correction, p<0.05). Specifically, the analysis results of the transition probability matrix show that the abnormal transformation in the fatigue state has increased significantly, including the OTP of microstate Map C→Map A is less than ETP and the OTP of microstate Map A→Map B is greater than ETP, while the abnormal transformation of microstates in the normal state has decreased, such as the OTP of Map D→Map B is less than ETP and the OTP of Map D→Map C is greater than ETP. In addition, this experiment also used the paired T test to compare the values ​​of each group (OTP-ETP) between groups. The statistical results showed that compared with the normal state, the mutual transfer of fatigue state in Map A⟷Map C has a significant upward trend, Map A→Map D has a significant downward trend, and Map D→Map B has a significant upward trend. Among them, Fig.10 (a) is the normal state transition probability matrix, and (b) is the fatigue state transition probability matrix.

[0068] Fig.11 (a) is a schematic diagram of DFA for a subject. The slope of the linear regression line in the figure is α. The Hurst index is an indicator used to evaluate whether there is long-range dependence in the time series. In detrended fluctuation analysis (DFA), the relationship between the Hurst index and the slope α of the linear regression is Hurst index = α / 2.

[0069] Fig.11 (b) Statistics of the Hurst index of all subjects in normal and fatigue states. The Hurst index is greater than 0.5, which has long-term dependence. In the fatigue state, the Hurst index decreases significantly, and long-term memory is weakened.

[0070] In this study, the KSS scale was used to obtain the subjective evaluation of the subjects' fatigue level, and the scores ranged from 2 to 9. The paired T test showed that the subjects' fatigue level was significantly improved after the fatigue induction experiment. The Spearman correlation analysis of the KSS subjective scale and the microstate characteristic parameters showed that fatigue was significantly positively correlated with the B-type microstate GEV and Occ, significantly positively correlated with the C-type microstate GEV, Dur, and Cov, and significantly negatively correlated with the D-type microstate GEV, Dur, Occ, and Cov. The correlation between the KSS scale and the entropy value and the Hurst index was also analyzed, and it was found that fatigue was significantly positively correlated with the entropy value and significantly negatively correlated with the Hurst index.

[0071] The present invention uses the microstate analysis method to compare the changes in EEG activity of healthy people in normal and fatigue states, and finds that the microstate clustering results are similar to the traditional resting EEG clustering results, and the microstate topographic maps of the subjects in normal and fatigue states are not much different. The combination of two-step clustering analysis and meta-standards produced 4 template topographic maps, which best described the entire data set and explained about 65% of the global explained variance. This paper uses T-AAHC to more accurately capture the spatial characteristics of brain activity by considering the topological relationship between electrodes. The 4 microstate template topographic maps match the standard microstates (Class A, Class B, Class C, Class D). The results show that the EEG indicators obtained by the microstate analysis method have high repeatability.

[0072] The microstate parameters of the subjects in the normal state and fatigue state were compared and analyzed. In the time characteristic analysis, compared with the normal state, the presence of microstate categories Map B and Map C in the fatigue state increased significantly, and the presence of Map D decreased significantly. The Spearman correlation analysis of the subjective sleepiness scale and microstate parameters showed that the sleepiness characteristics were significantly positively and negatively correlated with the microstate categories Map BCD.

[0073] In the present invention, the occurrence of microstate category Map D in the fatigue state is systematically lower than that in the normal state, and the GEV, average duration and time coverage of microstate category Map C are significantly increased compared with the normal state. In many studies, it has been shown that microstate D is related to the dorsal attention network (DAN), which mainly originates from the frontoparietal control network area. Studies on the effects of mental fatigue on human brain function have found that mental fatigue can lead to inattention and decreased brain information processing ability, reducing the ability of the dorsal attention network to interact with information. It is inferred that mental fatigue may be the reason why the GEV, average duration, time coverage and occurrence frequency of microstate D in the resting state are lower than those in the normal state. In addition, the GEV, average duration and time coverage of microstate C in the fatigue state caused by continuous working memory tasks are significantly lower than those in the normal state, which is consistent with the previous one. Microstate C reflects the changes in the activity of the salient network, which is positively correlated with the activation of the bilateral temporal gyrus, the posterior cingulate cortex and the insula. The anterior insula is sensitive to changes in physiological signals such as heart rate and blood pressure, and is an important structure for processing sensory stimuli and transmitting physiological signals. Hilty et al. found that in continuous high-load tasks, the information interaction between the middle insula and the anterior insula and the sensorimotor cortex is enhanced. This may be the reason for the significant increase in the occurrence of microstate C. In addition, microstate C also reflects the activation of the default mode network (DMN). DMN is one of the main networks involved in the processing of self-focus and self-reference. Many scholars have confirmed that abnormalities in the self-reference process of DMN are closely related to depression, anxiety and schizophrenia. In the fatigue state, the increase in the global explained variance and occurrence frequency of microstate B may be due to the increased burden of the visual network in the fatigue state, the chain reaction of the cognitive control network, and the combined effect of changes in the attention network and sleep pressure.

[0074] The subjects' brain activity was higher in the fatigue state, indicating that the complexity and uncertainty of the brain increased. Fatigue may lead to an imbalance in the brain's internal regulatory mechanism and a decrease in the stability of the neural network, thereby increasing the complexity of the system. This reflects changes in the system's internal regulatory mechanism and cognitive resource allocation. The GEV of microstates B and C increased significantly, while the GEV of microstate D decreased significantly, indicating that mental fatigue increases the burden on visual and cognitive control: fatigue increases the burden on the visual and cognitive control networks, as shown by the increase in the GEV of microstates B and C. This reflects that the brain needs more resources and efforts to maintain visual processing and cognitive control functions in a fatigued state. Difficulty in maintaining attention and increased sleep pressure: Fatigue makes it more difficult to maintain attention, and the activity of the attention network decreases, which is reflected in the decrease in the GEV of microstate D. At the same time, sleep pressure increases, and the brain tends to reduce the activity of the attention network. Resource reallocation: The brain reallocates resources in a fatigued state, giving priority to visual and cognitive control functions and reducing resource investment in the attention network.

[0075] The dynamic properties (entropy rate) of the subjects in normal and fatigue states were compared. Entropy is often used to measure the complexity of EEG microstate sequences. Low entropy values ​​indicate that the data is highly predictable and regular, otherwise it is the opposite. The entropy rate in the fatigue state increased significantly, indicating that the brain's microstate sequence has become more complex and unpredictable, reflecting the increased instability of the brain in information processing and cognitive control. This phenomenon suggests that it is difficult for the brain to maintain stable and efficient functions in a fatigued state, emphasizing the importance of fatigue management. By monitoring the changes in the entropy rate, fatigue states can be identified in a timely manner, and effective intervention measures can be taken to ensure work efficiency and safety. The attribute value of the microstate sequence can also be used as a new indicator to distinguish between normal and fatigue states.

[0076] Matrix analysis showed that the transition probability between microstate categories Map A and Map C increased significantly under fatigue, the one-way transition probability from Map D to Map B increased significantly, and the one-way transition probability from Map A to Map D decreased significantly. Finally, after working memory induced mental fatigue in the subjects, the transition probability of microstate D related to the attention network and microstate B related to the visual network also increased significantly, which is consistent with the findings of Britz et al. The transition probability of microstate A related to the language processing network and microstate D related to attention decreased significantly. Microstate D is usually related to attention and executive function. The reduced probability of transition to microstate D may mean difficulty in maintaining attention, which may be due to the intensity of cognitive load tasks and the differences caused by the fatigue state of the subjects. Khanna et al. showed that the transition between microstates may represent the transition and activation of different neural networks. In the study of neurological diseases, changes in microstates are often not corresponding to the activity or abnormal connection of functional networks. Therefore, mental fatigue may lead to increased transitions between speech processing networks and salience networks, as well as reduced activation of the attention network as a whole.

[0077] The Hurst index result of the present invention is calculated under a long time series. The short time series of 16-32ms proposed by Zhang et al. can explain the ability of neural network changes in a long time range. It can be seen in the figure that the Hurst index in the fatigue state is significantly reduced compared to the normal state. In this way, the short time series in the resting state can also explain the ability of neural network changes in a long time range. The long-term memory ability of the subjects in the fatigue state is worse than that in the normal state. The change in the Hurst index may reflect the decreased coordination and stability of the neural network due to fatigue. Studies have shown that the Hurst index can quantify the degree of relaxation of the brain. Ma et al. studied the difference between the Hurst index of schizophrenia patients and healthy people, and believed that the Hurst index can reflect the difference between the EEG signals of patients and healthy people, and help to better understand the neural mechanism of the disease. Therefore, studying the Hurst index of EEG signals is of great value to EEG analysis.

[0078] In the claims, specification and drawings of the present invention, the term "multiple" refers to two or more than two. Unless otherwise clearly defined, the orientation or positional relationship indicated by the terms "upper" and "lower" is based on the orientation or positional relationship shown in the drawings, which is only for the purpose of more conveniently describing the present invention and making the description process simpler, rather than indicating or implying that the device or element referred to must have the specific orientation described, be constructed and operated in a specific orientation, so these descriptions cannot be understood as limiting the present invention; the terms "connect", "install", "fix" and the like should be understood in a broad sense. For example, "connection" can be a fixed connection between multiple objects, or a detachable connection between multiple objects, or an integral connection; it can be a direct connection between multiple objects, or an indirect connection between multiple objects through an intermediate medium. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood based on the specific circumstances of the above data.

[0079] In the claims, specification and drawings of the present invention, the description of the terms "one embodiment", "some embodiments", "specific embodiments" and the like means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In the claims, specification and drawings of the present invention, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner.

[0080] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for testing the degree of mental fatigue, characterized in that: include: Conducting an experiment on a target, obtaining an experimental result of the target, and extracting electroencephalogram data from the experimental result; Preprocessing the EEG data to obtain pre-analysis EEG data; Analyzing the pre-analysis EEG data to obtain microstate characteristic parameters; Performing real-time monitoring on the target to obtain real-time micro-state characteristic parameters of the target; The degree of mental fatigue is determined based on the real-time micro-state characteristic parameters.

2. The method for testing the degree of mental fatigue according to claim 1, characterized in that: The step of performing an experiment on a target, obtaining an experimental result of the target, and extracting electroencephalogram data from the experimental result includes: Acquiring a normal electroencephalogram of a target in a normal state, collecting first behavioral data of the target, and recording a first result of a subjective fatigue questionnaire of the target; Inducing the target from the normal state to a fatigue state through a mental fatigue inducing task; Acquiring a fatigue electroencephalogram of the target in the fatigue state, collecting second behavioral data of the target, and recording a second result of a subjective fatigue questionnaire of the target; The normal electroencephalogram and the fatigue electroencephalogram are extracted as the electroencephalogram data.

3. The method for testing the degree of mental fatigue according to claim 2, characterized in that: Preprocessing the EEG data to obtain pre-analysis EEG data includes: Performing channel positioning on the electroencephalogram data to obtain first electroencephalogram data; Eliminate four active electrodes in the first electroencephalogram data to obtain second electroencephalogram data; Performing filtering processing on the second electroencephalogram data to obtain third electroencephalogram data; Downsampling the third electroencephalogram data to obtain fourth electroencephalogram data; Performing baseline correction processing on the fourth electroencephalogram data to obtain fifth electroencephalogram data; Eliminate bad segments in the fifth electroencephalogram data and replace bad channels to obtain sixth electroencephalogram data; Separating and removing eye movement components, heart components, and muscle artifact components from the sixth electroencephalogram data by an independent component analysis method to obtain seventh electroencephalogram data; removing artifacts of the seventh electroencephalogram data to obtain eighth electroencephalogram data; Taking the average value of each electrode of the eighth electroencephalogram data as the reference electrode, obtaining ninth electroencephalogram data; The ninth electroencephalogram data is subjected to a bandpass filtering process from 2 Hz to 20 Hz to obtain the pre-analysis electroencephalogram data.

4. The method for testing the degree of mental fatigue according to claim 3, characterized in that: The filtering process on the second electroencephalogram data to obtain third electroencephalogram data includes: Using a FIR digital filter, the 50 Hz power frequency interference in the second electroencephalogram data is removed to obtain the primary filtered electroencephalogram data; The first processed brain wave is subjected to a bandpass filtering process of 0.5 Hz to 30 Hz to obtain the third electroencephalogram data.

5. The method for testing the degree of mental fatigue according to claim 1, characterized in that: The microstate characteristic parameters include: Entropy rate, global explained variance, temporal characteristic parameters, transition probability, and Hurst exponent; The time domain characteristic parameters include average duration, time coverage and occurrence frequency, and the transition probability is the deviation value between the observed transition probability and the expected transition probability.

6. The method for testing the degree of mental fatigue according to claim 5, characterized in that: The analyzing the pre-analysis EEG data to obtain microstate characteristic parameters includes: obtaining a global field power of the pre-analyzed electroencephalogram data; Based on the global field power, obtaining a first time domain point of a maximum value of the global field power; Bringing the first time domain point into the electroencephalogram to obtain a first time series; Performing a second preprocessing on the first time series to obtain a microstate time series; Based on the microstate time series, the microstate characteristic parameters are obtained.

7. The method for testing the degree of mental fatigue according to claim 6, characterized in that: The performing a second preprocessing on the first time series to obtain a microstate time series includes: Applying the T-AAHC algorithm to perform cluster analysis on the first time series to determine the microstate category of the first time series; Based on the microstate category, obtaining a microstate template topographic map; Based on the microstate template topographic map, the microstate time series is acquired.

8. The method for testing the degree of mental fatigue according to claim 6, characterized in that: The step of obtaining the microstate characteristic parameters based on the microstate time series includes: Based on the microstate time series, obtaining the global explained variance, the time domain characteristic parameter and the transition probability; determining static properties of the pre-analyzed EEG data based on the global explained variance; Based on the observed transition probability and the expected transition probability, obtaining a deviation value between the observed transition probability and the expected transition probability; Based on the deviation values, construct a transition probability matrix; Based on the transition probability matrix, the short-range correlation of the pre-analysis electroencephalogram data is determined.

9. The method for testing the degree of mental fatigue according to claim 8, characterized in that: The step of obtaining the microstate characteristic parameters based on the microstate time series further includes: Based on the number of state transitions in the microstate time series, construct a state transition count matrix; Convert the counts in the state transition count matrix into probabilities to construct a state transition probability matrix; Based on the state transition probability matrix, obtaining Shannon entropy; Based on the Shannon entropy and the step size, obtaining the entropy rate; Based on the entropy rate, a dynamic property of the pre-analysis EEG data is determined.

10. The method for testing the degree of mental fatigue according to claim 8, characterized in that: The step of obtaining the microstate characteristic parameters based on the microstate time series further includes: Perform binary classification on the first time series to obtain a binary time series; Accumulating and processing the binary time series to obtain a random walk signal; Performing discrete wavelet transform on the random walk signal to decompose and extract wavelet coefficients, and obtaining a structure function using the wavelet coefficients; Plotting the value and scale of the structure function value to form a log-log plot; Performing linear fitting on the points of the log-log graph to obtain the slope of the fitted line; Obtaining a Hurst exponent based on the slope and the order of the structure function; Based on the Hurst exponent, the long-range correlation of the pre-analysis EEG data is determined.

Images