A psychological state evaluation system based on the asymmetry of physiological signal complexity

Abstract

The present invention discloses a psychological state evaluation system based on the complexity asymmetry of physiological signals, which includes a basic information input module, a physiological signal acquisition module, a data preprocessing module, a feature extraction module, a knowledge base construction module, an intelligent evaluation module, a result feedback module, and a cloud storage module; the feature extraction module utilizes the complexity asymmetry algorithm and other linear and nonlinear analysis methods proposed based on the nonlinear characteristics of physiological signals, and the complexity asymmetry algorithm can extract complexity area asymmetry indexes, complexity slope asymmetry indexes, and complexity aggregation degree asymmetry indexes. The present invention uses portable wearable devices to acquire physiological signals, is suitable for daily use in non-hospital environments such as at home or in laboratory scientific research environments, and can realize the functions of rapid evaluation and long-term monitoring.

Description

Technical Field

[0001] The present invention belongs to the technical fields of physiological signal feature extraction, psychological state evaluation, complexity analysis methods, etc. Specifically, it relates to a psychological state evaluation system based on the complexity asymmetry of physiological signals. Background Art

[0002] In modern society, people's mental health problems are becoming increasingly prominent. With the progress and development of science and technology, people's attention to mental health is also constantly deepening, and the demand for psychological state evaluation, which is closely related to mental health, has increased rapidly. Humans can generate various psychological states, including excitement, joy, satisfaction, calmness, sadness, stress, depression, anxiety, anger, fear, etc., and the internal psychological state keeps changing with the changes of the internal and external environment. A long-term negative psychological state will further lead to serious mental health problems, bringing a great burden to public health resources and the global economy. The identification of psychological states has become an important research content in the field of mental health and has attracted more and more attention from doctors and researchers. At present, there are many psychological state evaluation schemes mainly based on questionnaires, which can help users understand their current mental health status and encourage users to further seek the help of professional technical personnel. However, the evaluation results of the above schemes are often too subjective and cannot give reliable psychological state evaluation results, while the quantitative evaluation based on human physiological signals can overcome the above limitations and achieve the quantitative evaluation of psychological states.

[0003] All kinds of physiological signals output by the human physiological system at all times can reflect the operating state inside the system. As a metric characterizing the balance characteristics of complex systems, complexity asymmetry is applicable to various physiological signals output by the human system. The complexity curve of a time series represents the change process of the entropy value of a time series with increasing coarse-graining degree as the time scale increases. When the time scale is 1, the entropy value represents the uncertainty of the original time series. The increase of the time scale means that the high-frequency information in the original time series is continuously weakened, and the entropy value of the corresponding coarse-grained time series represents the uncertainty of the remaining information in the series. According to previous studies on multi-scale entropy and its various improved algorithms, the complexity curves of various physiological signals generated by healthy living systems are often similar and usually consist of two parts with different slopes. In the small time scale range to the left of the critical point, its entropy value corresponds to the complexity of the short-range high-frequency information of the time series; in the large time scale range to the right of the critical point, its entropy value corresponds to the complexity of the long-range low-frequency information. The ratio of the high-frequency information complexity to the low-frequency information complexity represents the balance between the short-range high-frequency and long-range low-frequency complexities of this complex system. Therefore, complexity asymmetry is proposed to characterize this system balance metric.

[0004] Professional large-scale physiological signal acquisition devices are often not suitable for monitoring large samples and non-hospital or laboratory environments. Portable and low-load consumer-grade wearable devices can achieve long-term home monitoring of one-dimensional or multi-dimensional physiological signals of the human body, and can form long-term electronic health records by continuously collecting various physiological signals of users. Long-term tracking of the changes in users' mental states and intervention effects is an important means to maintain a healthy and stable mental state. Digital health has become a new development trend in mental health management. Based on intelligent wearable devices that can acquire human physiological signals, it is possible to further extract relevant features of physiological signals as reliable and effective criteria for mental state evaluation. The combination of wearable devices and mental state evaluation APPs in mobile devices can better achieve the goals of intelligent evaluation and mental health management, and can assist professional technicians in the early detection, early diagnosis, and early treatment of related mental abnormalities. The software and hardware combined mental state intelligent evaluation system can timely remind users to carry out self-management or receive professional intervention, assist professional technicians in evaluation prediction and tracking management, achieve personalized and precise intervention, and is conducive to the stability of users' long-term mental health status.

[0005] In summary, it has very important practical significance to study a mental state evaluation system based on the asymmetry of physiological signal complexity. Summary of the Invention

[0006] Object of the Invention: The present invention provides a mental state evaluation system based on the asymmetry of physiological signal complexity. According to the one-dimensional or multi-dimensional physiological signals of the human body acquired by an integrated intelligent wearable device, feature extraction related to emotions and psychology is realized to evaluate the mental state of users, and finally the evaluation results are fed back to the users.

[0007] Technical Solution: The present invention provides a mental state evaluation system based on the asymmetry of physiological signal complexity, including a basic information input module, a physiological signal acquisition module, a data preprocessing module, a feature extraction module, a knowledge base construction module, an intelligent evaluation module, a result feedback module, and a cloud storage module;

[0008] The basic information input module is used to collect the basic information and medical history background information of users for health management and to assist in the evaluation of different mental states subsequently, and send the acquired information to the intelligent evaluation module;

[0009] The physiological signal acquisition module is used to collect one-dimensional or multi-dimensional physiological signals of the human body for a preset duration by using a wearable portable device, and send the acquired information to the data preprocessing module;

[0010] The data preprocessing module, according to a preset program, completes preprocessing of different categories or sequences of one-dimensional or multi-dimensional physiological signals of different types of the human body, and sends the preprocessed results to the feature extraction module;

[0011] The feature extraction module is used to extract the asymmetry features of the large time scale and small time scale complexity of the preprocessed one-dimensional or multi-dimensional physiological signals by using the complexity asymmetry feature extraction algorithm, and extract other features by using other linear and non-linear analysis methods, and send the feature extraction results to the intelligent evaluation module;

[0012] The knowledge base construction module is used to train the mental state evaluation model by using the data set containing the one-dimensional or multi-dimensional physiological signals and other relevant basic information of individuals with different mental states;

[0013] The intelligent evaluation module is used to evaluate the mental state of the user based on the basic information and physiological characteristics of the user by using the mental state evaluation model, and send the result to the result feedback module;

[0014] The result feedback module is used to feedback the current state and change condition result report of the user's psychology to the user according to the mental state evaluation result obtained by the intelligent evaluation module;

[0015] The cloud storage module is used to store the existing mental state evaluation data set, the mental state evaluation model and the data of new users, and continuously improve the prediction accuracy of the mental state evaluation model by using the data of new users.

[0016] Further, the physiological signal acquisition module completes the acquisition of the user's one-dimensional or multi-dimensional physiological signals according to the user's needs and selections, mainly including electroencephalogram, electrocardiogram, electromyogram, pulse wave, acceleration, skin electricity and skin temperature.

[0017] Further, the preprocessing steps include data segmentation, invalid data cleaning, filtering, denoising, segmentation, bad segment elimination and signal quality assessment.

[0018] Further, the process of the feature extraction module using the complexity asymmetry feature extraction algorithm to extract the asymmetry features is as follows:

[0019] S1: Calculate the multi-scale complexity of the one-dimensional physiological signal by using the multi-scale entropy and its improved algorithm; calculate the multi-scale complexity of the multi-dimensional physiological signal by using the multi-variable multi-scale entropy and its improved algorithm;

[0020] S2: According to the sampling rate and the number of data points of various physiological signals in the healthy life system, use the mean value theorem to find the stationary point position, or find the position with the largest curve curvature, for adaptively determining the critical point of the complexity curve;

[0021] S3: Based on the critical point of the complexity curve, calculate the complexity asymmetry according to the complexity characteristics on both sides of the critical point, including the complexity area asymmetry index, the complexity slope asymmetry index and the complexity aggregation degree asymmetry index, etc.

[0022] Furthermore, the specific calculation process of the complexity area asymmetry index is as follows:

[0023] Calculate the area under the curve area within the small-scale range to the left of the critical point 1 and the area under the curve area within the large-scale range to the right of the critical point 2 ; Based on the area 1 and area 2 of the complexity curve, a complexity area asymmetry index is proposed to characterize the complexity asymmetry between the small time scale and the large time scale of the system. This index is calculated by the following formula:

[0024] Complexity area asymmetry index = ±(area 1 - area 2 ) / (area 1 + area 2 ) or

[0025]

[0026] where area 1 represents the complexity of the fine scale, i.e., high-frequency information, of the complex system, and area 2 represents the complexity of the rough scale, i.e., low-frequency information, of the complex system.

[0027] Furthermore, the specific calculation process of the complexity slope asymmetry index is as follows:

[0028] Calculate the curve slope a 1 within the small-scale range to the left of the critical point and the curve slope a 2 within the large-scale range to the right of the critical point; Based on the a 1 and a 2 of the complexity curve, a complexity slope asymmetry index is proposed to characterize the complexity asymmetry between the small time scale and the large time scale of the system. This index is calculated by the following formula:

[0029] Complexity slope asymmetry index = ±(a 1 - a 2 ) / (a 1 + a 2 ) or

[0030]

[0031] where a 1 represents the complexity change rate of the fine scale of the complex system, and a 2 represents the complexity change rate of the rough scale of the complex system.

[0032] Further, the specific calculation process of the complexity aggregation degree asymmetry index is as follows:

[0033] Calculate the standard deviation b of the curve within a small scale range to the left of the critical point 1 and the standard deviation b of the curve within a large scale range to the right of the critical point 2 ; Based on b 1 and b 2 of the complexity curve, an asymmetry index of complexity aggregation degree is proposed to characterize the asymmetry of complexity between the small time scale and the large time scale of the system. This index is calculated by the following formula:

[0034] Complexity aggregation degree asymmetry index = ±(b 1 -b 2 ) / (b 1 +b 2 ) or

[0035]

[0036] where b 1 represents the concentration degree of the complexity of the fine scale of the complex system, and b 2 represents the concentration degree of the complexity of the rough scale of the complex system.

[0037] Beneficial effects: Compared with the prior art, the beneficial effects of the present invention are as follows: The present invention can use intelligent wearable devices to achieve short-term or long-term monitoring of one-dimensional or multi-dimensional physiological signals including but not limited to electroencephalogram, electrocardiogram, pulse, etc. according to user needs; the present invention can use the complexity asymmetry algorithm and other linear and non-linear analysis methods to extract features related to emotions, psychology, etc. in human physiological signals to evaluate the psychological state of users; the present invention can combine user historical data to feedback the change status of the psychological state of users and show the intervention effect; the present invention can use the cloud storage module to establish long-term health records and management of the psychological state of users for long-term evaluation, prediction and tracking management; the present invention can use the newly added user information in the cloud storage module to continuously optimize the prediction and evaluation methods of various psychological states in the intelligent evaluation model. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 is a schematic framework diagram of the present invention;

[0039] Figure 2 is a working flow chart of the psychological state evaluation system;

[0040] Figure 3 is a schematic diagram of the user operation interface of the basic information input module;

[0041] Figure 4 is a flow chart of electroencephalogram signal preprocessing;

[0042] Figure 5 It is a flowchart for the preprocessing of electrocardiogram signals;

[0043] Figure 6 It is a flowchart of the algorithm for calculating the complexity asymmetry feature of physiological signals;

[0044] Figure 7 It is an example result diagram of the complexity area asymmetry feature;

[0045] Figure 8 It is a flowchart for the construction of the knowledge base module;

[0046] Figure 9 It is a schematic step flowchart of the intelligent evaluation method;

[0047] Figure 10 It is a schematic diagram of the user interface of the feedback module. Specific implementation manners

[0048] The present invention will be further described in detail below with reference to the accompanying drawings:

[0049] The present invention provides a psychological state evaluation system based on the complexity asymmetry of physiological signals. As Figure 1 shown, it mainly includes 8 modules, namely the basic information input module, the physiological signal acquisition module, the data preprocessing module, the feature extraction module, the knowledge base construction module, the intelligent evaluation module, the result feedback module, and the cloud storage module. The working process of the psychological state evaluation system is as Figure 2 shown.

[0050] The basic information input module is used to collect the user's basic information and the background information of the medical history, for health management and assisting in the evaluation of different psychological states, and send the above content to the intelligent evaluation module. Specifically, this module can obtain the user's basic information and the background information of the medical history. The former includes name, gender, date of birth, height, weight, marital status, fertility status, ethnicity, education level, occupation, and current residence, etc.; the latter includes chief complaint, current medical history, past medical history, family history, and allergy history, etc. The specific user operation interface is as Figure 3 shown. The system will allocate a new account for new users internally, and historical users can directly select the existing account to log in. This basic information input module can use the internal APP of the intelligent mobile device to complete the function of inputting basic information, and transmit the user's relevant information to the cloud storage module as the basic information features of the individual, for the training and evaluation prediction of the psychological state evaluation model.

[0051] A physiological signal acquisition module is used to collect one-dimensional or multi-dimensional physiological signals of the human body for a preset duration by using a wearable portable device, and send the above content to the data preprocessing module. Specifically, this module can include various types of intelligent wearable devices, including a portable multi-channel electrode system and a smart bracelet, etc.; in addition, according to the user's needs and choices, the acquisition of one-dimensional or multi-dimensional physiological signals of the user can be completed, which can include electroencephalogram (EEG), electrocardiogram (ECG), electromyogram (EMG), pulse wave, acceleration, galvanic skin response (GSR), and skin temperature, etc. After the physiological signals are transmitted to the APP through wired or wireless transmission, they are synchronously uploaded to the cloud storage module for subsequent processing and analysis.

[0052] A data preprocessing module is used to complete preprocessing steps such as segmentation, cleaning, and signal quality assessment of one-dimensional or multi-dimensional physiological signals that can reflect the physiological state of the human body, and send the above content to the feature extraction module. This module can complete different types or sequences of preprocessing steps for different types of human physiological signals according to a preset program. The preprocessing steps can include data segmentation, cleaning of invalid data, filtering, denoising, segmentation, rejection of bad segments, and signal quality assessment, etc. As Figure 4 , Figure 5 shown, taking the preprocessing of EEG and ECG signals as an example, the EEG preprocessing process can include rejecting bad leads, resampling to the required sampling rate, 50 / 60 Hz notch filtering to remove power frequency interference, high-pass filtering and low-pass filtering at specific frequencies to remove baseline drift and high-frequency noise, removing eye movement artifacts, data segmentation, rejecting bad segments, etc.; the ECG preprocessing process can include rejecting bad leads, resampling to the required sampling rate, 50 / 60 Hz notch filtering to remove power frequency interference, high-pass filtering and low-pass filtering at specific frequencies to remove baseline drift and high-frequency noise, rejecting bad segments, extracting RR interval data, data segmentation, rejecting outliers, etc. This data preprocessing module completes the above functions in the cloud storage module by using the stored one-dimensional or multi-dimensional physiological signals.

[0053] A feature extraction module is used to extract the asymmetry features of the large time scale and small time scale complexity of the preprocessed one-dimensional or multi-dimensional physiological signals by using the complexity asymmetry algorithm, and extract other features by using other linear and nonlinear analysis methods, and send the above content to the intelligent evaluation module. This module can use the complexity asymmetry index algorithm to extract the asymmetry features of the complexity of the high-frequency components and the complexity of the low-frequency components in one-dimensional or multi-dimensional physiological signals. In addition, other linear and nonlinear features in different physiological signals can be extracted. Other linear feature indicators can include statistical features in the time domain, band energy features in the frequency domain, and time-frequency combined analysis features, such as the maximum value, minimum value, mean value, standard deviation, variance, median, percentile, skewness, kurtosis, zero-crossing rate, and energy in the time domain, the band energy of Theta, Alpha, Beta, etc. in the frequency domain, and the features obtained by time-frequency combined analysis methods such as short-time Fourier transform and wavelet transform; other nonlinear feature indicators can include Hurst exponent, Lyapunov exponent, detrended fluctuation analysis exponent, fractal dimension, complexity, and various other entropies, etc. This feature extraction module is in the cloud storage module and uses the stored preprocessed one-dimensional or multi-dimensional physiological signals to complete the above functions. As Figure 6 shown, the flow of the complexity asymmetry feature extraction algorithm is as follows:

[0054] First, use multi-scale entropy and its improved algorithms (including multi-scale sample entropy, multi-scale fuzzy entropy, multi-scale distribution entropy, etc.) to calculate the multi-scale complexity of one-dimensional physiological signals; use multi-variable multi-scale entropy and its improved algorithms to calculate the multi-scale complexity of multi-dimensional physiological signals.

[0055] For various physiological signals generated by a healthy life system, the complexity curve often consists of two parts with different slopes. Usually, the absolute slope of the complexity curve in the small time scale part is larger, while the absolute slope of the complexity curve in the large time scale part is smaller. According to the sampling rate and the number of data points of various physiological signals in the healthy life system, use the mean value theorem to find the stationary point position, or find the position with the maximum curve curvature, for adaptively determining the critical point of the complexity curve.

[0056] Based on the critical point of the complexity curve, calculate the complexity asymmetry according to the complexity characteristics on both sides of the critical point, including complexity area asymmetry index, complexity slope asymmetry index, complexity aggregation degree asymmetry index, etc. The calculation methods of the three types of complexity asymmetry features are as follows:

[0057] Complexity area asymmetry index: As Figure 7 shown, calculate the area under the curve area 1 in the small scale range on the left side of the critical point and the area under the curve area 2 in the large scale range on the right side of the critical point. Based on the area 1and area 2 , a complexity area asymmetry index is proposed to characterize the complexity asymmetry between the small time scale and the large time scale of the system. This index is calculated by the following formula:

[0058] Complexity area asymmetry index = ±(area 1 - area 2 ) / (area 1 + area 2 ) or

[0059]

[0060] where area 1 represents the complexity of the fine scale, i.e., high-frequency information, of the complex system, and area 2 represents the complexity of the rough scale, i.e., low-frequency information, of the complex system.

[0061] Complexity slope asymmetry index: Calculate the curve slope a 1 within the small scale range to the left of the critical point and the curve slope a 2 within the large scale range to the right of the critical point. Based on the a 1 and a 2 of the complexity curve, a complexity slope asymmetry index is proposed to characterize the complexity asymmetry between the small time scale and the large time scale of the system. This index is calculated by the following formula:

[0062] Complexity slope asymmetry index = ±(a 1 - a 2 ) / (a 1 + a 2 ) or

[0063]

[0064] where a 1 represents the complexity change rate of the fine scale of the complex system, and a 2 represents the complexity change rate of the rough scale of the complex system.

[0065] Complexity aggregation degree asymmetry index: Calculate the curve standard deviation b 1 within the small scale range to the left of the critical point and the curve standard deviation b 2 within the large scale range to the right of the critical point. Based on the b 1 and b 2 of the complexity curve, a complexity aggregation degree asymmetry index is proposed to characterize the complexity asymmetry between the small time scale and the large time scale of the system. This index is calculated by the following formula:

[0066] Complexity aggregation degree asymmetry index = ±(b1 -b 2 ) / (b 1 +b 2 ) or

[0067]

[0068] where b 1 represents the concentration degree of the fine-scale complexity of the complex system, and b 2 represents the concentration degree of the rough-scale complexity of the complex system.

[0069] Knowledge base construction module, used to train a mental state evaluation model by using a data set containing one-dimensional or multi-dimensional physiological signals and other relevant basic information of individuals with different mental states. Specifically, this module can construct different mental state evaluation models, and the categories of mental states that can be evaluated include excitement, joy, satisfaction, calmness, sadness, stress, depression, anxiety, anger, fear, etc. In addition, more data sets of individuals with different types of mental states can be incorporated into the knowledge base in the future, and the mental state evaluation model can be continuously updated according to the information of new users. The knowledge base construction process can be completed in the cloud storage module. By continuously uploading the data set to the cloud storage module for synchronization, the training of the mental state evaluation model is completed. The knowledge base construction process can be implemented in a PC using MATLAB, C++, Python or other languages. Specifically, as Figure 8 shown, the internal knowledge base construction of this module includes the following steps:

[0070] 1) Obtain one-dimensional or multi-dimensional physiological signals, basic information, and medical history background information of a large number of individuals with various mental states, and complete case screening and sample labeling.

[0071] 2) Complete the data preprocessing process for the above different types of physiological signals, and this process can be completed using the data preprocessing module.

[0072] 3) Extract the complexity asymmetry index and other linear and non-linear indexes of one-dimensional or multi-dimensional physiological signals of individuals with various mental states, and this process can be completed using the feature extraction module.

[0073] 4) Screen the features with strong correlation with specific mental states, determine the physiological signal categories with the best evaluation effect corresponding to different mental states, and retain the basic information or medical history background information related to different mental states.

[0074] 5) Use the screened physiological signal features, basic information, and medical history background information features to train the mental state evaluation model, and common machine learning models in multi-class supervised learning can be used, including support vector machines, random forests, naive Bayes, K-nearest neighbors, and ensemble learning, etc.

[0075] 6) Evaluate the effects of the above-mentioned multiple types of machine learning models. Use the k-fold cross-validation method for model evaluation, and use multiple common metrics such as the ROC curve, F1-score, specificity, sensitivity, recall, precision, accuracy, confusion matrix, etc. to measure the model performance.

[0076] 7) Considering the above multiple metrics for measuring model performance comprehensively, select the machine learning model with the best psychological state evaluation effect.

[0077] 8) The constructed psychological state evaluation model can be used to evaluate the psychological state of users in the intelligent evaluation module, and further output the psychological state result report of the patient.

[0078] The intelligent evaluation module, as Figure 9 shown, is used to evaluate the psychological state of users based on the basic information and physiological characteristics of users, using the psychological state evaluation model, and send the results to the result feedback module. Specifically, this module takes the one-dimensional or multi-dimensional physiological signal characteristics of users (including complexity asymmetry and other linear and non-linear characteristics), basic information, and background information characteristics of the medical history as the input of the psychological state evaluation model trained by the knowledge base construction module, and the model outputs the psychological state result of this user. This intelligent evaluation module uses the one-dimensional or multi-dimensional physiological signal characteristics stored in the cloud storage module to complete the above functions, and sends the corresponding result report to the internal APP of the intelligent mobile device.

[0079] The result feedback module, as Figure 10 shown, is used to feedback the current state and change status result report of the psychology to the user according to the psychological state evaluation result obtained by the intelligent evaluation module. Specifically, this module can feedback the current state of the psychology to the user, and remind the user to intervene in their own psychological state in time. It can also combine the user's historical data to feedback the change status of the user's psychological state and show the intervention effect. In addition, it can assist the evaluation prediction and tracking management of professional and technical personnel. This result feedback module can visually feedback the intelligent evaluation result to the user through the internal APP interface of the intelligent mobile device.

[0080] The cloud storage module is used to store the existing psychological state evaluation data set, psychological state evaluation model, and data of new users, and continuously improve the prediction accuracy of the psychological state evaluation model using the data of new users. Specifically, this module stores computer programs, the constructed knowledge base, and various historical information of users.

Claims

1. A psychological state evaluation system based on the asymmetry of physiological signal complexity, characterized in that, it includes a basic information input module, a physiological signal acquisition module, a data preprocessing module, a feature extraction module, a knowledge base construction module, an intelligent evaluation module, a result feedback module, and a cloud storage module; The basic information input module is used to collect the user's basic information and medical history background information for health management and to assist in the evaluation of different psychological states subsequently, and send the acquired information to the intelligent evaluation module; The physiological signal acquisition module is used to collect one-dimensional or multi-dimensional physiological signals of the human body for a preset duration by using wearable portable devices, and send the acquired information to the data preprocessing module; The data preprocessing module, according to a preset program, completes preprocessing of different categories or time series of one-dimensional or multi-dimensional physiological signals of the human body of different types, and sends the preprocessed results to the feature extraction module; The feature extraction module is used to extract the asymmetry features of the large time scale and small time scale complexity of the preprocessed one-dimensional or multi-dimensional physiological signals by using the complexity asymmetry feature extraction algorithm, and send the feature extraction results to the intelligent evaluation module; The knowledge base construction module is used to train a psychological state evaluation model by using a data set containing one-dimensional or multi-dimensional physiological signals, age, gender, and symptom information of individuals in different psychological states; The intelligent evaluation module is used to evaluate the user's psychological state based on the user's basic information and physiological signals by using the psychological state evaluation model, and send the results to the result feedback module; The result feedback module is used to feedback the current state and change status result report of the user's psychology to the user according to the psychological state evaluation result obtained by the intelligent evaluation module; The cloud storage module is used to store the existing psychological state evaluation data set, the psychological state evaluation model, and the data of new users, and continuously improve the prediction accuracy of the psychological state evaluation model by using the data of new users; The process of the feature extraction module extracting the asymmetry features by using the complexity asymmetry feature extraction algorithm is as follows: S1: Calculate the multi-scale complexity of one-dimensional physiological signals by using multi-scale sample entropy, multi-scale permutation entropy, and multi-scale fuzzy entropy; calculate the multi-scale complexity of multi-dimensional physiological signals by using multi-variable multi-scale entropy and its improved algorithm; S2: According to the sampling rate and the number of data points of various physiological signals in the healthy life system, use the mean value theorem to find the stationary point position, or find the position with the maximum curve curvature to adaptively determine the critical point of the complexity curve; S3: Based on the critical point of the complexity curve, calculate the complexity asymmetry according to the complexity characteristics on both sides of the critical point, including the complexity area asymmetry index, the complexity slope asymmetry index, and the complexity aggregation degree asymmetry index.

2. The psychological state evaluation system based on the asymmetry of physiological signal complexity according to claim 1, characterized in that, The physiological signal acquisition module completes the acquisition of the user's one-dimensional or multi-dimensional physiological signals according to the user's needs and selections, mainly including electroencephalogram, electrocardiogram, electromyogram, pulse wave, acceleration, skin electricity, and skin temperature.

3. The mental state evaluation system based on the complexity asymmetry of physiological signals according to claim 1, wherein, the preprocessing of one-dimensional or multi-dimensional physiological signals of different types of human bodies for different categories or time series includes data segmentation, invalid data cleaning, filtering, denoising, segmentation, bad segment removal, and signal quality evaluation.

4. The mental state evaluation system based on the complexity asymmetry of physiological signals according to claim 1, wherein, the specific calculation process of the complexity area asymmetry index is as follows: Calculate the area under the curve within a small scale range to the left of the critical point area 1 and the area under the curve within a large scale range to the right of the critical point area 2 ; Based on the area of the complexity curve 1 and area 2 , a complexity area asymmetry index is proposed to characterize the complexity asymmetry between the small time scale and the large time scale of the system. This index is calculated by the following formula: Complexity area asymmetry index = ±(area 1 -(area 2 ) / (area 1 +area 2 ) or Among them, area 1 represents the complexity of the fine scale, i.e., high-frequency information, of the complex system, and area 2 represents the complexity of the rough scale, i.e., low-frequency information, of the complex system.

5. The mental state evaluation system based on the complexity asymmetry of physiological signals according to claim 1, wherein, the specific calculation process of the complexity slope asymmetry index is as follows: Calculate the curve slope a within a small scale range to the left of the critical point 1 and the curve slope a within a large scale range to the right of the critical point 2 ; Based on a 1 and a 2 of the complexity curve, a complexity slope asymmetry index is proposed to characterize the complexity asymmetry between the small time scale and the large time scale of the system. This index is calculated by the following formula: Complexity slope asymmetry index = ±(a 1 - a 2 ) / (a 1 + a 2 ) or Among them, a 1 represents the complexity change rate of the fine scale of the complex system, and a 2 represents the complexity change rate of the coarse scale of the complex system.

6. The mental state evaluation system based on the complexity asymmetry of physiological signals according to claim 1, wherein, the specific calculation process of the complexity aggregation degree asymmetry index is as follows: Calculate the standard deviation b of the curve within a small scale range on the left side of the critical point 1 and the standard deviation b of the curve within a large scale range on the right side of the critical point 2 ; Based on b 1 and b 2 of the complexity curve, an asymmetry index of complexity aggregation degree is proposed to characterize the asymmetry of system complexity between small time scales and large time scales. This index is calculated by the following formula: Complexity aggregation degree asymmetry index = ±(b 1 -b 2 ) / (b 1 +b 2 ) or Among them, b 1 represents the degree of concentration of the fine-scale complexity of the complex system, and b 2 represents the degree of concentration of the coarse-scale complexity of the complex system.

Images