Depression identification method based on multi-dimensional feature integration

Abstract

The invention discloses a depression identification method based on multi-dimensional feature integration, and relates to the technical field of mental health, and the method comprises the steps: S1, playing different types of music, including happy, sadness and passion music, of a to-be-identified patient, obtaining brain wave data of the to-be-identified patient based on a brain wave device, and storing the brain wave data in a database; constructing an electroencephalogram judgment model based on the big data, training the electroencephalogram judgment model through an electroencephalogram data set of a historical patient, inputting the electroencephalogram data into the electroencephalogram judgment model, and outputting to obtain a preliminary recognition result; s2, collecting facial feature data of a patient to be recognized, and uploading the obtained data to a recognition module for storage; and S3, acquiring behavior data of the patient to be identified. According to the method, the multi-dimensional features are fused, the state of the patient can be reflected more comprehensively, the accuracy of depression identification is improved, the accuracy of judgment is improved, the depression patient is treated earlier, and targeted intervention measures are taken.

Description

Technical Field

[0001] The present invention relates to the field of mental health technology, and in particular to a depression identification method based on multi-dimensional feature integration. Background Art

[0002] Depression is a common mental disorder that brings great pain and distress to patients. The symptoms of depression are diverse, mainly including emotional, cognitive, and physical aspects. Emotionally, patients often feel persistent sadness, loss, helplessness, lose interest and fun in life, and lack pleasure. They may cry for no reason, and the state of depression lasts for a long time, and it is difficult to improve through self-regulation, which seriously affects the quality of life and social function of patients. At present, the diagnosis of depression mainly depends on the subjective judgment and questionnaire survey of clinicians, lacking objective and accurate identification methods. The process of identifying depression is highly subjective and relies on the experience and subjective judgment of doctors, which is prone to misdiagnosis and missed diagnosis. In this regard, we propose a depression identification method based on multidimensional feature integration. Summary of the invention

[0003] In order to solve the above technical problems, a depression identification method based on multi-dimensional feature integration is provided. This technical solution solves the above problems of misdiagnosis and missed diagnosis.

[0004] To achieve the above objectives, the technical solution adopted by the present invention is: a depression identification method based on multidimensional feature integration, the identification steps are: S1. Play different kinds of music for the patient to be identified, including happy, sad and passionate music. Obtain the EEG data of the patient to be identified based on the EEG device, build an EEG judgment model based on big data, train the EEG judgment model through the EEG data set of historical patients, input the EEG data into the EEG judgment model, and output a preliminary recognition result; S2, collect facial feature data of the patient to be identified, and upload the acquired data to the identification module for storage; S3, obtaining the behavior data of the patient to be identified, and uploading the behavior data to the identification module for storage; S4. Analyze the preliminary recognition results in combination with the facial feature data and the behavioral data, and judge the preliminary recognition results to determine their accuracy. If yes, output depression, otherwise repeat the above steps to continue the judgment; S5. After identifying depression, the doctor will provide psychological counseling to the patient based on the patient's condition.

[0005] Preferably, in step S1, different types of music are played in sequence for the patient to be identified, including happy, sad and passionate music. Each type of music is paused after playing for a period of time to give the patient to be identified a buffer time. In the process of playing music, the EEG data of the patient to be identified is acquired in real time based on the EEG device, and the changes in the patient's EEG signals under different music stimulations are recorded, including frequency, amplitude and phase characteristics; the EEG judgment model is trained on the data set based on deep learning, and the EEG data under different music stimulations are analyzed. After the training is completed, the EEG data is input into the model and the results are output.

[0006] Preferably, the EEG judgment model training process assumes that the model is in the form of: M=(x, y), where x is the data input into the model, y is the parameter value of the model's minimized loss function, and the historical patient EEG data set is: where a i is the EEG data combination of the i-th historical patient under different music stimulations, b i is the corresponding label to determine whether depression exists; The formula for minimizing the loss function is defined as: Where y is the parameter value of the model's minimized loss function, arg represents the value of the variable when the formula reaches the minimum value, n is the number of EEG data combinations, M is the EEG judgment model, and the formula is used to obtain the parameter value of y.

[0007] Preferably, the model is optimized after training, and the hyperparameters of the model are optimized using a hyperparameter optimization method, and the best hyperparameter settings are obtained based on a cross-validation method.

[0008] Preferably, in step S2, based on the image acquisition device, the patient to be identified is photographed at multiple angles and multiple time periods, the photographed images are uploaded to the recognition module, and based on the data transmission channel, the collected facial feature data of the patient to be identified is transmitted in an encrypted form, and the data is classified and stored; the recognition module uses image analysis technology to pre-process the collected images, extract key feature points of the face, including the position, shape and proportion of the eyes, nose and mouth, analyze the facial feature data, and store the analysis results.

[0009] Preferably, in step S3, the behavioral data of the patient to be identified is obtained through monitoring equipment, and the patient's limb movements in different scenarios are recorded. After the behavioral data is collected, it is uploaded based on the transmission channel, and the joints of the limbs are tracked based on computer vision technology, the motion trajectory is drawn, and the movement frequency of the movement is analyzed to obtain the analysis data results.

[0010] Preferably, the action frequency analysis of the patient to be identified sets the time period as T, and the number of occurrences of a specific action is c, which is expressed as: within the time period T, the number of occurrences c is obtained by counting, and the frequency k of the specific action is calculated based on the total number of actions and the number of occurrences of the specific action, and the behavior feature value F is evaluated. The evaluation formula is: F=jk+h Where F is the behavioral characteristic value obtained by evaluation, and j and h are constants.

[0011] Preferably, in step S4, the judgment step sets the output result of the preliminary recognition to O, the facial feature data to P, the behavior data to F, and the analysis function R (P, F), which represents the process of combining the facial feature data and the behavior data for analysis. The judgment process is in the form of: Among them, Y represents the judgment of depression, N represents the judgment of not depression, indicating that there is an error, then continue to monitor, identify and judge, S is the output result; if the preliminary identification result is that there is a tendency of depression, after analyzing the facial feature data and behavioral data, it is also concluded that there is a tendency of depression, then the final result is depression; if the analysis result is inconsistent with the preliminary identification result, re-collect data, perform preliminary identification and combine facial features and behavioral data for analysis steps until an accurate result is obtained.

[0012] Preferably, after the patient is identified as suffering from depression in step S5, the physician provides psychological counseling to the patient, creates a safe, comfortable and private environment, allows the patient to relax, open up, communicate, analyze and discuss the causes of depression, and take measures to address the causes.

[0013] Preferably, during the psychological counseling process, the physician uses cognitive behavioral therapy to help the patient identify and correct negative thinking patterns and behavioral habits, cultivate positive thinking patterns, and encourage the patient to take positive actions.

[0014] Compared with the prior art, the present invention has the following beneficial effects: The present invention obtains and analyzes the brain waves of the patient to be identified by playing music, gives a preliminary analysis result, and combines the collected behavior and facial features to conduct in-depth analysis and judgment on the results of the preliminary analysis, gives an accurate judgment result, and integrates multi-dimensional features. It can more comprehensively reflect the patient's condition, improve the accuracy of depression identification, improve the accuracy of judgment, treat depression patients earlier, and take targeted intervention measures. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 This is a flow chart of the depression identification steps of the present invention. DETAILED DESCRIPTION

[0016] The following description is used to disclose the present invention so that those skilled in the art can implement the present invention. The preferred embodiments described below are only examples, and those skilled in the art may think of other obvious variations.

[0017] Reference Figure 1 As shown in the figure, the depression recognition method based on multi-dimensional feature integration has the following recognition steps: S1. Play different kinds of music for the patient to be identified, including happy, sad and passionate music. Obtain the EEG data of the patient to be identified based on the EEG device, build an EEG judgment model based on big data, train the EEG judgment model through the EEG data set of historical patients, input the EEG data into the EEG judgment model, and output a preliminary recognition result; S2, collect facial feature data of the patient to be identified, and upload the acquired data to the identification module for storage; S3, obtaining the behavior data of the patient to be identified, and uploading the behavior data to the identification module for storage; S4. Analyze the preliminary recognition results in combination with the facial feature data and the behavioral data, and judge the preliminary recognition results to determine their accuracy. If yes, output depression, otherwise repeat the above steps to continue the judgment; S5. After identifying depression, the doctor will provide psychological counseling to the patient based on the patient's condition.

[0018] This application can obtain the patient's response to different emotional stimuli from a neurophysiological level by playing different types of music and collecting brain wave data, providing objective physiological indicators for the identification of depression. The EEG judgment model is based on big data training and integrates the data characteristics of many historical patients, making the initial identification results more scientific and reliable; Different patients have unique responses to music, facial features, and behavioral performance. This method comprehensively considers multi-dimensional data and can analyze the specific situation of each patient to achieve personalized diagnosis, avoiding the limitations of generalizations in traditional methods. The collection and analysis of multi-dimensional data can enable doctors to have a more comprehensive understanding of the patient's physical and mental state. They can not only know whether the patient suffers from depression, but also gain an in-depth understanding of the patient's emotional triggers, behavioral patterns, and potential psychological needs.

[0019] In step S1, different types of music are played in turn for the patient to be identified, including happy, sad and passionate music. Each type of music is paused after playing for a period of time to give the patient to be identified buffer time. During the music playing process, the EEG data of the patient to be identified is obtained in real time based on the EEG device, and the changes in the patient's EEG signals under different music stimulations are recorded, including frequency, amplitude and phase characteristics; the EEG judgment model is based on deep learning to learn and train the data set, and analyze the EEG data under different music stimulations. After the training is completed, the EEG data is input into the model and the results are output.

[0020] By playing different types of music, this application can observe the patient's EEG response under different emotional stimulations, enriching the data source. Happy, sad and passionate music can stimulate different emotional states respectively, so as to have a more comprehensive understanding of the patient's neural response pattern, record the frequency, amplitude and phase characteristics of EEG signals, and provide multi-dimensional physiological indicators for the identification of depression. These characteristics can reflect the brain's activity state and the state of neural transmission, which is helpful for a more in-depth analysis of the patient's brain function.

[0021] The EEG judgment model training process assumes that the model is in the form of: M = (x, y), where x is the data input into the model, and y is the parameter value of the model's minimized loss function. The historical patient EEG data set is: where a i is the EEG data combination of the i-th historical patient under different music stimulations, b i is the corresponding label to determine whether depression exists; The formula for minimizing the loss function is defined as: Where y is the parameter value of the model's minimized loss function, arg represents the value of the variable when the formula reaches the minimum value, n is the number of EEG data combinations, M is the EEG judgment model, and the formula is used to obtain the parameter value of y.

[0022] The model is optimized after training. The hyperparameter optimization method is used to optimize the hyperparameters of the model, and the best hyperparameter settings are obtained based on cross-validation.

[0023] In step S2, based on the image acquisition device, the patient to be identified is photographed from multiple angles and at multiple time periods, the photographed images are uploaded to the recognition module, and based on the data transmission channel, the collected facial feature data of the patient to be identified is transmitted in an encrypted form, and the data is classified and stored; the recognition module uses image analysis technology to pre-process the collected images, extract key feature points of the face, including the position, shape and proportion of the eyes, nose and mouth, analyze the facial feature data, and store the analysis results.

[0024] The facial feature data analyzed in this application is based on image recognition and is a prior art. It will not be elaborated here. By shooting at multiple angles and multiple time periods, the facial features of the patient in different states can be captured, avoiding the limitations that may be brought about by single-angle or instantaneous shooting. This can more comprehensively reflect the patient's facial expression changes and feature stability, and provide a rich data basis for subsequent analysis. Facial features are closely related to emotional state. Analysis of facial feature data can provide important clues for the identification of depression. For example, patients with depression may have reduced facial expressions, dull eyes, and drooping corners of the mouth. The analysis of these features can assist in determining whether the patient is suffering from depression and improve the accuracy of identification.

[0025] In step S3, the behavioral data of the patient to be identified is obtained through monitoring equipment, and the patient's body movements in different scenarios are recorded. After the behavioral data is collected, it is uploaded based on the transmission channel, and the joints of the limbs are tracked based on computer vision technology, the motion trajectory is drawn, and the movement frequency of the movement is analyzed to obtain the analysis data results.

[0026] This application uses computer vision technology to track limb joints and draw motion trajectories, which can intuitively display the patient's movement method and range, which helps to analyze the patient's movement smoothness, coordination and direction, and discover abnormal movement patterns. Analyzing the movement frequency can quantify the patient's behavioral activity. Patients with depression often have a low movement frequency. By analyzing the movement frequency, objective data support can be provided for the judgment of depression. The combination of behavioral data with facial feature data, brain wave data and multi-dimensional data can more comprehensively evaluate the patient's condition and improve the accuracy of depression diagnosis.

[0027] The frequency analysis of the patient's actions to be identified sets the time period as T, and the number of occurrences of a specific action as c, which is expressed as: within the T time period, the number of occurrences c is counted, and based on the total number of actions and the number of occurrences of the specific action, the frequency k of the specific action is calculated, and the behavioral feature value F is evaluated. The evaluation formula is: F=jk+h Where F is the behavioral characteristic value obtained by evaluation, and j and h are constants.

[0028] In step S4, the judgment step sets the output result of the preliminary recognition as O, the facial feature data as P, the behavior data as F, and the analysis function R (P, F), which represents the process of combining the facial feature data and the behavior data for analysis. The judgment process is in the form of: Among them, Y represents the judgment of depression, N represents the judgment of not depression, indicating that there is an error, then continue to monitor, identify and judge, S is the output result; if the preliminary identification result is that there is a tendency of depression, after analyzing the facial feature data and behavioral data, it is also concluded that there is a tendency of depression, then the final result is depression; if the analysis result is inconsistent with the preliminary identification result, re-collect data, perform preliminary identification and combine facial features and behavioral data for analysis steps until an accurate result is obtained.

[0029] After identifying that the patient has depression in step S5, the physician conducts psychological counseling for the patient, creating a safe, comfortable, and private environment so that the patient can relax, open up, and communicate. The physician analyzes and discusses the causes of depression and takes measures to address the causes. During the psychological counseling process, the physician uses cognitive behavioral therapy to help the patient identify and correct negative thinking patterns and behavioral habits, cultivate a positive way of thinking, and encourage the patient to take positive actions.

[0030] This application creates a safe, comfortable and private environment that allows patients to quickly let down their psychological defenses and be more willing to reveal their true feelings and thoughts. Physicians can have in-depth communication based on each patient's unique situation, understand their specific life experiences, personality traits and psychological needs, and thus provide more targeted psychological guidance. In the process of psychological guidance, close communication and interaction between physicians and patients can establish a doctor-patient relationship of trust and support, which is crucial to the patient's recovery. Patients will be more willing to cooperate with treatment and follow the physician's advice. A good doctor-patient relationship can also make patients feel cared for and warm, and reduce loneliness and helplessness.

[0031] The above shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments, and the above embodiments and descriptions only describe the principles of the present invention. The present invention may be subject to various changes and improvements without departing from the spirit and scope of the present invention, and these changes and improvements fall within the scope of the present invention claimed.

Claims

1. A depression recognition method based on multidimensional feature integration, characterized in that: The identification steps are: S1. Play different kinds of music for the patient to be identified, including happy, sad and passionate music. Obtain the EEG data of the patient to be identified based on the EEG device, build an EEG judgment model based on big data, train the EEG judgment model through the EEG data set of historical patients, input the EEG data into the EEG judgment model, and output a preliminary recognition result; S2, collect facial feature data of the patient to be identified, and upload the acquired data to the identification module for storage; S3, obtaining the behavior data of the patient to be identified, and uploading the behavior data to the identification module for storage; S4. Analyze the preliminary recognition results in combination with the facial feature data and the behavioral data, and judge the preliminary recognition results to determine their accuracy. If yes, output depression, otherwise repeat the above steps to continue the judgment; S5. After identifying depression, the doctor will provide psychological counseling to the patient based on the patient's condition.

2. The depression identification method based on multidimensional feature integration according to claim 1 is characterized in that: In step S1, different types of music are played to the patient to be identified in turn, including happy, sad and passionate music. Each type of music is paused after playing for a period of time to give the patient to be identified a buffer time. During the music playing process, the EEG data of the patient to be identified is obtained in real time based on the EEG device, and the changes of the patient's EEG signals under different music stimulations are recorded, including frequency, amplitude and phase characteristics; The EEG judgment model trains the data set based on deep learning and analyzes the EEG data under different music stimulations. After the training is completed, the EEG data is input into the model and the results are output.

3. The depression identification method based on multidimensional feature integration according to claim 2 is characterized in that: The EEG judgment model training process assumes that the model is in the form of: M = (x, y), where x is the data input into the model, and y is the parameter value of the model's minimized loss function. The historical patient EEG data set is: where a i is the EEG data combination of the i-th historical patient under different music stimulations, b i is the corresponding label to determine whether depression exists; The formula for minimizing the loss function is defined as: Where y is the parameter value of the model's minimized loss function, arg represents the value of the variable when the formula reaches the minimum value, n is the number of EEG data combinations, M is the EEG judgment model, and the formula is used to obtain the parameter value of y.

4. The depression identification method based on multidimensional feature integration according to claim 2 is characterized in that: The model is optimized after training. The hyperparameter optimization method is used to optimize the hyperparameters of the model, and the best hyperparameter settings are obtained based on cross-validation.

5. The depression identification method based on multidimensional feature integration according to claim 1, characterized in that: In step S2, based on the image acquisition device, the patient to be identified is photographed from multiple angles and at multiple time periods, the photographed images are uploaded to the recognition module, and based on the data transmission channel, the facial feature data of the patient to be identified is transmitted in an encrypted form, and the data is classified and stored; The recognition module uses image analysis technology to pre-process the collected images, extract key facial feature points, including the position, shape and proportion of the eyes, nose and mouth, analyze the facial feature data, and store the analysis results.

6. The depression identification method based on multidimensional feature integration according to claim 1, characterized in that: In step S3, the behavioral data of the patient to be identified is obtained through monitoring equipment, and the patient's body movements in different scenarios are recorded. After the behavioral data is collected, it is uploaded based on the transmission channel, and the joints of the limbs are tracked based on computer vision technology, the motion trajectory is drawn, and the movement frequency of the movement is analyzed to obtain the analysis data results.

7. The depression identification method based on multidimensional feature integration according to claim 6 is characterized in that: The frequency analysis of the patient's actions to be identified sets the time period as T, and the number of occurrences of a specific action as c, which is expressed as: within the T time period, the number of occurrences c is counted, and based on the total number of actions and the number of occurrences of the specific action, the frequency k of the specific action is calculated, and the behavioral feature value F is evaluated. The evaluation formula is: F=jk+h Where F is the behavioral characteristic value obtained by evaluation, and j and h are constants.

8. The depression identification method based on multidimensional feature integration according to claim 1, characterized in that: In step S4, the judgment step sets the output result of the preliminary recognition as O, the facial feature data as P, the behavior data as F, and the analysis function R (P, F), which represents the process of combining the facial feature data and the behavior data for analysis. The judgment process is in the form of: Among them, Y represents the judgment of depression, N represents the judgment of not depression, indicating that there is an error, then continue to monitor, identify and judge, S is the output result; if the preliminary identification result is that there is a tendency of depression, after analyzing the facial feature data and behavioral data, it is also concluded that there is a tendency of depression, then the final result is depression; if the analysis result is inconsistent with the preliminary identification result, re-collect data, perform preliminary identification and combine facial features and behavioral data for analysis steps until an accurate result is obtained.

9. The depression identification method based on multidimensional feature integration according to claim 1, characterized in that: After identifying that the patient suffers from depression in step S5, the physician will provide psychological counseling to the patient and create a safe, comfortable and private environment so that the patient can relax, open up, communicate, analyze and discuss the causes of depression, and take measures to address the causes.

10. The depression identification method based on multidimensional feature integration according to claim 9, characterized in that: During the psychological counseling process, physicians use cognitive behavioral therapy to help patients identify and correct negative thinking patterns and behavioral habits, cultivate positive thinking, and encourage patients to take positive actions.

Images