An auxiliary assessment system for anhedonia in patients with depression using multimodal physiological parameters

Abstract

The present invention relates to the technical field of multi-dimensional auxiliary evaluation, and specifically to an auxiliary evaluation system for anhedonia in depression patients using multi-modal physiological parameters. The system includes a data acquisition module to obtain quantitative anhedonia evaluation values ​​in different dimensions; a basic weight analysis module to analyze the data distribution of multiple patients in the dimensions to obtain basic weights; a low representation analysis module, which is used to obtain low representation weights through continuous changes in multiple anhedonia evaluation values ​​and the correlation between low representation dimensions and high representation dimensions; a high representation analysis module, which is used to cluster patients through low representation weights, analyze the deviation of anhedonia evaluation values ​​of high representation dimensions between clustered patients to obtain high representation weights; and an evaluation module combines basic weights, low representation weights and high representation weights for evaluation. The present invention distinguishes different dimensions through data collection methods to perform data credibility analysis, comprehensively obtains more appropriate weights for each dimension, and improves the accuracy of multi-dimensional evaluation results.

Description

Technical Field

[0001] The present invention relates to the technical field of multi-dimensional auxiliary assessment, and in particular to an auxiliary assessment system for anhedonia in depressed patients using multi-modal physiological parameters. Background Art

[0002] Anhedonia is one of the core symptoms of depression and a difficult point in the treatment of depression. It is one of the important predictors of a long course of depression, a significant decline in quality of life, and severe impairment of social function. By quantifying or evaluating the degree of anhedonia, professionals can more accurately diagnose the severity of the disease, develop personalized treatment plans, and evaluate changes in treatment effectiveness. In addition, the assessment of the degree of anhedonia can also help identify potential mental health problems and prevent the occurrence or deterioration of mood disorders, thereby improving the individual's overall mental health level and quality of life.

[0003] However, when using multimodal physiological parameters to evaluate the degree of anhedonia in patients with depression, multiple physiological parameters are often combined for evaluation using fixed weights. However, for different patients, the contribution of different parameters to the final diagnosis may be different. Using constant weights without considering the actual differences between individuals will lead to fluctuations in the accuracy of the evaluation results and large errors in patient evaluation. Summary of the invention

[0004] In order to solve the technical problems of fluctuations in the accuracy of evaluation results and large errors in patient evaluation in the prior art, the present invention aims to provide an auxiliary evaluation system for anhedonia in depressed patients using multimodal physiological parameters. The technical solutions adopted are as follows:

[0005] The present invention provides an auxiliary evaluation system for anhedonia in patients with depression using multimodal physiological parameters, the system comprising:

[0006] A data acquisition module is used to obtain the anhedonia evaluation value of each patient in different dimensions; the dimensions are divided into low-representation dimensions and high-representation dimensions according to different data acquisition methods of each dimension; the data acquisition methods include subjective acquisition and objective acquisition;

[0007] A basic weight analysis module is used to obtain the basic weight of each dimension according to the distribution disorder of the anhedonia evaluation values ​​in each dimension among different patients and the degree of disorder of the anhedonia evaluation values ​​of each patient in different dimensions;

[0008] The low-representation analysis module is used to continuously collect data for each dimension of each patient to obtain a preset number of historical anhedonia evaluation values ​​for each dimension; according to the changes in the historical anhedonia evaluation values ​​of each patient in each low-representation dimension, as well as the changes in the historical anhedonia evaluation values ​​and the consistency of data distribution between each low-representation dimension and the high-representation dimension, the low-representation weight of each patient in each dimension is obtained;

[0009] A high representation analysis module is used to cluster patients based on their low representation weights in low representation dimensions to obtain patient clusters; according to the deviation of each high representation dimension of each patient in the patient cluster and the corresponding high representation dimension of other patients in the anhedonia evaluation value, determine the high representation weight of each patient in each dimension;

[0010] The evaluation module is used to obtain the evaluation weight of each patient in each dimension according to the low representation weight, high representation weight and basic weight of each patient in each dimension; based on the evaluation weight combined with the anhedonia evaluation value of each patient in each dimension, the evaluation result of each patient is obtained.

[0011] Furthermore, the method for obtaining the basic weight includes:

[0012] The entropy values ​​of each patient's anhedonia evaluation values ​​in all dimensions are negatively correlated and mapped as the influence weight of each patient;

[0013] For any dimension, the frequency of occurrence of the anhedonia evaluation value corresponding to each patient of the dimension in the anhedonia evaluation values ​​corresponding to all patients of the dimension is taken as the distribution frequency of the dimension in each patient;

[0014] The entropy value is calculated by weighting the distribution frequency corresponding to the dimension with the influence weight of each patient to obtain the basic weight of the dimension.

[0015] Furthermore, the method for obtaining the low representation weight includes:

[0016] For any patient, each low-representation dimension of the patient is taken as the analysis dimension in turn;

[0017] According to the numerical consistency of all historical anhedonia evaluation values ​​of the patient under the analysis dimension, and the stability of the temporal changes of all historical anhedonia evaluation values, the first factor of the patient under the analysis dimension is obtained;

[0018] According to the consistency of the change of the patient's historical anhedonia evaluation value between the analysis dimension and each high representation dimension and the closeness of the numerical distribution, the second factor of the patient under the analysis dimension is obtained;

[0019] The first factor and the second factor of the patient in each low representation dimension are combined to determine the low representation weight of the patient in each dimension.

[0020] Furthermore, the method for obtaining the first factor includes:

[0021] Negative correlation mapping is performed on the standard deviation of all historical anhedonia evaluation values ​​of the patient under the analysis dimension to obtain the numerical stability of the patient under the analysis dimension;

[0022] Arrange all the historical anhedonia evaluation values ​​of the patient under the analysis dimension in chronological order to obtain an evaluation value sequence; take the difference between each two adjacent historical anhedonia evaluation values ​​in the evaluation value sequence as the change difference; calculate the difference between the two change differences, calculate the difference sum and perform negative correlation mapping to obtain the change stability of the patient under the analysis dimension;

[0023] The first factor of the patient under the analysis dimension is obtained by combining the numerical stability and change stability of the patient under the analysis dimension.

[0024] Furthermore, the method for obtaining the second factor includes:

[0025] All historical anhedonia evaluation values ​​of the patient in each dimension are sorted in chronological order to obtain the evaluation value sequence of the patient in each dimension; the mean of all historical anhedonia evaluation values ​​of the patient in each dimension is taken as the mean evaluation value of the patient in each dimension;

[0026] After calculating the Pearson correlation coefficient between the evaluation value sequence of the patient in the analysis dimension and each high representation dimension, the sum of the Pearson correlation coefficients is calculated to obtain the correlation of the changes in the analysis dimension of the patient;

[0027] After calculating the difference between the patient's evaluation mean in the analysis dimension and each high-representation dimension, the difference and value are calculated and negative correlation mapping is performed to obtain the patient's numerical correlation in the analysis dimension;

[0028] The change correlation and value correlation of the patient in the analysis dimension are combined to obtain the second factor of the patient in the analysis dimension.

[0029] Furthermore, combining the first factor and the second factor of the patient in each low representation dimension to determine the low representation weight of the patient in each dimension includes:

[0030] Normalizing the product of the first factor and the second factor of the patient in each low representation dimension to obtain the low representation weight of the patient in each low representation dimension;

[0031] The low representation weight of the patient in each high representation dimension is set to a preset weight value.

[0032] Furthermore, the method for obtaining the patient clusters includes:

[0033] The low representation weights of each patient in all low representation dimensions are combined into a representation vector for each patient; the similarity of the representation vectors is used as a metric to cluster the patients through a clustering algorithm to obtain patient clusters.

[0034] Furthermore, the method for obtaining the high representation weight includes:

[0035] For any patient, each high representation dimension of the patient is taken as the target dimension in turn;

[0036] Calculate the difference between the anhedonia evaluation value of the patient and that of every other patient in the patient cluster in the target dimension as the deviation value of the patient and every other patient in the target dimension; normalize the sum of the deviation values ​​of the patient and all other patients in the target dimension as the high representation dimension of the patient in the target dimension;

[0037] The high representation weight of the patient in each low representation dimension is set to a preset weight value.

[0038] Furthermore, the method for obtaining the evaluation weight includes:

[0039] The product of the basic weight, low representation weight and high representation weight of each patient in each dimension was normalized to obtain the evaluation weight of each patient in each dimension.

[0040] Furthermore, the evaluation result of each patient is obtained based on the evaluation weight and the anhedonia evaluation value of each patient in each dimension, including:

[0041] The product of each patient's anhedonia evaluation value in each dimension and the corresponding evaluation weight is taken as the evaluation value of each patient in each dimension;

[0042] The sum of the evaluation values ​​of each patient in all dimensions is taken as the evaluation result of each patient.

[0043] The present invention has the following beneficial effects:

[0044] The present invention quantifies the anhedonia evaluation value under different dimensions, combines multiple patients to analyze the data distribution under different dimensions to obtain the basic weight, and comprehensively determines the initial weight by combining the data distribution of each dimension and the stability of the patient's own data. Considering the different trust situations brought by the data collection method, the trust situation of the anhedonia evaluation value is analyzed separately by the low representation dimension and the high representation dimension distinguished by the data collection method. Combined with the patient's multiple collection results, by quantifying the continuous change degree of the anhedonia evaluation value, and the continuous correlation between the low representation dimension and the high representation dimension, the trust level of each dimension under the low representation dimension analysis is determined to obtain the low representation weight, and the patients are clustered by the low representation weight to reflect the trustworthy categories that may be similar for multiple patients under the low representation dimension collection method. At this time, the deviation of the anhedonia evaluation value under the high representation dimension between the clustered patients is analyzed, and the trust level of the data is analyzed by the high representation dimension to obtain the high representation weight. Finally, the evaluation weight of each dimension is obtained by combining the basic weight, the low representation weight and the high representation weight to reflect the contribution ratio, and the evaluation result is calculated by the evaluation weight. The present invention distinguishes different dimensions by the data collection method to analyze the data trust level, comprehensively obtains a more appropriate weight for each dimension, and improves the accuracy of the multi-dimensional evaluation results. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0046] Figure 1 A structural diagram of a system for assisting in assessing anhedonia in depressed patients using multimodal physiological parameters provided by one embodiment of the present invention;

[0047] Figure 2 A flow chart of a method for obtaining a low representation weight provided by an embodiment of the present invention;

[0048] Figure 3 A schematic diagram of changes in historical anhedonia evaluation values ​​in a low representation dimension provided by an embodiment of the present invention;

[0049] Figure 4 A schematic diagram of changes in historical anhedonia evaluation values ​​in another low representation dimension provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0050] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the following is a detailed description of the specific implementation method, structure, features and effects of a depression patient anhedonia auxiliary assessment system using multimodal physiological parameters proposed by the present invention in combination with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.

[0051] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0052] The following is a detailed description of a specific scheme of the anhedonia auxiliary assessment system for depression patients using multimodal physiological parameters provided by the present invention in conjunction with the accompanying drawings.

[0053] See also Figure 1 , which shows a structural diagram of a system for auxiliary assessment of anhedonia in depressive patients using multimodal physiological parameters provided by an embodiment of the present invention, the system includes a data acquisition module 101, a basic weight analysis module 102, a low representation analysis module 103, a high representation analysis module 104 and an assessment module 105.

[0054] The data acquisition module 101 is used to obtain the anhedonia evaluation value of each patient in different dimensions; the dimensions are divided into low-representation dimensions and high-representation dimensions according to different data acquisition methods of each dimension; the data acquisition methods include subjective acquisition and objective acquisition.

[0055] Anhedonia, as one of the core symptoms of depression, is closely related to a variety of physiological parameters. For example, when depressed patients experience pleasant experiences, the reward circuits of the brain, such as the nucleus accumbens or prefrontal cortex, usually show lower activity levels. By analyzing EEG signals, especially changes in specific frequency bands, such as theta waves or alpha waves, it is possible to reflect abnormalities in the brain's emotional processing, motivation, and reward mechanisms. Usually, low-frequency EEG activity is highly correlated with the severity of anhedonia symptoms. In addition, since ECG signals can reflect the activity of the sympathetic and parasympathetic nerves, and patients with depression usually show disorders of the autonomic nervous system, anhedonia can cause patients to show lower heart rate variability (HRV) when they are depressed.

[0056] It can be seen that anhedonia is highly correlated with the human neurophysiological mechanism. Therefore, the monitoring and analysis of multimodal physiological parameter signals can effectively reveal the neurophysiological mechanism behind anhedonia. However, due to individual differences, when different patients experience anhedonia, there are differences in the degree of representation of different neurophysiological signals. Therefore, different degrees of weight can be given to the patient's physiological signals in the auxiliary assessment of anhedonia to improve the assessment effect.

[0057] In the embodiment of the present invention, when a patient with depression sees a doctor, professional equipment is required to monitor the relevant physiological parameters, and the patient is asked to fill in relevant scales and other information according to the doctor's requirements. The data of physiological parameters collected by the equipment is objectively collected data, and the data collected by the patient filling in the data by himself is subjectively collected data. When collecting data, each type is used as a dimension, such as EEG signals, ECG signals, pulse wave signals, skin electrical signals, gastric electrical signals, myoelectric signals or sleep frequency, etc.

[0058] Since subjectively collected data are more easily affected by the patient's thoughts, while objectively collected data are not affected by thoughts, objectively collected data are more accurate than subjectively collected data in terms of representation accuracy. Therefore, the subjectively collected dimension is regarded as the low representation dimension, and the objectively collected data is regarded as the high representation dimension.

[0059] Physiological parameters can also reveal the anhedonia of patients with depression. For example, electrodes are placed on the patient's scalp to record EEG signals, and the potential changes are converted into graphics, which are transmitted to the data recording terminal to obtain the EEG. EEGs usually show abnormalities in different frequency bands. For example, Alpha waves are related to relaxation and emotional regulation. Depressed patients usually show reduced low-frequency alpha wave activity. Theta waves are related to emotional regulation, attention and self-reflection. Depressed patients often have excessive theta wave activity. By analyzing the changes in various waves in the EEG, the patient's anhedonia level can be reflected from the side.

[0060] The anhedonia evaluation value of each dimension is obtained through the data collected in each dimension. In the embodiment of the present invention, taking the ECG signal as an example, the specific process of the corresponding anhedonia level evaluation value is as follows: the ECG signal data is converted into a format suitable for neural network input through standardization, the mean square error loss function is selected to design the neural network model, the input layer receives the ECG signal data, the output layer outputs the anhedonia level evaluation value, and the model is trained using the training data set to obtain the training completion model to realize the evaluation value of the patient's anhedonia level quantified according to the ECG signal. The data of the corresponding dimension of the collected ECG signal is input into the corresponding training completion model to obtain the anhedonia evaluation value of the corresponding dimension. The same is true for other dimensions, which will not be repeated here.

[0061] At this point, the initial data acquisition is completed.

[0062] The basic weight analysis module 102 is used to obtain the basic weight of each dimension according to the distribution disorder of the anhedonia evaluation value in each dimension among different patients and the degree of disorder of the anhedonia evaluation value of each patient in different dimensions.

[0063] First, the distribution disorder of anhedonia evaluation values ​​is analyzed for all dimensions combined with multiple patients, and the weight of each dimension is determined. When the data distribution disorder of a certain dimension is greater, it indicates that for different levels of anhedonia, the difference of this dimension is greater, and it can better distinguish the anhedonia levels of different patients. Therefore, the higher weight of this dimension should be based on when determining the weight. However, at the same time, if the difference between the anhedonia evaluation values ​​obtained by a patient in different dimensions is greater, it can be said that the data accuracy of this patient is lower, and the relevant data of this patient should be given a smaller weight when calculating.

[0064] Preferably, in some possible implementations of the embodiments of the present invention, the method for obtaining the basic weight includes:

[0065] First, the entropy value of each patient's anhedonia evaluation value in all dimensions is negatively correlated and mapped as the influence weight of each patient, reflecting the credibility of each patient's corresponding anhedonia evaluation value. The smaller the entropy value of the patient's anhedonia evaluation value in multiple dimensions, the more consistent the multi-dimensional quantitative evaluation value is, reflecting the higher accuracy of the patient's data in quantification. In the subsequent calculation of the patient's related data, a higher weight is given, so the influence weight is greater. It should be noted that negative correlation mapping and entropy value calculation are technical means well known to those skilled in the art. Negative correlation mapping can be in the form of inverse proportional value or negative exponential power, which will not be elaborated here.

[0066] Furthermore, for any dimension, the frequency of occurrence of the anhedonia evaluation value corresponding to each patient of this dimension in the anhedonia evaluation values ​​corresponding to all patients of this dimension is used as the distribution frequency of this dimension in each patient. The subsequent entropy value calculation is performed based on the frequency of occurrence of the numerical value corresponding to this dimension in each patient, so as to facilitate the analysis of different contribution influences on the credibility of different patients.

[0067] Finally, the entropy value is calculated by weighting the distribution frequency corresponding to the dimension with the influence weight of each patient, and the basic weight of the dimension is obtained. The data entropy is calculated according to the frequency of occurrence, and the chaotic distribution of the anhedonia evaluation value corresponding to each dimension is reflected in the form of entropy. The larger the entropy value, the more significant the difference is, and the higher the basic weight is. As an example, the expression of the basic weight is:

[0068] ;

[0069] In the formula, Expressed as The basic weight of the dimension, Expressed as The impact weight of each patient, Expressed as the total number of patients, Expressed as The dimension in The distribution frequency of patients, It is expressed as a logarithmic function with base 2. It should be noted that the use of the entropy function formula is a technical means well known to those skilled in the art, so the meaning of the specific formula will not be repeated.

[0070] At this point, the data entropy of the anhedonia evaluation value of each dimension is obtained, and the basic weight is obtained by weighting the chaotic distribution of the numerical values ​​of each patient in the multi-dimensional evaluation value.

[0071] The low-representation analysis module 103 is used to continuously collect data for each dimension of each patient to obtain a preset number of historical anhedonia evaluation values ​​for each dimension; according to the changes in the historical anhedonia evaluation values ​​of each patient in each low-representation dimension, as well as the changes in the historical anhedonia evaluation values ​​and the consistency of data distribution between each low-representation dimension and the high-representation dimension, the low-representation weight of each patient in each dimension is obtained.

[0072] The patient's level of anhedonia should be a process with a relatively stable changing trend. For low-representation and low-dimensional data, the more unstable a certain data of the patient is, the more it proves that these low-representation dimensions of the patient are not accurately responding to the level of anhedonia, and the change may be caused by other reasons. Therefore, the anhedonia evaluation value obtained based on such low-representation dimensions should have a smaller weight when data is fused, so that the final result is as unaffected as possible by other irrelevant influences.

[0073] The greater the fluctuation in the evaluation value of the patient's anhedonia level obtained from the low-representational dimension data under multiple different collection situations, it proves that the test results of this dimension may be affected by other factors. At this time, the credibility of this low-representational dimension is low. Therefore, the patient's low-representational dimension data is further analyzed in combination with multiple collection situations.

[0074] In an embodiment of the present invention, a preset number of data are collected from the patient in different time periods to obtain data for each dimension and obtain an anhedonia evaluation value as a historical anhedonia evaluation value. The patient has a preset number of historical anhedonia evaluation values ​​in each dimension. The preset number can be set to 5, and the value is adjusted by the implementer according to the specific implementation scenario and is not limited here.

[0075] When the patient's anhedonia level of a low representation dimension is less stable, it means that the low representation dimension is less credible. Even if the deviation between the anhedonia level of a low representation dimension with a certain stability and the corresponding patient's high representation dimension is high, it means that the credibility of the low representation dimension is also low. Therefore, the credibility of the low representation dimension is evaluated from two aspects to obtain the low representation weight of each dimension.

[0076] Preferably, in some possible implementations of the embodiments of the present invention, the method for obtaining the low representation weight is described in Figure 2 , which shows a flow chart of a method for obtaining a low representation weight provided by an embodiment of the present invention, the method comprising the following steps:

[0077] S301: For any patient, each low representation dimension of the patient is taken as an analysis dimension in turn; according to the numerical consistency of all historical anhedonia evaluation values ​​of the patient under the analysis dimension, and the stability of the temporal changes of all historical anhedonia evaluation values, the first factor of the patient under the analysis dimension is obtained.

[0078] For the low representation dimension with strong subjectivity, when the overall fluctuation of the patient's anhedonia level data obtained by the low representation dimension is large and the changes are more chaotic, it means that the subjective test results are not credible, and the patient's anhedonia level evaluation through this low representation dimension is inaccurate, and a smaller weight needs to be assigned in the future. Therefore, the numerical size and change of the anhedonia evaluation value collected from the time series are stable, and the first factor is obtained comprehensively, reflecting the possibility of assigning weights from the stable change.

[0079] See also Figure 3 , which shows a schematic diagram of the change of historical anhedonia evaluation values ​​in a low representation dimension provided by an embodiment of the present invention, Figure 4 , which shows another schematic diagram of the change of historical anhedonia evaluation values ​​with low representation dimensions provided by an embodiment of the present invention, wherein the horizontal axis represents the temporal distribution order of the historical anhedonia evaluation values, and the vertical axis represents the historical anhedonia evaluation values. The change of the anhedonia evaluation values ​​in the figure is relatively unstable.

[0080] Therefore, for any patient, each low representation dimension is used as an analysis dimension for analysis. In an embodiment of the present invention, the method for obtaining the first factor includes:

[0081] First, the standard deviation of all historical anhedonia evaluation values ​​of the patient in the analysis dimension is negatively correlated to obtain the numerical stability of the patient in the analysis dimension. The stability is reflected numerically in the form of standard deviation. The smaller the standard deviation, the higher the stability of the numerical value and the greater the numerical stability.

[0082] Further, all the historical anhedonia evaluation values ​​of the patient under the analysis dimension are arranged in chronological order to obtain the evaluation value sequence, and analyzed from the perspective of time series change. The difference between each two adjacent historical anhedonia evaluation values ​​in the evaluation value sequence is taken as the change difference, reflecting the degree of change of the continuous value. After calculating the difference between the two change differences, the difference sum is calculated and negative correlation mapping is performed to obtain the change stability of the patient under the analysis dimension. When the overall difference between the change degrees is larger, it means that the change state is more unstable, and the change stability is lower.

[0083] Finally, the numerical stability and change stability of the patient in the analysis dimension are combined to obtain the first factor of the patient in the analysis dimension. In an embodiment of the present invention, the product of the numerical stability and change stability of the patient in the analysis dimension is used as the first factor of the patient in the analysis dimension. Through the product combination, the greater the numerical stability and change stability, the more accurate the characterization ability of the analysis dimension, the greater the weight that can be assigned later, and therefore the greater the first factor. As an example, the expression of the first factor is:

[0084] ;

[0085] In the above formula, Expressed as The patient in The first factor of low representational dimension, Expressed as The patient in Numerical stability of low representation dimensions, Expressed as The difference in changes, Expressed as the total number of change differences, Expressed as The patient in The stability of changes in a low representation dimension.

[0086] S302: Obtain the second factor of the patient in the analysis dimension according to the consistency of the change of the patient's historical anhedonia evaluation value in the analysis dimension and the patient's historical anhedonia evaluation value in each high representation dimension and the closeness of the numerical distribution.

[0087] The more similar the change pattern of the patient's historical anhedonia evaluation value of the low representation dimension is to the change pattern of the patient's high representation dimension, the more accurate the anhedonia evaluation level is under the corresponding low representation dimension. In the subsequent evaluation of the patient's anhedonia level, the low representation dimension should be given a higher weight. The second factor is obtained from the similarity between the time series change and the overall value distribution and the high representation dimension.

[0088] In an embodiment of the present invention, a method for obtaining the second factor includes:

[0089] First, all the historical anhedonia evaluation values ​​of the patient in each dimension are sorted in chronological order to obtain the evaluation value sequence of the patient in each dimension and the time series change sequence. The mean of all the historical anhedonia evaluation values ​​of the patient in each dimension is taken as the mean evaluation value of the patient in each dimension, reflecting the overall distribution of the values.

[0090] Then, after calculating the Pearson correlation coefficient between the evaluation value sequence of the patient in the analysis dimension and each high representation dimension, the Pearson correlation coefficient and value are calculated to obtain the change correlation of the patient in the analysis dimension. The larger the Pearson correlation coefficient, the higher the consistency of the change, the higher the similarity between the analysis dimension and the high representation dimension in the change of the quantitative anhedonia level, and the greater the change correlation. It should be noted that the Pearson correlation coefficient is a technical means well known to those skilled in the art and will not be described in detail here.

[0091] Then, after calculating the difference between the patient's evaluation mean in the analysis dimension and each high-representation dimension, the difference and value are calculated and negative correlation mapping is performed to obtain the patient's numerical correlation in the analysis dimension. Through the degree of closeness of the numerical distribution, the smaller the difference between the cumulative evaluation means, the closer the numerical distribution of the analysis dimension and the high-representation dimension in the quantitative level of anhedonia is, so the greater the numerical correlation.

[0092] Finally, the change correlation and numerical correlation of the patient in the analysis dimension are combined to obtain the second factor of the patient in the analysis dimension. In the embodiment of the present invention, the product of the change correlation and numerical correlation of the patient in the analysis dimension is used as the second factor of the patient in the analysis dimension. The greater the change correlation and numerical correlation, the higher the consistency between the analysis dimension and the high representation dimension, the greater the credibility of the analysis dimension, and the higher the weight that can be assigned subsequently, so the second factor is larger.

[0093] S303: Determine the low representation weight of the patient in each dimension by combining the first factor and the second factor of the patient in each low representation dimension.

[0094] Finally, the two aspects are combined to obtain the low representation weight of the dimension. Since the main focus is on the low representation dimension, the low representation dimension and the high representation dimension are weighted separately. In an embodiment of the present invention, the product of the first factor and the second factor of the patient in each low representation dimension is normalized to obtain the low representation weight of the patient in each low representation dimension. It should be noted that normalization is a technical means well known to technicians in this field. The normalization option can be linear normalization or standard normalization, etc. The specific normalization method is not limited here.

[0095] The low representation weight of the patient in each high representation dimension is further set to a preset weight value. In the embodiment of the present invention, the preset weight value is 1, that is, the low representation dimension analysis does not affect the weight assignment of the high representation dimension.

[0096] At this point, the analysis of the low representation dimensions is completed, and the low representation weight is set for each dimension.

[0097] The high representation analysis module 104 is used to cluster patients based on their low representation weights in low representation dimensions to obtain patient clusters; and determine the high representation weight of each patient in each dimension according to the deviation of the anhedonia evaluation value between each high representation dimension of each patient in the patient cluster and the corresponding high representation dimension of other patients.

[0098] Further corrections can be made to the weighting of high-representation dimensions. For patients with similar low-representation dimensions, due to insufficient representation ability, a more accurate high-representation dimension analysis will be more important. At this time, if a patient's high-representation dimension is significantly different from the overall evaluation level of other patients, it means that the high-representation data of this dimension is more targeted in the evaluation of this patient. In order to ensure that this dimension receives more attention during patient evaluation, a higher weight should be assigned to this high-representation dimension.

[0099] First, patients with relatively similar low representation dimensions are clustered to reduce the impact of different evaluation levels of the low representation dimensions. In an embodiment of the present invention, the low representation weights of each patient in all low representation dimensions are combined into a representation vector of each patient. The similarity of the representation vectors is used as a metric to cluster the patients through a clustering algorithm to obtain patient clusters. The clustering algorithm can select the k-means clustering algorithm. It should be noted that the clustering algorithm and vector similarity calculation are both well-known technical means well known to those skilled in the art and will not be elaborated here.

[0100] The evaluation level of each high-representation dimension of each patient is analyzed in each cluster. Preferably, in some possible implementations of the embodiments of the present invention, the method for obtaining the high-representation weight includes:

[0101] First, for any patient, each high representation dimension of the patient is taken as the target dimension in turn, and each high representation dimension is analyzed. The difference between the anhedonia evaluation value of the patient and each other patient in the patient cluster in the target dimension is calculated as the deviation value of the patient and each other patient in the target dimension, reflecting the degree of deviation in the anhedonia evaluation of patients in the same patient cluster under the target dimension.

[0102] The sum of the deviation values ​​between the patient and all other patients on the target dimension is further normalized as the high representation dimension of the patient on the target dimension. The overall deviation between the patient and all patients reflects the patient's evaluation representation ability on the target dimension. The greater the overall deviation, the more attention is paid to the evaluation results, and therefore the larger the normalized high representation dimension.

[0103] The high representation weight of the patient in each low representation dimension is further set to a preset weight value, and the weight assignment of the low representation dimension is not affected during the analysis of the high representation dimension.

[0104] At this point, the analysis of the low representation dimensions is completed, and the low representation weight is set for each dimension.

[0105] The evaluation module 105 is used to obtain the evaluation weight of each patient in each dimension according to the low representation weight, high representation weight and basic weight of each patient in each dimension; based on the evaluation weight combined with the anhedonia evaluation value of each patient in each dimension, obtain the evaluation result of each patient.

[0106] Finally, the basic weight, low representation weight and high representation weight of each dimension of each patient are comprehensively considered, and the weighting is adjusted from the overall analysis and the differentiated analysis, respectively, to adjust the contribution of the low representation dimension with greater influence, and to adjust the contribution of the high representation dimension with high evaluation influence. In an embodiment of the present invention, the product of the basic weight, low representation weight and high representation weight of each patient in each dimension is normalized to obtain the evaluation weight of each patient in each dimension. The higher the weight, the more attention should be paid to the corresponding dimension in the evaluation of anhedonia.

[0107] Therefore, based on the evaluation weight and the anhedonia evaluation value, the evaluation result is obtained comprehensively. In the embodiment of the present invention, the product of the anhedonia evaluation value of each patient in each dimension and the corresponding evaluation weight is used as the evaluation value of each patient in each dimension, and weighted by the evaluation weight. And the sum of the evaluation values ​​of each patient in all dimensions is used as the evaluation result of each patient, which comprehensively reflects the anhedonia evaluation situation.

[0108] In the embodiment of the present invention, the evaluation result of anhedonia can be fed back to the doctor, and the data corresponding to the evaluation result can be displayed to assist the doctor's diagnosis and improve the efficiency of diagnosis.

[0109] The present invention quantifies the anhedonia evaluation value under different dimensions, combines multiple patients to analyze the data distribution under each dimension to obtain the basic weight, and comprehensively determines the initial weight by combining the data distribution of each dimension and the stability of the patient's own data. Considering the different trust situations brought by the data collection method, the trust situation of the anhedonia evaluation value is analyzed separately by the low representation dimension and the high representation dimension distinguished by the data collection method. Combined with the patient's multiple collection results, by quantifying the continuous change degree of the anhedonia evaluation value, and the continuous correlation change and distribution between the low representation dimension and the high representation dimension, the trust level of each dimension under the low representation dimension analysis is determined to obtain the low representation weight, and the patients are clustered by the low representation weight to reflect the trustworthy categories that may be similar for multiple patients under the low representation dimension collection method. At this time, the deviation of the anhedonia evaluation value under the high representation dimension between the clustered patients is analyzed, and the trust level of the data is analyzed by the high representation dimension to obtain the high representation weight. Finally, the evaluation weight of each dimension is obtained by combining the basic weight, the low representation weight and the high representation weight to reflect the contribution ratio, and the evaluation result is calculated by the evaluation weight. The present invention distinguishes different dimensions by the data collection method to analyze the data trust level, comprehensively obtains a more appropriate weight for each dimension, and improves the accuracy of the multi-dimensional evaluation results.

[0110] It should be noted that the sequence of the above embodiments of the present invention is only for description and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0111] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.

Claims

1. A system for assisting the assessment of anhedonia in patients with depression using multimodal physiological parameters, characterized in that: The system comprises: A data acquisition module is used to obtain the anhedonia evaluation value of each patient in different dimensions; the dimensions are divided into low-representation dimensions and high-representation dimensions according to different data acquisition methods of each dimension; the data acquisition methods include subjective acquisition and objective acquisition; A basic weight analysis module is used to obtain the basic weight of each dimension according to the distribution disorder of the anhedonia evaluation values ​​in each dimension among different patients and the degree of disorder of the anhedonia evaluation values ​​of each patient in different dimensions; The low-representation analysis module is used to continuously collect data for each dimension of each patient to obtain a preset number of historical anhedonia evaluation values ​​for each dimension; according to the changes in the historical anhedonia evaluation values ​​of each patient in each low-representation dimension, as well as the changes in the historical anhedonia evaluation values ​​and the consistency of data distribution between each low-representation dimension and the high-representation dimension, the low-representation weight of each patient in each dimension is obtained; A high representation analysis module is used to cluster patients based on their low representation weights in low representation dimensions to obtain patient clusters; according to the deviation of each high representation dimension of each patient in the patient cluster and the corresponding high representation dimension of other patients in the anhedonia evaluation value, determine the high representation weight of each patient in each dimension; An evaluation module, for obtaining an evaluation weight of each patient in each dimension according to the low representation weight, high representation weight and basic weight of each patient in each dimension; obtaining an evaluation result of each patient based on the evaluation weight combined with the anhedonia evaluation value of each patient in each dimension; The method for obtaining the evaluation weight includes: The product of the basic weight, low representation weight and high representation weight of each patient in each dimension is normalized to obtain the evaluation weight of each patient in each dimension; The evaluation result of each patient is obtained based on the evaluation weight and the anhedonia evaluation value of each patient in each dimension, including: The product of each patient's anhedonia evaluation value in each dimension and the corresponding evaluation weight is taken as the evaluation value of each patient in each dimension; The sum of the evaluation values ​​of each patient in all dimensions is taken as the evaluation result of each patient.

2. According to claim 1, a system for assisting the assessment of anhedonia in patients with depression using multimodal physiological parameters, characterized in that: The method for obtaining the basic weight includes: The entropy values ​​of each patient's anhedonia evaluation values ​​in all dimensions are negatively correlated and mapped as the influence weight of each patient; For any dimension, the frequency of occurrence of the anhedonia evaluation value corresponding to each patient of the dimension in the anhedonia evaluation values ​​corresponding to all patients of the dimension is taken as the distribution frequency of the dimension in each patient; The entropy value is calculated by weighting the distribution frequency corresponding to the dimension with the influence weight of each patient to obtain the basic weight of the dimension.

3. According to claim 1, a system for assisting the assessment of anhedonia in patients with depression using multimodal physiological parameters, characterized in that: The method for obtaining the low representation weight includes: For any patient, each low-representation dimension of the patient is taken as the analysis dimension in turn; According to the numerical consistency of all historical anhedonia evaluation values ​​of the patient under the analysis dimension, and the stability of the temporal changes of all historical anhedonia evaluation values, the first factor of the patient under the analysis dimension is obtained; According to the consistency of the change of the patient's historical anhedonia evaluation value between the analysis dimension and each high representation dimension and the closeness of the numerical distribution, the second factor of the patient under the analysis dimension is obtained; The first factor and the second factor of the patient in each low representation dimension are combined to determine the low representation weight of the patient in each dimension.

4. According to claim 3, a system for assisting the assessment of anhedonia in patients with depression using multimodal physiological parameters, characterized in that: The method for obtaining the first factor includes: Negative correlation mapping is performed on the standard deviation of all historical anhedonia evaluation values ​​of the patient under the analysis dimension to obtain the numerical stability of the patient under the analysis dimension; Arrange all the historical anhedonia evaluation values ​​of the patient under the analysis dimension in chronological order to obtain an evaluation value sequence; take the difference between each two adjacent historical anhedonia evaluation values ​​in the evaluation value sequence as the change difference; calculate the difference between the two change differences, calculate the difference sum and perform negative correlation mapping to obtain the change stability of the patient under the analysis dimension; The first factor of the patient under the analysis dimension is obtained by combining the numerical stability and change stability of the patient under the analysis dimension.

5. According to claim 3, a system for assisting the assessment of anhedonia in patients with depression using multimodal physiological parameters, characterized in that: The method for obtaining the second factor includes: All historical anhedonia evaluation values ​​of the patient in each dimension are sorted in chronological order to obtain the evaluation value sequence of the patient in each dimension; the mean of all historical anhedonia evaluation values ​​of the patient in each dimension is taken as the mean evaluation value of the patient in each dimension; After calculating the Pearson correlation coefficient between the evaluation value sequence of the patient in the analysis dimension and each high representation dimension, the sum of the Pearson correlation coefficients is calculated to obtain the correlation of the changes in the analysis dimension of the patient; After calculating the difference between the patient's evaluation mean in the analysis dimension and each high-representation dimension, the difference and value are calculated and negative correlation mapping is performed to obtain the patient's numerical correlation in the analysis dimension; The change correlation and value correlation of the patient in the analysis dimension are combined to obtain the second factor of the patient in the analysis dimension.

6. According to claim 3, a system for assisting the assessment of anhedonia in patients with depression using multimodal physiological parameters, characterized in that: The combining the first factor and the second factor of the patient in each low representation dimension to determine the low representation weight of the patient in each dimension includes: Normalizing the product of the first factor and the second factor of the patient in each low representation dimension to obtain the low representation weight of the patient in each low representation dimension; The low representation weight of the patient in each high representation dimension is set to a preset weight value.

7. According to claim 1, a system for assisting the assessment of anhedonia in patients with depression using multimodal physiological parameters, characterized in that: The method for obtaining the patient clusters includes: The low representation weights of each patient in all low representation dimensions are combined into a representation vector for each patient; the similarity of the representation vectors is used as a metric to cluster the patients through a clustering algorithm to obtain patient clusters.

8. According to claim 1, a system for assisting the assessment of anhedonia in patients with depression using multimodal physiological parameters, characterized in that: The method for obtaining the high representation weight includes: For any patient, each high representation dimension of the patient is taken as the target dimension in turn; Calculate the difference between the anhedonia evaluation value of the patient and that of every other patient in the patient cluster in the target dimension as the deviation value of the patient and every other patient in the target dimension; normalize the sum of the deviation values ​​of the patient and all other patients in the target dimension as the high representation dimension of the patient in the target dimension; The high representation weight of the patient in each low representation dimension is set to a preset weight value.

Images