Management system based on VBT short-range psychotherapy

Abstract

The invention provides a management system based on VBT short-range psychotherapy, and relates to the technical field of psychotherapy. Comprising a data acquisition module, a data preprocessing module and a data analysis module which are connected in sequence, the data analysis module is connected with a personalized treatment module and an early warning module, the early warning module is connected with a report evaluation module, and an adjustment module is firstly connected with the personalized treatment module; a complete, efficient and intelligent mental health management solution is provided by integrating a multi-source data acquisition module, an advanced data preprocessing and analysis module, a personalized treatment scheme generation and optimization module, a dynamic adjustment module, a real-time early warning module and the like. The psychotherapy system solves the problems of poor efficiency and low individuation of the psychotherapy system in the prior art.

Description

Technical Field

[0001] The present invention relates to the technical field of psychotherapy, and particularly to a management system based on VBT short-term psychotherapy. Background Art

[0002] With the increase of social pressure, the incidence of mental diseases shows an upward trend, and effective monitoring and intervention means are urgently needed. The existing mental disease monitoring systems mainly include the methods and systems described in the publication numbers: CN 111599441 B and CN101485574 B. CN 111599441 B proposes a mental disease monitoring method and system, which realizes the automatic monitoring of the patient's mental state through data collection, analysis and early warning modules, reduces the influence of doctors' subjective factors, and is convenient for patients' self-management. However, this system has deficiencies in the depth of data analysis and personalized intervention, and it is difficult to provide targeted treatment plans.

[0003] CN 101485574 B introduces a rapid mental adjustment intelligent system based on deep learning. This system combines the client and the server, and stores and retrieves psychotherapy plans through deep learning technology, aiming to improve the scientific and personalized level of mental intervention. However, this system still relies on preset treatment plans, lacks a dynamic adjustment and real-time feedback mechanism, and cannot fully adapt to the immediate needs of patients and the changes in mental states.

[0004] Although the existing technologies have made certain progress in the monitoring and preliminary intervention of mental diseases, there are still obvious deficiencies in personalization, dynamic adjustment and short-term and efficient treatment. Therefore, based on the VBT (Value-Based Therapy) theory, there is an urgent need for a short-term psychotherapy management system that can combine automatic monitoring, deep learning and value orientation to achieve more efficient and personalized mental health management. Summary of the Invention

[0005] In order to overcome the deficiencies of the prior art, the purpose of the present invention is to provide a management system based on VBT short-term psychotherapy.

[0006] To achieve the above purpose, the present invention provides the following solutions:

[0007] A management system based on VBT short-term psychotherapy, comprising:

[0008] A data collection module, a data preprocessing module, a data analysis module, a personalized treatment module, an adjustment module, an early warning module and a report evaluation module;

[0009] The data acquisition module, the data preprocessing module, and the data analysis module are connected in sequence. The data analysis module is respectively connected to the personalized treatment module and the warning module. The warning module is connected to the report evaluation module. The adjustment module and the personalized treatment module are first connected to each other;

[0010] The data acquisition module is used to obtain multi-source data of the patient to be tested. The data preprocessing module is used to preprocess the multi-source data to obtain preprocessed multi-source data. The data analysis module is used to convert the preprocessed multi-source data into multi-dimensional tensor data and perform feature extraction to obtain high-order features, and input the high-order features into an improved multi-modal deep neural network to obtain the matching degree of values and behavior patterns and the psychological state score, and obtain the psychological state evaluation result according to the matching degree of values and behavior patterns and the psychological state score. The personalized treatment module is used to generate an initial treatment plan according to the matching degree of values and behavior patterns and the psychological state score, and optimize the initial data plan according to historical treatment data and the VBT theory to obtain an optimized treatment plan. The adjustment module is used to track the response of the patient to be tested or receiving the optimized treatment plan and adjust the optimized treatment plan according to the response. The warning module is used to receive the psychological state evaluation result and compare it with the adaptive threshold to obtain a comparison result and issue different levels of alarms according to the comparison result. The report evaluation module is used to generate visual warning data according to the alarms of the warning module.

[0011] Preferably, the data preprocessing module includes:

[0012] An original data acquisition sub-module, a data conversion sub-module, a fusion sub-module, and a feature extraction sub-module;

[0013] The original data acquisition sub-module is used to extract the original data set from the multi-modal data. Among them, the multi-modal data includes: physiological data, behavioral data, self-report data, and social media data. The data conversion sub-module is used to convert the data in the original data set into an input format data set. The fusion sub-module is used to perform multi-dimensional tensor representation on the data in the input format data set and perform tensor fusion to obtain fusion data. The feature extraction sub-module is used to extract the high-order features of the fusion data.

[0014] Preferably, the data analysis module includes:

[0015] A data cleaning sub-module and a model building sub-module;

[0016] The data cleaning sub-module is used to clean the high-order features to obtain the cleaned high-order features. The model construction sub-module is used to construct a multi-modal neural network based on the cleaned high-order features. Among them, the multi-modal neural network includes an input layer, a branch layer, and a fusion layer. Among them, the input layer is used to input the cleaned high-order features. The branch layer includes a mental state branch and a values matching branch. Among them, the mental state branch contains multiple LSTM layers, which are used to process the time series data in the cleaned high-order features to obtain a mental state score. The values matching branch contains multiple GNN layers, which are used to process the social interaction data in the cleaned high-order features to obtain the matching degree between values and behavior patterns. The fusion layer is used to obtain a mental state evaluation result according to the matching degree between values and behavior patterns and the mental state score.

[0017] Preferably, the expression for the matching degree between values and behavior patterns is:

[0018]

[0019] Where V m is the matching degree between values and behavior patterns, V i is the importance score of the i-th type of values, B i is the compliance score of the i-th type of behavior pattern, Similarity is the similarity function between values and behavior patterns, and N is a natural number.

[0020] 5. A management system based on VBT short-term psychotherapy according to claim 4, wherein the expression for the mental state evaluation result is:

[0021] S c = αS p + βV m ;

[0022] Where α and β are the first weight coefficient and the second weight coefficient respectively, and S p is the mental state score.

[0023] Preferably, the adjustment module includes:

[0024] A feedback data acquisition sub-module, a response sub-module, a reinforcement learning sub-module, and a transmission sub-module;

[0025] The feedback data acquisition sub-module is used to acquire the feedback data of the patient to be tested. Among them, the feedback data includes: activity level, emotional change, and self-report. The response sub-module is used to perform a response evaluation on the current treatment status of the patient to be tested according to the feedback data, and obtain a response evaluation result. The reinforcement learning sub-module is used to perform reinforcement learning adjustment on the initial treatment plan according to the response evaluation result, and obtain an adjusted treatment plan. The transmission sub-module is used to send the current adjusted treatment plan to the patient to be tested.

[0026] Preferably, the early warning module includes:

[0027] An early warning score calculation sub-module, a threshold setting sub-module, and an early warning determination sub-module;

[0028] The early warning score calculation sub-module is used to calculate an early warning score according to the psychological state evaluation result, and obtain an early warning score. The threshold setting sub-module is used to set a threshold based on historical early warning data, and obtain an early warning threshold. The early warning determination sub-module is used to perform an early warning judgment according to the early warning score and the early warning threshold, and obtain an early warning level.

[0029] Preferably, the calculation expression of the early warning score is:

[0030] P = αS′ c + βV′ m + γR′;

[0031] Among them, α, β, and γ are the first weight coefficient, the second weight coefficient, and the third weight coefficient respectively, P is the early warning score, and S′ c is the standardized psychological state score, V′ m is the standardized value matching degree, and R′ is the standardized score of the most recent risk-related data.

[0032] Preferably, the calculation expression of the threshold is:

[0033] θ P = μ P + kσ P ;

[0034] Among them, μ P is the historical mean of the early warning score, k is an adjustment factor, and σ P is the historical standard deviation of the early warning score.

[0035] The present invention discloses the following technical effects:

[0036] The present invention provides a management system based on VBT short-term psychotherapy, including: a data acquisition module, a data preprocessing module, a data analysis module, a personalized treatment module, an adjustment module, an early warning module, and a report evaluation module; the data acquisition module, the data preprocessing module, and the data analysis module are connected in sequence, the data analysis module is respectively connected to the personalized treatment module and the early warning module, the early warning module is connected to the report evaluation module, and the adjustment module and the personalized treatment module are first connected to each other; the data acquisition module is used to obtain multi-source data of the patient to be tested, the data preprocessing module is used to preprocess the multi-source data to obtain preprocessed multi-source data, the data analysis module is used to convert the preprocessed multi-source data into multi-dimensional tensor data and perform feature extraction to obtain high-order features and input the high-order features into an improved multi-modal deep neural network to obtain the matching degree of values and behavior patterns and the psychological state score and obtain a psychological state evaluation result according to the matching degree of values and behavior patterns and the psychological state score, the personalized treatment module is used to generate an initial treatment plan according to the matching degree of values and behavior patterns and the psychological state score and optimize the initial data plan according to historical treatment data and VBT theory to obtain an optimized treatment plan, the adjustment module is used to track the response of the patient to be tested or receiving the optimized treatment plan and adjust the optimized treatment plan according to the response, the early warning module is used to receive the psychological state evaluation result and compare it with an adaptive threshold to obtain a comparison result and issue different levels of alarms according to the comparison result, and the report evaluation module is used to generate visual warning data according to the alarms of the early warning module. The management system based on VBT short-term psychotherapy of the present invention provides a complete, efficient, and intelligent mental health management solution by integrating modules such as multi-source data acquisition, advanced data preprocessing and analysis, personalized treatment plan generation and optimization, dynamic adjustment, and real-time early warning. Its main beneficial effects include: comprehensiveness: the integration and analysis of multi-source data provide a comprehensive mental health assessment. Precision: deep neural networks and multi-modal data processing improve the accuracy of assessment and early warning. Personalization: personalized treatment plans and dynamic adjustment mechanisms meet the needs of different patients. Real-time: real-time early warning mechanisms ensure timely risk management and intervention. Optimization: continuous reinforcement learning optimization improves the adaptive ability and long-term effects of the system. Visualization: visual reports enhance the effectiveness of information communication and decision support. Description of the Drawings

[0037] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required in the embodiments. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0038] Figure 1 It is a schematic structural diagram of a management system based on VBT short-term psychotherapy provided by an embodiment of the present invention.

[0039] Explanation of reference numerals:

[0040] 1 - Data acquisition module, 2 - Data preprocessing module, 3 - Data analysis module, 4 - Personalized treatment module, 5 - Adjustment module, 6 - Early warning module, 7 - Report evaluation module. Detailed implementation manners

[0041] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the drawings in the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, rather than all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present invention.

[0042] To make the above objects, features, and advantages of the present invention more obvious and understandable, the present invention will be further described in detail below with reference to the drawings and specific implementation manners.

[0043] As Figure 1 shown, the present invention provides a management system based on VBT short-term psychotherapy, including:

[0044] Data acquisition module 1, data preprocessing module 2, data analysis module 3, personalized treatment module 4, adjustment module 5, early warning module 6, and report evaluation module 7;

[0045] The data acquisition module 1, the data preprocessing module 2, and the data analysis module 3 are connected in sequence. The data analysis module 3 is respectively connected to the personalized treatment module 4 and the early warning module 6. The early warning module 6 is connected to the report evaluation module 7. The adjustment module 5 and the personalized treatment module 4 are connected to each other first;

[0046] The data acquisition module 1 is used to obtain multi-source data of the patient to be tested. The data preprocessing module 2 is used to preprocess the multi-source data to obtain preprocessed multi-source data. The data analysis module 3 is used to convert the preprocessed multi-source data into multi-dimensional tensor data and perform feature extraction to obtain high-order features, and input the high-order features into an improved multi-modal deep neural network to obtain the matching degree of values and behavior patterns and the psychological state score, and obtain the psychological state evaluation result according to the matching degree of values and behavior patterns and the psychological state score. The personalized treatment module 4 is used to generate an initial treatment plan according to the matching degree of values and behavior patterns and the psychological state score, and optimize the initial data plan according to historical treatment data and VBT theory to obtain an optimized treatment plan. The adjustment module 5 is used to track the response of the patient to be tested or receiving the optimized treatment plan and adjust the optimized treatment plan according to the response. The warning module 6 is used to receive the psychological state evaluation result and compare it with an adaptive threshold to obtain a comparison result and issue alerts of different levels according to the comparison result. The report evaluation module 7 is used to generate visual warning data according to the alerts of the warning module 6.

[0047] Further, the data preprocessing module 2 includes:

[0048] An original data acquisition sub-module, a data conversion sub-module, a fusion sub-module, and a feature extraction sub-module;

[0049] The original data acquisition sub-module is used to extract an original data set from the multi-modal data. Among them, the multi-modal data includes: physiological data, behavioral data, self-report data, and social media data. The data conversion sub-module is used to convert the data in the original data set into an input format data set. The fusion sub-module is used to perform multi-dimensional tensor representation on the data in the input format data set and perform tensor fusion to obtain fused data. The feature extraction sub-module is used to extract high-order features of the fused data.

[0050] More specific, multi-source data types: Physiological data: heart rate, blood pressure, respiratory rate, body temperature, sleep quality, activity level, steps, skin conductance response, sweat secretion; Behavioral data: daily activity records, exercise volume, work and rest time allocation, social interaction frequency and quality, self-report data, emotion diaries, emotion scores; Psychological assessment questionnaires (such as PHQ-9, GAD-7): behavioral records and reflection logs, social media and communication data, post content analysis, sentiment tendency, interaction frequency, social relationship network, voice call emotion analysis; Voice and text data: voice call recordings, intonation changes, text chat records, emotion analysis results; Virtual reality (VR) interaction data: user movement trajectories in the VR environment, interaction behavior records, reaction times, changes in physiological indicators in VR; Environmental data: noise level, light intensity, temperature, humidity, air quality index, layout and changes of the living environment.

[0051] Specifically, data acquisition channels: Reception: Visitors can fill in form materials in the VR scene, including basic information, personality, living environment of the visitor, and the intimacy of family members, so that therapists can understand the overall situation of the visitor. There is an important goal in this module - to further enhance the visitor's treatment motivation by discussing the possibility of change with the visitor.

[0052] Assessment: Visitors can complete scale evaluations in the VR scene.

[0053] Analysis and tracing: Help visitors understand their own personality psychological characteristics and match corresponding audio-visual files according to the results of some scales.

[0054] Adjusting core beliefs: Scene 1: Show two value orientation models; Scene 2: Show "through awareness"; Scene 3: Show "boundary awareness"; Scene 4: Show "psychological energy system"; Scene 5: Show "hot coping" and "cold coping"; Scene 6: Show "expectation / actual feedback problem".

[0055] Problem presentation: Visitors can select their existing problem labels in the VR scene. The problems are divided into two categories, one is symptom-level problems and the other is reality-level problems.

[0056] Family talks: Present questionnaires in voice form.

[0057] Summary: According to the choices of visitors in problem presentation, corresponding to symptom problem modules such as insomnia, anxiety, depression, compulsion, physical discomfort, etc., and reality problem modules such as interpersonal relationships, learning pressure, family problems, self-cognition, etc. Visitors click on each module to match the corresponding summary audio-visual files.

[0058] Furthermore, the data analysis module 3 includes:

[0059] Data cleaning sub-module and model construction sub-module;

[0060] The data cleaning sub-module is used to clean the high-order features to obtain the cleaned high-order features. The model construction sub-module is used to construct a multi-modal neural network based on the cleaned high-order features. Among them, the multi-modal neural network includes an input layer, a branch layer, and a fusion layer. Among them, the input layer is used to input the cleaned high-order features. The branch layer includes a mental state branch and a values matching branch. Among them, the mental state branch contains multiple LSTM layers, which are used to process the time series data in the cleaned high-order features to obtain a mental state score. The values matching branch contains multiple GNN layers, which are used to process the social interaction data in the cleaned high-order features to obtain the matching degree between values and behavior patterns. The fusion layer is used to obtain a mental state evaluation result based on the matching degree between values and behavior patterns and the mental state score.

[0061] Specifically, for data verification and integrity check: Step 1.1: Receive the high-order feature dataset from the data preprocessing module 2. Step 1.2: Check whether each feature in the dataset is complete to ensure that all necessary features of each sample have values. Step 1.3: Identify and mark missing values (NaN), outliers (such as extreme outliers), and duplicate data.

[0062] Missing value handling

[0063] Step 2.1: Statistically calculate the proportion of missing values for each feature. Step 2.2: For features with a relatively low proportion of missing values, use interpolation methods (such as mean interpolation, median interpolation) to fill in the missing values. Step 2.3: For features with a relatively high proportion of missing values that cannot be reasonably filled by interpolation, consider deleting the feature or using model prediction methods to fill it.

[0064] Outlier detection and handling

[0065] Step 3.1: Use statistical methods (such as Z-Score, IQR) to detect outliers in each feature. Step 3.2: For the detected outliers, evaluate whether they are real extreme values or data entry errors.

[0066] Step 3.3: For outliers that are data entry errors, correct or delete them; for real extreme values, decide whether to retain them according to the specific situation.

[0067] Data standardization and normalization

[0068] Step 4.1: According to the data distribution, select a suitable standardization method (such as Z-Score standardization) or normalization method (such as Min-Max normalization).

[0069] Step 4.2: Standardize or normalize continuous features to eliminate the impact of different dimensions on model training.

[0070] Step 4.3: Encode categorical features (such as one-hot encoding, label encoding) to meet the model input requirements.

[0071] Data deduplication and consistency verification

[0072] Step 5.1: Check whether there are duplicate samples in the dataset and merge or delete the duplicate samples.

[0073] Step 5.2: Ensure the consistency within the dataset, such as the order of timestamps, the logical relationship between features, etc.

[0074] Feature selection and dimensionality reduction

[0075] Step 6.1: Use statistical methods (such as correlation coefficient analysis, chi-square test) or model methods (such as Lasso regression, tree model) for feature selection to eliminate redundant or irrelevant features.

[0076] Step 6.2: Adopt dimensionality reduction techniques (such as principal component analysis PCA, t-SNE) to reduce the feature dimensionality, retain the main information, and improve the model training efficiency.

[0077] Furthermore, the expression for the matching degree between the values and behavior patterns is:

[0078]

[0079] where V m is the matching degree between the values and behavior patterns, V i is the importance score of the i-th type of values, B i is the compliance score of the i-th type of behavior patterns, Similarity is the similarity function between the values and behavior patterns, and N is a natural number.

[0080] Furthermore, the expression for the psychological state assessment result is:

[0081] S c = αS p + βV m ;

[0082] where α and β are the first weight coefficient and the second weight coefficient respectively, and S p is the psychological state score.

[0083] Specifically, model architecture design

[0084] Step 1.1: Define the overall architecture of the multimodal neural network, including the input layer, branch layer, and fusion layer.

[0085] Step 1.2: Design the input layer, determine the dimensions and formats of each input feature, and ensure they match the cleaned high-order feature data.

[0086] Step 1.3: Design the branch layer, including the mental state branch and the values matching branch respectively, and determine the network structures and parameters of each branch.

[0087] Step 1.4: Design the fusion layer, define the fusion method for the mental state score and the values matching degree, and generate the final mental state assessment result.

[0088] Construction of the input layer

[0089] Step 2.1: Configure the input layer to accept the cleaned high-order feature data from the data cleaning sub-module.

[0090] Step 2.2: According to the data type, set different input processing methods, such as setting the time step for sequence data and configuring the adjacency matrix for graph data, etc.

[0091] Construction of the mental state branch

[0092] Step 3.1: Build a multi-layer LSTM (Long Short-Term Memory Network) structure in the branch layer to process time series data.

[0093] Step 3.2: Set an appropriate number of hidden units and activation functions (such as tanh or ReLU) for each LSTM layer to capture time dependencies and patterns.

[0094] Step 3.3: Add a fully connected layer (Dense Layer) after the LSTM layer to map the LSTM output to the mental state score space.

[0095] Step 3.4: Apply an appropriate activation function (such as sigmoid or softmax) to output the mental state score.

[0096] Construction of the values matching branch

[0097] Step 4.1: Build multiple GNN (Graph Neural Network) layers in the branch layer to process the graph structure information of social interaction data.

[0098] Step 4.2: Set the node feature dimension, adjacency matrix, and aggregation function (such as GraphConvolution, GraphAttention) for each GNN layer.

[0099] Step 4.3: Add a fully connected layer after the GNN layer to map the GNN output to the values and behavior patterns matching degree space.

[0100] Step 4.4: Apply an appropriate activation function to output the matching degree between values and behavior patterns.

[0101] Fusion layer construction

[0102] Step 5.1: Collect the output results of the psychological state scores and the value matching degree as the input of the fusion layer.

[0103] Step 5.2: Design a fusion mechanism, which can be simple concatenation, weighted sum, or a more complex attention mechanism.

[0104] Step 5.3: Further process the fused features through one or more fully connected layers to generate the final psychological state evaluation result.

[0105] Step 5.4: Apply an appropriate activation function (such as linear activation or sigmoid) to output the final evaluation result.

[0106] Model compilation and configuration

[0107] Step 6.1: Select a suitable loss function (such as mean squared error MSE, cross-entropy loss, etc.) according to the specific task requirements.

[0108] Step 6.2: Select an optimization algorithm (such as Adam, SGD, RMSprop) to optimize the model parameters.

[0109] Step 6.3: Set evaluation metrics (such as accuracy, F1 score, AUC, etc.) to monitor the model performance.

[0110] Model training

[0111] Step 7.1: Divide the cleaned high-order feature dataset into a training set, a validation set, and a test set.

[0112] Step 7.2: Use the training set to train the model and optimize the model parameters through multiple iterations (Epochs).

[0113] Step 7.3: During the training process, use the validation set for real-time monitoring to prevent the model from overfitting and adjust the hyperparameters (such as learning rate, batch size).

[0114] Model evaluation and optimization

[0115] Step 8.1: Evaluate the performance of the model on the test set and calculate the preset evaluation metrics.

[0116] Step 8.2: According to the evaluation results, adjust the model architecture or optimize the parameters to improve the accuracy and robustness of the model.

[0117] Step 8.3: Use cross-validation technology to further verify the generalization ability of the model.

[0118] Model deployment and integration

[0119] Step 9.1: Deploy the trained multimodal neural network model to the production environment for practical application.

[0120] Step 9.2: Integrate the data interface between the model and other modules (such as personalized treatment module 4 and early warning module 6) to ensure smooth transmission of data flow.

[0121] Step 9.3: Set up a model monitoring mechanism to monitor the performance of the model in actual applications in real time and identify and resolve potential problems in a timely manner.

[0122] Model maintenance and updates

[0123] Step 10.1: Regularly collect new data and feedback to evaluate the long-term performance of the model.

[0124] Step 10.2: Retrain or fine-tune the model based on the latest data and requirements to keep the model advanced and adaptable.

[0125] Step 10.3: Record model versions and change logs to ensure traceability and transparency of the model maintenance process.

[0126] Furthermore, the adjustment module 5 includes:

[0127] Feedback data acquisition submodule, response submodule, reinforcement learning submodule and transmission submodule;

[0128] The feedback data acquisition submodule is used to obtain feedback data from the patient to be tested, wherein the feedback data includes: activity level, emotional changes, and self-reports; the response submodule is used to perform a response evaluation on the current treatment status of the patient to be tested based on the feedback data to obtain a response evaluation result; the reinforcement learning submodule is used to perform reinforcement learning adjustment on the initial treatment plan based on the response evaluation result to obtain an adjusted treatment plan; and the transmission submodule is used to send the current adjusted treatment plan to the patient to be tested.

[0129] Furthermore, the early warning module 6 includes:

[0130] Early warning score calculation submodule, threshold setting submodule and early warning determination submodule;

[0131] The early warning score calculation sub-module is used to calculate the early warning score based on the psychological state evaluation result to obtain the early warning score. The threshold setting sub-module is used to set the threshold based on historical early warning data to obtain the early warning threshold. The early warning determination sub-module is used to make an early warning judgment based on the early warning score and the early warning threshold to obtain the early warning level.

[0132] Specifically, when P is greater than θ P the early warning condition is triggered:

[0133]

[0134] According to the early warning level, the corresponding early warning information is generated and pushed to the patient and relevant medical staff through multiple channels (such as mobile applications, text messages, emails).

[0135] Furthermore, the calculation expression of the early warning score is:

[0136] P = αS' c + βV' m + γR';

[0137] where α, β, and γ are the first weight coefficient, the second weight coefficient, and the third weight coefficient respectively, P is the early warning score, and S' c is the standardized psychological state score, V' m is the standardized value matching degree, and R' is the standardized score of the most recent risk-related data.

[0138] Furthermore, the calculation expression of the threshold is:

[0139] θ P = μ P + kσ P ;

[0140] where μ P is the historical mean of the early warning score, k is the adjustment factor, and σ P is the historical standard deviation of the early warning score.

[0141] In this specification, each embodiment is described in a progressive manner. The key point of each embodiment is to illustrate the differences from other embodiments. The same or similar parts among the embodiments can be referred to each other.

[0142] Specific examples are used in this article to elaborate on the principle and implementation manner of the present invention. The description of the above embodiments is only used to help understand the method and its core idea of the present invention; at the same time, for those of ordinary skill in the art, there will be changes in the specific implementation manner and application scope according to the idea of the present invention. In summary, the content of this specification should not be construed as a limitation to the present invention.

Claims

1. A management system based on VBT short-term psychotherapy, characterized in that: include: Data collection module, data preprocessing module, data analysis module, personalized treatment module, adjustment module, early warning module and report evaluation module; The data acquisition module, the data preprocessing module, and the data analysis module are connected in sequence, the data analysis module is connected to the personalized treatment module and the early warning module respectively, the early warning module is connected to the report evaluation module, and the adjustment module is first connected to the personalized treatment module; The data acquisition module is used to obtain multi-source data of the patient to be tested, the data preprocessing module is used to preprocess the multi-source data to obtain preprocessed multi-source data, the data analysis module is used to convert the preprocessed multi-source data into multi-dimensional tensor data and perform feature extraction to obtain high-order features and input the high-order features into the improved multimodal deep neural network to obtain the matching degree of values ​​and behavior patterns and the psychological state score, and obtain the psychological state evaluation result according to the matching degree of values ​​and behavior patterns and the psychological state score, the personalized treatment module is used to generate an initial treatment plan according to the matching degree of values ​​and behavior patterns and the psychological state score, and optimize the initial data plan according to historical treatment data and VBT theory to obtain an optimized treatment plan, the adjustment module is used to track the response of the patient to be tested or receiving the optimized treatment plan and adjust the optimized treatment plan according to the response, the early warning module is used to receive the psychological state evaluation result and compare it with the adaptive threshold to obtain the comparison result and issue different degrees of alarms according to the comparison result, and the report evaluation module is used to generate visual alarm data according to the alarm of the early warning module.

2. A management system based on VBT short-term psychotherapy according to claim 1, characterized in that: The data preprocessing module comprises: Raw data acquisition submodule, data conversion submodule, fusion submodule and feature extraction submodule; The raw data acquisition submodule is used to extract the raw data set according to the multimodal data, wherein the multimodal data includes: physiological data, behavioral data, self-report data, and social media data. The data conversion submodule is used to convert each data in the raw data set into an input format data set. The fusion submodule is used to represent the data in the input format data set as a multidimensional tensor and perform tensor fusion to obtain fused data. The feature extraction submodule is used to extract high-order features of the fused data.

3. A management system based on VBT short-term psychotherapy according to claim 1, characterized in that: The data analysis module comprises: Data cleaning submodule and model building submodule; The data cleaning submodule is used to clean the high-order features to obtain the cleaned high-order features, and the model building submodule is used to build a multimodal neural network based on the cleaned high-order features, wherein the multimodal neural network includes an input layer, a branch layer and a fusion layer, wherein the input layer is used to input the cleaned high-order features, the branch layer includes a psychological state branch and a value matching branch, wherein the psychological state branch contains a multi-layer LSTM for processing the time series data in the cleaned high-order features to obtain a psychological state score, the value matching branch contains a plurality of GNN layers for processing the social interaction data in the cleaned high-order features to obtain a matching degree between values ​​and behavior patterns, and the fusion layer is used to obtain a psychological state evaluation result based on the matching degree between the values ​​and behavior patterns and the psychological state score.

4. A management system based on VBT short-term psychotherapy according to claim 1, characterized in that: The expression of the matching degree between the values ​​and the behavior patterns is: Among them, V m is the matching degree between values ​​and behavior patterns, V i Score the importance of the i-th category of values, B i is the conformity score of the i-th behavior pattern, Similarity is the similarity function between values ​​and behavior patterns, and N is a natural number.

5. A management system based on VBT short-term psychotherapy according to claim 4, characterized in that: The expression of the psychological state evaluation result is: S c =αS p +βV m ; Among them, α and β are the first weight coefficient and the second weight coefficient respectively, S p Score your mental state.

6. A management system based on VBT short-term psychotherapy according to claim 4, characterized in that: The adjustment module comprises: Feedback data acquisition submodule, response submodule, reinforcement learning submodule and transmission submodule; The feedback data acquisition submodule is used to obtain feedback data from the patient to be tested, wherein the feedback data includes: activity level, emotional changes, and self-reports. The response submodule is used to perform a response evaluation on the current treatment status of the patient to be tested based on the feedback data to obtain a response evaluation result. The reinforcement learning submodule is used to perform reinforcement learning adjustment on the initial treatment plan based on the response evaluation result to obtain an adjusted treatment plan. The transmission submodule is used to send the current adjusted treatment plan to the patient to be tested.

7. A management system based on VBT short-term psychotherapy according to claim 1, characterized in that: The early warning module comprises: Early warning score calculation submodule, threshold setting submodule and early warning determination submodule; The early warning score calculation submodule is used to calculate the early warning score according to the psychological state assessment result to obtain the early warning score. The threshold setting submodule is used to set the threshold based on the historical early warning data to obtain the early warning threshold. The early warning judgment submodule is used to make an early warning judgment based on the early warning score and the early warning threshold to obtain the early warning degree.

8. A management system based on VBT short-term psychotherapy according to claim 7, characterized in that: The calculation expression of the early warning score is: P=αS′ c +βV′ m +γR′; Among them, α, β, γ are the first weight coefficient, the second weight coefficient, and the third weight coefficient respectively, P is the early warning score, S′ c is the standardized psychological state score, V′ m is the standardized value match, and R′ is the standardized score of the most recent risk-related data.

9. A management system based on VBT short-term psychotherapy according to claim 7, characterized in that: The calculation expression of the threshold is: i P =μ P +kσ P ; Among them, μ P is the historical mean of the early warning score, k is the adjustment factor, σ P The historical standard deviation of the warning score.

Images