Artificial intelligence assisted psychological assessment system

Abstract

The invention relates to the technical field of artificial intelligence, and discloses an artificial intelligence assisted psychological assessment system. The artificial intelligence assisted psychological assessment system comprises a data collection module, a data analysis module, a scale generation module, a result presentation module, a user interaction module, an online AI consultation module and a modal selection and decision optimization module, according to the artificial intelligence assisted psychological assessment method, personalized and accurate psychological health assessment and intervention are realized through the steps of multi-modal data collection, psychological state modeling, data analysis, scale generation, result presentation, modal optimization, online consultation and the like, and the assessment efficiency and accuracy are improved; according to the method, accurate and personalized psychological assessment and intervention are provided through multi-modal data collection and analysis, information entropy optimization, variational reasoning data analysis and personalized scale generation; compared with the prior art, the accuracy and efficiency of evaluation and the user experience are improved.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and specifically to an artificial intelligence-assisted psychological assessment system. Background Art

[0002] In traditional psychological assessments, self-assessment questionnaires are usually adopted, and the psychological state is evaluated by the user answering a series of standardized questions; although this method can provide certain references, due to the limitation of a single data source, its assessment results are often not comprehensive enough; for example, many assessment methods only rely on the user's text input and ignore the potential psychological information of the user in voice, facial expressions or behaviors; such a method cannot comprehensively capture the user's psychological fluctuations and is prone to limited accuracy of the assessment.

[0003] The existing technologies generally have the problem of ignoring individual differences; most psychological assessment tools adopt unified assessment criteria and fixed scales and cannot be adjusted according to the unique needs of each user; the user's psychological state may be affected by various factors, but the existing methods mostly rely on fixed assessment templates and problem dimensions and fail to effectively make personalized adjustments according to the actual situation and real-time feedback of the user; this makes many assessment results at risk of being overly simplified and unable to comprehensively reflect the user's true psychological state.

[0004] In addition, the existing psychological assessment methods mostly rely on manual or traditional statistical analysis methods, with low calculation efficiency; as the amount of user data increases, the traditional methods have obvious bottlenecks in the processing of high-dimensional data; the existing technologies usually cannot quickly process and analyze multi-modal data, especially when facing a large number of users, the generation and feedback speed of the assessment results are slow; in contrast, the lack of an efficient automatic analysis and optimization mechanism limits the application of the existing technologies in large-scale mental health management; therefore, the present invention proposes an artificial intelligence-assisted psychological assessment system to solve the deficiencies of the existing technologies. Summary of the Invention

[0005] Aiming at the deficiencies of the existing technologies, the present invention provides an artificial intelligence-assisted psychological assessment system, which solves the limitations of traditional assessment methods in multi-modal data fusion, personalized adjustment and efficient data analysis, and improves the accuracy, personalization and processing efficiency of psychological assessments.

[0006] To achieve the above objectives, the present invention is realized through the following technical solutions: An artificial intelligence-assisted psychological assessment system, comprising:

[0007] A data collection module, used for collecting the psychological assessment data of the user and transmitting the collected data to the data analysis module for processing;

[0008] A data analysis module, which is used to analyze the collected data using machine learning algorithms and provide the analysis results to the scale generation module and the result presentation module;

[0009] A scale generation module, which is used to automatically generate scales, questions, dimensions and answers required for psychological assessment, and dynamically adjust the assessment content according to the results of the data analysis module;

[0010] A result presentation module, which is used to present the analysis results to the user in a graphical manner and provide assessment feedback information to the user interaction module;

[0011] A user interaction module, which is used to allow the user to access the analysis results, assessment content, assessment feedback information and data to obtain more detailed psychological assessment information, and provide interaction data to the data collection module and the online AI consultation module;

[0012] An online AI consultation module, which is used to provide online artificial intelligence consultation services, provide personalized suggestions according to user needs, and optimize consultation feedback in combination with the data of the result presentation module;

[0013] A modality selection and decision optimization module, which is used to calculate the contribution degrees of different data modalities, optimize the psychological assessment decision scheme, and provide the optimal modality combination information to the data analysis module.

[0014] The present invention also provides an artificial intelligence-assisted psychological assessment method, including the following steps:

[0015] S1. Data collection: Collect the text, voice, expression and drawing data of the user, and preprocess the multi-modal data to prepare for subsequent psychological state modeling and data analysis;

[0016] S2. Psychological state modeling: Use a probabilistic graphical model to construct a hierarchical representation of the psychological state, and map the collected multi-modal data to the potential psychological feature space;

[0017] S3. Data analysis: Use machine learning algorithms to analyze the user's psychological data, and optimize the calculation efficiency through variational inference to obtain the user's psychological state distribution;

[0018] S4. Scale generation: Dynamically generate a personalized psychological assessment scale according to the analysis results and the user's individual characteristics to meet the user's needs;

[0019] S5. Result presentation: Display the analysis results of the psychological assessment through data visualization methods, and provide an analysis of the psychological change trend in combination with the user's historical data;

[0020] S6. Modality selection and optimization: Calculate the contribution degrees of multi-modal data based on information entropy, and optimize the modality weights to ensure the accuracy and personalization of the decision-making process;

[0021] S7. Online consultation. Based on the results of psychological state analysis, personalized mental health advice is provided to users through the online AI consultation module, helping users further understand their psychological conditions and make corresponding adjustments.

[0022] The present invention provides an artificial intelligence-assisted psychological assessment system, which has the following beneficial effects:

[0023] 1. The present invention adopts multi-modal data collection and analysis technologies, combines text, voice, facial expressions and drawing data, comprehensively evaluates the psychological state of users, and achieves a more comprehensive and accurate mental health assessment effect; compared with the assessment methods with a single data source in the prior art, the present invention can capture the psychological changes of users from multiple dimensions, and solves the problem that the traditional methods fail to fully reflect the multi-faceted emotional fluctuations of users.

[0024] 2. The present invention dynamically optimizes the contribution degrees of different data modalities through information entropy calculation, ensuring that the weight distribution in the system decision-making process is more scientific and personalized; this technical solution can adjust the modality weights according to different situations. Compared with the fixed weight method in the prior art, the present invention effectively solves the problem of insufficient contribution degree of a specific modality, enabling the system to make more accurate assessments and suggestions in different situations.

[0025] 3. The present invention adopts an efficient data analysis algorithm based on variational inference, optimizes the calculation efficiency of the psychological state distribution, and realizes the rapid analysis of large-scale user data; compared with the data analysis methods with large computational amounts and low efficiency in the prior art, the solution of the present invention solves the bottleneck of traditional methods in high-dimensional data processing while ensuring the accuracy of the results, and improves the response speed of the system and the user experience.

[0026] 4. The present invention introduces a personalized scale generation and online AI consultation module, provides customized mental health advice to users, and improves the effectiveness of psychological assessment and intervention; different from the unified scale and standardized advice in the prior art, the present invention can generate a scale according to the specific psychological state of users and provide real-time adjustment suggestions, solving the problem of insufficient consideration of individual differences in the existing methods. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] Figure 1 is the system architecture diagram of the present invention;

[0028] Figure 2 is the architecture diagram of the data collection module of the present invention;

[0029] Figure 3 is the architecture diagram of the data analysis module of the present invention;

[0030] Figure 4It is the architecture diagram of the scale production module of the present invention;

[0031] Figure 5 It is the architecture diagram of the result presentation module of the present invention;

[0032] Figure 6 It is the method flow chart of the present invention. Detailed implementation manners

[0033] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0034] Please refer to Figures 1 - 5 , an artificial intelligence-assisted psychological assessment system provided by an embodiment of the present invention includes:

[0035] A data collection module, configured to collect psychological assessment data of a user and transmit the collected data to a data analysis module for processing;

[0036] In this system, the data collection module is one of the core components; its main task is to collect psychological assessment data of users from different sources, including text data, voice data, facial expression data, drawing data, etc.; the collection of each type of data requires different technical means and equipment support, and these data will be further transmitted to the data analysis module for subsequent analysis and processing.

[0037] In this embodiment, the data collection module realizes data collection through the following methods:

[0038] Text data collection: Text data is mainly obtained through the input of users in psychological assessments; for example, when a user fills out a psychological assessment questionnaire, the system will guide the user to gradually input text descriptions; these text data are transmitted to the data analysis module and processed here through natural language processing (NLP) technology; the main task of text analysis is to extract information related to emotional tendency and psychological state from the user's text; through an emotion analysis model, the emotion score of the text data can be converted into a part of the psychological state, reflecting the emotional characteristics of the user.

[0039] In a possible implementation manner, the processing of text data includes a combination of an emotion dictionary method and a deep learning method; the emotion dictionary method is mainly used for preliminary analysis of the emotional direction of the user's text, while the deep learning method can help the system more deeply understand the emotional fluctuations and potential psychological conditions in the context; through such technical processing, the text data can provide valuable input for subsequent psychological state modeling.

[0040] Voice data collection: Voice data is obtained through the user's voice input or voice emotion analysis; for example, the user can interact with the system through voice or answer assessment questions through voice; these voice data are collected by an audio collection device and further transmitted to the data analysis module.

[0041] In one implementation, the analysis of voice data is completed through acoustic feature extraction technology; the Mel spectrum is a commonly used feature for analyzing voice emotions, and the system will extract the Mel spectrum features in the voice and use them for subsequent emotion classification; the voice emotion analysis method is based on a hybrid model of convolutional neural network (CNN) and recurrent neural network (RNN); these models can effectively identify features such as speech rate, pitch change, and pause from the voice signal to help the system judge the user's current mental state.

[0042] Through the following formula, the system can calculate the user's emotion features:

[0043] X s = f(Mel spectrum features, voice pause duration, pitch change);

[0044] where X s represents the emotion features extracted from the voice, f is the emotion classification function, and the Mel spectrum features, voice pause duration, and pitch change are all input voice parameters.

[0045] Facial expression data collection: Facial expression data is collected through computer vision technology; this process is usually achieved through a camera installed on the device, and the system will capture the user's facial expressions in real time and extract expression features through image processing technology.

[0046] In a specific implementation, the system uses a convolutional neural network (CNN) to recognize the user's facial expressions; facial expressions such as smiling, frowning, and opening the mouth will be marked and converted into numerical features to reflect the user's emotional state; these emotion features will be transmitted as input data to the data analysis module to participate in further emotion analysis and mental state assessment.

[0047] In some embodiments, the system uses OpenCV and deep learning models, combined with the changes in facial landmark points, to perform detailed expression dynamic analysis; for example, the system will calculate the expression change through the following formula:

[0048] X e = g(Δ expression points, facial movement trajectory);

[0049] where X e represents the extracted expression features, g is the expression feature extraction function, and Δ expression points and facial movement trajectory are important parameters of the facial expression data.

[0050] Drawing data collection: Drawing data collection usually involves psychological tests such as the Rorschach inkblot test; during this process, the user will freely draw according to the system's instructions, and the system collects drawing data through devices.

[0051] These drawing data will be analyzed through computer vision technology; through image processing algorithms, the system will analyze features such as the shape, symmetry, and line density of the drawing pattern; through these features, the system can infer the user's potential psychological state.

[0052] In an embodiment, the system calculates the psychological characteristics of the drawing data through the following formula:

[0053] X p = h(drawing pattern shape, line density, symmetry);

[0054] where, X p represents the psychological characteristics extracted from the drawing data, h is an image analysis function, and the drawing pattern shape, line density, and symmetry are input features.

[0055] The data analysis module is used to analyze the collected data using machine learning algorithms and provide the analysis results to the scale generation module and the result presentation module;

[0056] In this system, the data analysis module is a key link connecting the data collection module and the scale generation module; its main task is to receive multi-modal data from the data collection module and deeply analyze this data to extract useful information to support the subsequent psychological assessment process; through the data analysis module, the system can generate accurate assessment results for the user's psychological state and provide the necessary input to the scale generation module to ensure the personalization and pertinence of the scale.

[0057] In this embodiment, the data analysis module realizes its functions in the following ways:

[0058] First, the data analysis module will receive and integrate multi-modal data from the data collection module; these data include various forms such as text, voice, facial expressions, drawings, etc.; when each type of data is transmitted to the data analysis module, it has undergone preliminary processing and preprocessing to facilitate accurate analysis in subsequent steps.

[0059] Data preprocessing: The data collected by the data collection module from different sources is usually raw data that has not been processed. In the data analysis module, all data is first preprocessed to ensure the accuracy and effectiveness of the analysis. Text data will undergo sentiment analysis through natural language processing (NLP) techniques, speech data will undergo emotion recognition through acoustic feature extraction, facial expression data will undergo feature extraction through computer vision techniques, and painting data will obtain relevant psychological features through image recognition methods.

[0060] In some embodiments, text data will undergo sentiment classification through a sentiment dictionary or a deep learning model; speech data will extract parameters such as Mel spectrogram features, pitch, and speech rate, and use a support vector machine (SVM) or a convolutional neural network (CNN) to identify the emotional state.

[0061] Data analysis model: The core function of the data analysis module is to analyze the collected data using machine learning algorithms to accurately judge the user's psychological state. Specifically, the system will apply various algorithms, such as decision trees, random forests, support vector machines (SVM), etc., to classify or perform regression prediction on the user's emotions and psychological state.

[0062] In a possible implementation, the data analysis module performs psychological state prediction through the following formula:

[0063] P(Y|X)=∑ Z P(Y|Z)P(Z|X);

[0064] Among them, P(Y|X) represents the posterior probability of the user's psychological state Y given the multi-modal data X; Y is the user's psychological state variable; X = {X t , X s , X e , X p} is multi-modal psychological assessment data; Z is a latent variable representing the individual's potential psychological characteristics; P(Y|Z) represents the conditional probability distribution of the latent variable Z on the psychological state Y; P(Z|X) represents the mapping probability of the multi-modal data X to the latent variable Z.

[0065] In a specific implementation, the data analysis module not only analyzes each type of data independently, but also performs fusion analysis on multi-modal data through a deep learning model (such as a convolutional neural network, a long short-term memory network LSTM) to further improve the accuracy of psychological state prediction.

[0066] Variational Inference and Optimization: To efficiently compute the posterior distribution in a high-dimensional data space, the data analysis module also includes a variational inference unit; Variational Inference (VI) aims to optimize the posterior distribution of a probability model for more efficient computation; Specifically, the system approximates the posterior distribution of latent variables through variational inference methods and optimizes the prediction of mental states.

[0067] The core formula of variational inference is as follows:

[0068]

[0069] Where: P(Y|X) is the posterior probability of the user's mental state; P(Y,Z|X) is the joint probability distribution, representing the joint probability of mental state Y and latent variable Z under the input data X; Q(Z) is the approximate posterior distribution of latent variable Z; D KL (Q(Z)||P(Z|X)) is the Kullback-Leibler divergence, used to measure the difference between the approximate distribution Q(Z) and the true posterior distribution P(Z|X); represents the expectation calculation with respect to the distribution Q(Z).

[0070] Through this method, the system can reduce the computational amount when processing high-dimensional data and obtain effective mental state prediction results.

[0071] Modal Fusion: Since this system involves multi-modal data (text, speech, expression, painting, etc.), the data analysis module also needs to effectively fuse these multi-modal data; To ensure the accuracy of data fusion, the module optimizes each modal data through weighted fusion technology; In this implementation, each data modality has a weight value, reflecting its importance in the overall evaluation.

[0072] In some embodiments, the weights of the modalities are automatically adjusted based on the information entropy calculation method; Information entropy is a measure of the amount of information, used to calculate the contribution degree of different data modalities in the evaluation; According to the calculated weights, the data analysis module performs weighted averaging on the multi-modal data to obtain a comprehensive mental state evaluation result.

[0073] The formula is as follows:

[0074]

[0075] Where: H(X i ) represents the information entropy of modality X i , used to measure the data complexity of this modality; represents the probability distribution of category j in modality X i , satisfying Wi Indicates modality X i The optimized weight, obtained by calculating the normalized information entropy; H(X i ) is the entropy value of the i-th modality; H(X j ) represents the sum of the information entropies of all modalities X j respectively.

[0076] In this way, the system can dynamically adjust the influence of each modality data in the overall analysis to improve the accuracy of psychological assessment.

[0077] The scale generation module is used to automatically generate scales, questions, dimensions, and answers required for psychological assessment, and dynamically adjust the assessment content according to the results of the data analysis module;

[0078] In this system, the scale generation module is mainly responsible for generating personalized psychological assessment scales according to the output results of the data analysis module; generating assessment scales suitable for users through an automated method not only improves the efficiency of the assessment but also ensures the personalization and pertinence of the scale content; the work of the scale generation module is based on the analysis results of the data analysis module on the user's mental state, and at the same time considering the user's historical data, emotional characteristics, and current mental state, automatically adjusts the generated assessment content to ensure that the assessment questions for each user are highly relevant to their current mental state.

[0079] The scale generation module first receives the mental state analysis results from the data analysis module; the data analysis module analyzes the user's multi-modal data (text, voice, facial expressions, paintings, etc.) to generate prediction results about the user's current mental state, such as information on various dimensions such as anxiety, depression, and emotional stability; then, based on these analysis results, the scale generation module will automatically create corresponding assessment questions, dimensions, and answer options to meet the personalized needs of the user.

[0080] The scale generation module dynamically generates assessment questions suitable for the user according to the mental state assessment results provided by the data analysis module; generally, the content of the scale questions will involve multiple dimensions, including but not limited to emotions, behaviors, cognitions, etc.; for some users with obvious emotional fluctuations, the system will generate assessment questions specifically for emotion management; while for users with relatively stable mental states, the generated scale may focus more on the assessment of general mental health.

[0081] In some embodiments, the scale generation module automatically selects appropriate questions through a template matching method; the question selection process takes into account the following factors:

[0082] The user's emotional state. For example, if the user's mood is relatively low, the scale generation module may add some questions about depression and self-awareness;

[0083] Based on the user's historical psychological assessment data and past assessment results, the system can dynamically adjust the scale content to improve relevance and effectiveness;

[0084] For the current assessment purpose, the system can customize different questions according to different mental health goals (such as anxiety assessment, emotional stability assessment, etc.).

[0085] For example, when the user has a high anxiety index, the scale generation module will dynamically adjust the question content according to the following formula:

[0086] P(T|Y)=∑ Z P(T|Z)P(Z|Y);

[0087] Among them, P(T|Y) represents the probability of generating question T under the condition of the given user mental state Y; Z is a latent variable representing potential psychological characteristics; P(T|Z) represents the probability of generating a question under the condition of the latent variable Z; P(Z|Y) represents the probability of the latent variable Z under the condition of the given mental state Y.

[0088] The scale generation module is not only responsible for generating assessment questions but also automatically adjusts the dimensions of the questions; generally, the scale will include multiple dimensions such as emotion, cognition, behavior, etc.; however, these dimensions are not fixed for all users; for different user groups, especially users with different mental states, the scale generation module will dynamically adjust the dimensions according to the output of the data analysis module; for example, for users with obvious anxiety symptoms, the system will automatically increase anxiety-related dimensions (such as anxiety experience, emotional reaction, etc.); for users with relatively stable emotions, the system may reduce such dimensions and focus on self-evaluation and the assessment of routine emotional health.

[0089] In some embodiments, the scale generation module will dynamically adjust the dimensions according to the following formula:

[0090] V i =∑ k P(V i |X k )P(X k |Y);

[0091] Among them, V i represents the weight of the i-th dimension; P(V i |X k ) is the weight of the i-th dimension under the condition of the given data modality X k ; P(X k |Y) is the influence of the k-th data modality on the dimension V i under the condition of the given mental state Y.

[0092] The scale generation module not only generates questions based on the analysis results, but also automatically adjusts the content and scope of the answer options; for example, for self-report type questions (such as "Do I feel anxious?"), the scope of the answer options may be dynamically adjusted according to the user's mental state; the system may provide more detailed options (such as "never", "occasionally", "often", "always", etc.) for users with a lower mood, while for users with a more stable mood, the options may be more simplified.

[0093] In a possible implementation, the scale generation module automatically adjusts the answer options through the following formula:

[0094] A j = f(P(Y|X j ),α j );

[0095] where A j represents the answer options for the j-th question; P(Y|X j ) represents the probability of the user's mental state Y under the j-th data modality X j ; α j is a regulatory parameter for controlling the number of answer options, which is dynamically adjusted based on the user's specific mental state.

[0096] The result presentation module is used to present the analysis results to the user in a graphical way and provide evaluation feedback information to the user interaction module;

[0097] In this system, the result presentation module is mainly responsible for presenting the results of the data analysis module to the user in an intuitive and easy-to-understand way; this module not only displays the current mental state evaluation results of the user, but also helps the user understand the changes in their emotions and the fluctuation trend of their mental state through data visualization technology; through the result presentation module, the user can see the relevant information about their mental health in a friendly interface and can make self-adjustments or further psychological interventions based on this information.

[0098] First, the result presentation module receives the psychological evaluation results from the data analysis module, including the user's emotion score, mental health status, and relevant dimensional analysis results; generally, these data are presented in the form of numerical values, labels, or other quantitative information; then, the system will present them to the user through different visualization technologies based on these results so that the user can more clearly understand their mental state.

[0099] Data visualization: One of the most important functions of the result presentation module is to present the results of the psychological assessment to the user in a visual way; usually, this includes showing the user's mental health status through graphs, charts, curves, radar charts, etc.

[0100] Specifically, the system generates various types of charts based on the user's evaluation results; for example, dimensions such as the user's anxiety level and mood fluctuations can be presented through a radar chart, while the user's mood trend can be shown through a time series chart or a line chart to help the user understand the changes in their mental state over time.

[0101] In some embodiments, the result presentation module calculates the visual representation of the mental state through the following formula:

[0102] Score i =f(P(Y|X i ));

[0103] where Score i represents the score of the i-th mental state dimension; P(Y|X i ) is the probability of the predicted mental state Y given the input data X i ; and f is a function that converts the evaluation result into a visual graph.

[0104] In this way, the system can convert complex mental state data into a clear and understandable graphical representation, enabling the user to obtain key information more quickly.

[0105] Trend analysis: In addition to presenting the current mental state, the result presentation module also provides trend analysis of the changes in the user's mental health; generally, the system shows the changes in the user's mental state over a period of time (such as the past few months or the past year); in this way, the user can not only see their current mental state but also understand the fluctuations in their mental state at different time points.

[0106] For example, the system shows the trend of changes in dimensions such as the user's anxiety and depression through line charts, heat maps, etc.; these charts not only provide the mental health status on a "time axis" but also provide the user with a clear "mental map" to help the user identify their mental fluctuations during a specific period.

[0107] As an option, the system also combines the user's historical data and current data, and through comparative analysis, shows the trend of improvement or deterioration of mental health; for example, if the user has a high anxiety level, the system marks the change curve of the user's anxiety index and shows the changes in the user's mental state at some key time points (such as after treatment or after a life event) through color changes or markings.

[0108] Interactive User Feedback: To enhance the user experience and interactivity, the result presentation module also allows users to provide feedback and interact with the system; users can view detailed analysis results from different dimensions through this module, or gain a deeper understanding of certain specific issues; the system dynamically updates the feedback content according to the user's needs; for example, if a user is interested in the evaluation results of a certain dimension, the user can click on that dimension to view more detailed data analysis results.

[0109] In a specific implementation, the system can provide an interactive interface that allows users to select different analysis perspectives (e.g., view the details of dimensions such as anxiety, mood, stress, etc.), and automatically update the displayed content according to the user's selection.

[0110] Result Output and Report Generation: The result presentation module is also responsible for generating the final evaluation report; the report will include an overall mental health status assessment of the user, a detailed analysis of each mental health dimension, and corresponding chart displays; the system automatically generates a professional and personalized psychological evaluation report based on the user's evaluation results, which is provided to the user for download in the form of a PDF or electronic document, or directly shared with the user through the system platform.

[0111] Specifically, the report generation is based on the following formula:

[0112] R = g(P(Y|X), Visual(X));

[0113] Where R represents the finally generated evaluation report; P(Y|X) is the predicted result of the user's mental state; Visual(X) is the generated visual chart content; g is the function for generating the report, which is responsible for combining the evaluation results and chart content to generate a complete report.

[0114] After the report is generated, users can not only view it through the system, but also export it or share it with professionals such as doctors and psychologists.

[0115] The user interaction module is used to allow users to access analysis results, evaluation content, evaluation feedback information, and data to obtain more detailed psychological evaluation information, and provide interaction data to the data collection module and the online AI consultation module;

[0116] In this system, the user interaction module is the key interface between the entire psychological evaluation system and the user, undertaking the important responsibilities of interacting with the user, providing feedback, and collecting additional user input; it enables users not only to participate in psychological evaluations, but also to obtain personalized suggestions based on the evaluation results and provide feedback on the analysis results provided by the system; through this module, users can deeply understand the evaluation results, improve their mental health, and even adjust subsequent evaluation content or conduct more detailed evaluations.

[0117] First, the user interaction module receives various types of information from the system, including the user's psychological assessment results, emotional assessment reports, mental health trend charts, etc., and feeds this information back to the user; the user can understand their assessment results through the interaction interface and raise further questions or opinions about the suggestions and reports generated by the system; specifically, the user can view multi-dimensional data of the psychological assessment through a graphical interface and interact with the system in various ways such as input boxes, buttons, and voice.

[0118] The user interaction module first displays the user's assessment results; generally, the assessment results will be presented to the user in the form of charts or forms; for example, psychological dimensions such as the user's anxiety, depression, stress, etc. will be displayed in the form of bar charts, line charts, etc.; specifically, the psychological assessment results displayed by the user interaction module are based on the output of the data analysis module and clearly present these results on the graphical interface to help the user intuitively understand their mental state.

[0119] In some embodiments, the system uses the following formula to transform the results:

[0120] R i = g(P(Y|X i ));

[0121] where R i represents the assessment result of the i-th psychological dimension; P(Y|X i ) is the conditional probability of the mental state Y generated based on the input data X i ; g is a function that transforms the assessment result into a graphical feedback.

[0122] In addition, the interaction module also provides the user with in-depth exploration of different dimensions; the user can click on specific assessment results to view detailed analysis explanations and further understand the reasons or influencing factors for their emotional fluctuations.

[0123] Generally, the user interaction module is not limited to one-way result display. It also allows the user to provide feedback on the assessment results; the user can interact with the system through text boxes, voice input, etc.; for the mental state assessment report generated by the system, the user can ask questions or request explanations; for example, if the user does not understand the assessment result of a certain dimension, the system will automatically generate prompts to help the user understand its meaning; at the same time, the user can also ask specific questions, such as "Why do I have a high score in the anxiety dimension?" The system will further provide explanations based on the user's questions.

[0124] As an option, the system can also adjust subsequent assessment content based on user feedback; for example, if the user reports doubts about their assessment results, the interaction module can trigger a re - evaluation process to further optimize the assessment content and scale, or guide the user to conduct a more specific psychological assessment.

[0125] In some embodiments, based on the assessment results provided by the data analysis module, the system will provide personalized suggestions for emotion management and psychological intervention to the user; for example, if the user has a high level of anxiety, the system will push some strategies to cope with anxiety according to the analysis results, such as meditation practice, deep breathing method, etc.; these suggestions will be presented to the user through the interaction module, and the user can choose to accept or ignore them.

[0126] Specifically, the system will recommend suitable adjustment methods for the user according to the user's mental health data and emotional state, combined with a series of mental health resources; for example, in the case of a high level of anxiety, the interaction module may provide suggestions to the user in the following ways:

[0127] S j = h(P(Y|X j ),α j );

[0128] Where S j represents the j - th suggestion recommended to the user; P(Y|X j ) is the evaluation result of the user's emotional state; α j is the adjustment parameter, indicating the intensity or frequency of the intervention measures recommended by the system under a given mental state.

[0129] During the implementation process, the system will adjust the content and intensity of the intervention measures according to factors such as the user's historical data and current emotions to ensure personalized effects.

[0130] The user interaction module also interacts with the scale generation module; generally, the system will automatically adjust the subsequent scale content according to factors such as the user's psychological assessment results and emotional reactions; specifically, if indicators such as the user's anxiety and depression are high, the system can generate more questions related to psychological adjustment according to the user's feedback to further evaluate the user's mental health.

[0131] In some embodiments, the user interaction module can also automatically identify the emotion information input by the user and feedback it to the scale generation module; for example, if the user reports their recent emotional changes through voice or text, the interaction module will process this information through natural language processing (NLP) technology and feedback it to the scale generation module to optimize the subsequent assessment content.

[0132] In the user interaction module, the system not only provides the basic results of psychological assessment but also can offer personalized mental health counseling services according to the user's needs; generally, users can choose different mental health topics, such as anxiety management, stress relief, emotion regulation, etc., and the system will provide relevant counseling services based on these topics.

[0133] Specifically, the user interaction module simulates the role of a psychological counselor through natural language processing technology, conducts conversations with users, and provides personalized suggestions based on the user's mental state; for example, if a user shows relatively high anxiety symptoms, the system can automatically suggest effective psychological treatment methods such as cognitive behavioral therapy (CBT).

[0134] In some embodiments, the personalized feedback can be adjusted based on the real-time assessment results of the data analysis module to optimize the feedback content and suggestions in real time; users can obtain professional counseling and intervention suggestions for different mental states through the interaction module.

[0135] The online AI counseling module is used to provide online artificial intelligence counseling services, offer personalized suggestions according to user needs, and optimize counseling feedback by combining the data of the result presentation module;

[0136] In this system, the online AI counseling module is a key component, aiming to provide personalized suggestions for users based on their psychological assessment results; this module simulates an artificial intelligence psychological counselor, utilizes machine learning and natural language processing technologies to provide users with real-time mental health counseling and support; it can further formulate personalized intervention suggestions for users on the basis of the system's preliminary psychological state assessment, helping users better understand and manage their mental health.

[0137] First, the online AI counseling module obtains the user's psychological assessment results, emotion analysis data, and relevant evaluation information from the data analysis module and the scale generation module; the assessment results of the user's mental state provided by the data analysis module (such as scores of psychological dimensions like anxiety and depression) and the personalized questions generated by the scale generation module are the basis for the online AI counseling module to provide personalized suggestions for users; the system will generate mental health intervention suggestions suitable for users based on this data.

[0138] The core task of the online AI counseling module is to generate personalized suggestions based on the user's psychological assessment results; generally, these suggestions will cover the user's current mental state, emotional fluctuations, and targeted intervention measures; for example, if the system detects that the user has a relatively high anxiety index, the AI counseling module may suggest that the user try relaxation techniques, meditation, deep breathing, etc. to relieve anxiety; if the user has a relatively high depression score, the system may recommend cognitive behavioral therapy (CBT) or emotion regulation techniques.

[0139] In some embodiments, the online AI consultation module generates personalized suggestions according to the psychological state evaluation result of the user by applying the following formula:

[0140] S i = f(P(Y|X i ), α i );

[0141] where S i represents the intervention suggestion provided for the i-th mental health dimension; P(Y|X i ) represents the probability of the predicted mental state Y given the input data X i ; and α i is a regulation parameter representing the intensity or frequency of the suggestion given a certain mental state.

[0142] Specifically, the system generates personalized intervention strategies according to different mental state categories (such as anxiety, depression, etc.) so that users can choose appropriate mental health intervention methods according to their own needs.

[0143] The online AI consultation module interacts with users through natural language processing (NLP) technology; users can communicate with the system through text input or voice input, asking questions about mental health, emotion management, etc.; the system will perform emotion recognition and question understanding based on the user's input and then generate appropriate responses.

[0144] In some embodiments, the AI consultation module will simulate the conversation mode of a psychological counselor and communicate deeply with the user through sentiment analysis and context understanding; the system will parse the information input by the user, such as "I've been feeling very anxious lately. What should I do?" Then, based on the analysis results, the AI consultation module will generate feedback similar to the following: "Anxiety is a common emotional reaction. It can be helped by practicing deep breathing. You can try doing five minutes of meditation every day."

[0145] Specifically, the AI consultation module processes the user input through the following formula:

[0146] U f = g(P(Y|X u ), h(T x ));

[0147] where U f is the user feedback generated by the AI consultation module; P(Y|X u ) is the probability of the predicted mental state Y given the user input X u ; T x is the user input text or voice, and h(T x ) is to convert the user input into an emotion identifier that the system can understand.

[0148] This natural language processing-based interaction method enables the AI consultation module to provide a more user-friendly consultation experience.

[0149] The online AI consultation module not only provides preliminary mental health advice but also dynamically adjusts intervention measures based on the user's feedback and subsequent data. Under normal circumstances, if the user expresses dissatisfaction with a certain intervention advice or indicates that the effect is not obvious during the interaction, the system will regenerate personalized advice according to the user's feedback.

[0150] For example, if the user's feedback shows that their mood has improved after receiving anxiety relief techniques, the AI consultation module may adjust the intensity and content of subsequent advice based on the new mood assessment data. If the user feedbacks that some advice has poor effects, the system can also adjust the intervention method according to the feedback.

[0151] In some embodiments, the system adjusts the advice content based on the following formula:

[0152]

[0153] where is the adjusted intervention advice; P(Y|X i ) represents the probability of the current user's mental state; β i is a parameter controlling the intervention intensity; is the previously provided intervention advice.

[0154] Through continuous feedback and updates, the system ensures that more effective mental health intervention advice is provided to users, enhancing the users' positive experience.

[0155] The online AI consultation module can also provide users with relevant mental health resources, such as self-help books, video courses, meditation practices, etc. Specifically, the system recommends the most suitable mental health resources based on the user's psychological assessment data. For example, if the user has a high depression score, the system may recommend a video tutorial on self-management of depression. If the user's anxiety symptoms are more obvious, the system may recommend books or online courses on anxiety management techniques.

[0156] These resources can be recommended to users through the online AI consultation module to further help users regulate their emotions and improve their mental health level.

[0157] The modality selection and decision optimization module is used to calculate the contribution degrees of different data modalities, optimize the psychological assessment decision-making scheme, and provide the optimal modality combination information to the data analysis module;

[0158] In this system, the main task of the modality selection and decision optimization module is to optimize the impact of different data modalities on the final decision result during the psychological assessment based on the contribution degree of multi-modal data; this module evaluates the importance of each data modality in the psychological assessment through information entropy calculation and dynamically adjusts the weights of each modality according to this information; in this way, the system ensures that the final assessment result can fully consider the effectiveness of each data modality, thereby improving the accuracy and personalization of the psychological assessment.

[0159] First, the modality selection and decision optimization module receives multi-modal data from the data collection module, including text, voice, facial expressions, and drawing data; each data modality has a different impact on the final psychological assessment result, and the system needs to evaluate and allocate appropriate weights to each modality to maximize the accuracy of the decision; this module determines the information volume of each modality through information entropy calculation and uses the information volume as the basis for weight allocation.

[0160] Information entropy calculation: Information entropy is a standard used to measure the uncertainty or information volume of information; in psychological assessment, data of different modalities may have different information volumes, so it is necessary to quantify the effectiveness of each data modality through information entropy; by calculating the information entropy of each modality, the system can determine which modalities are the most important in the decision-making process and allocate higher weights to them.

[0161] In this embodiment, the calculation formula of information entropy is as follows:

[0162]

[0163] where H(X i ) represents the information entropy of modality X i , which is used to measure the data complexity of this modality; represents the probability distribution of category j in modality X i , satisfying

[0164] This formula is based on the probability distribution of each category in modality X i and measures the uncertainty of the modality. The higher the entropy value, the greater the information volume provided by this modality.

[0165] Modality weight allocation: By calculating the information entropy, the system can allocate weights to each data modality; generally speaking, modalities with larger information entropy will be given higher weights because they can provide more effective information; the purpose of weight allocation is to ensure that each data modality in the assessment result can play its maximum role in the decision-making process.

[0166] In this embodiment, the modality weights are allocated through a normalization method, and the formula is as follows:

[0167]

[0168] Where: W i represents the optimized weight of modality X i , which is obtained by calculating the normalized information entropy; H(X i ) is the entropy value of the i-th modality; H(X j ) represents the sum of the information entropies of all modalities X j respectively;

[0169] Through this formula, the system dynamically allocates weights according to the information entropy of each modality, ensuring that the most effective information is given priority consideration.

[0170] Dynamic weight adjustment: The modality selection and decision optimization module not only assigns weights in the initial stage, but also can dynamically adjust weights according to real-time data feedback; for example, in some cases, a certain data modality (such as speech) may provide more effective information in sentiment analysis, so the system will automatically increase the weight of this modality; in contrast, other modalities (such as text) may contribute less information in the current evaluation, and the system may accordingly reduce the weight of this modality.

[0171] In this dynamic adjustment process, the system optimizes the weights of each modality through an adaptive algorithm; for example, based on the aforementioned modality entropy calculation results, the weight adjustment can be carried out according to the following formula:

[0172]

[0173] Where, represents the updated weight of the i-th modality; W i represents the current weight of the i-th modality; γ represents the learning rate or step size parameter; H(X i ) represents the data entropy of the i-th modality; ∑ j H(X j ) represents the sum of the entropy values of all modalities; represents the normalized entropy value of the i-th modality;

[0174] If is equal to the current weight W i , it means that the weight has reached the ideal state, so the difference in the second formula is 0, and thus the weight will not be changed; this dynamic adjustment can ensure that each evaluation can flexibly adapt to the different needs and psychological states of users.

[0175] Comprehensive Decision Optimization: The ultimate goal of the modality selection and decision optimization module is to obtain a comprehensive mental health assessment result by weighted fusion of the results of each modality; the system will perform weighted averaging on the data of each modality according to the modality weights calculated above to form the final assessment result; through this weighted fusion, the system can comprehensively consider the data of different modalities, so as to provide more accurate and personalized psychological assessment results for users.

[0176] In this embodiment, the final comprehensive assessment result RRR is calculated by the following formula:

[0177] R = ∑ i W i ·X i ;

[0178] where R is the comprehensive psychological assessment result; W i is the weight of the i-th modality; X i is the assessment data of the i-th modality.

[0179] This weighted fusion method ensures that each modality can fully play its role in the decision-making process, thus improving the accuracy of the psychological assessment result.

[0180] Please refer to Figure 6 , an artificial intelligence-assisted psychological assessment method, including the following steps:

[0181] S1. Data collection: In the first stage of the system, the data collection module is responsible for collecting multi-modal data of users, including text, voice, facial expressions, and drawing data; these data can comprehensively reflect the psychological state of users and provide a rich information source; the collected data will go through necessary preprocessing steps, including noise removal, normalization, etc., to ensure the quality and accuracy of the data; the modal data after data collection will provide key inputs for subsequent psychological state modeling and data analysis;

[0182] S2. Psychological state modeling: In the second stage, the system constructs a hierarchical representation of the user's psychological state by adopting a probabilistic graphical model (e.g., Bayesian network or hidden Markov model); this model can extract potential psychological features from multi-modal data and map these features to the hierarchical structure of the psychological state; in this way, the system can transform complex psychological states into easy-to-process mathematical models, providing a basis for subsequent analysis and prediction;

[0183] S3. Data Analysis: The third stage is to analyze the user's psychological data. By applying machine learning algorithms, especially technologies such as deep learning and support vector machine (SVM), the system can mine valuable psychological information from the user's multimodal data. At the same time, variational inference technology is used to optimize the computational efficiency and avoid computational bottlenecks in the high-dimensional data space. Finally, the system will output a distribution of the user's psychological state to help accurately evaluate the user's emotions, anxiety, depression and other psychological dimensions.

[0184] S4. Scale Generation: After the analysis stage, the scale generation module dynamically generates a personalized psychological assessment scale based on the analysis results and the user's individual characteristics. The scale content will be tailored to the user's psychological state and needs to ensure that the assessment content is highly relevant and targeted. For example, if the user has a high anxiety score, the scale may include more anxiety-related questions and adapt to the user's psychological state by adjusting the question order or content.

[0185] S5. Result Presentation: The result presentation module is responsible for presenting the analysis results of the psychological assessment to the user in a data visualization way. Through charts, radar charts, trend charts, etc., the user can intuitively see their psychological assessment results. In addition, the system will combine the user's historical data to show the trend of their psychological changes, helping the user understand their emotional fluctuations and mental health changes. These visualization results enable the user to more clearly see their changes in various psychological dimensions, so as to make more scientific self-adjustments.

[0186] S6. Modal Selection and Optimization: The system optimizes the weights of the modalities by calculating the information entropy according to the contribution degrees of the individual data modalities. Each modality has its unique contribution in the decision-making process. Some modalities (such as voice or expression) may provide more valuable information in certain situations. By calculating the data contribution degrees of the multimodal data, the system can automatically optimize the weights of the modalities to ensure that the most valuable modality can be given priority in the decision-making process. This optimization process improves the personalization ability and decision-making accuracy of the system.

[0187] S7. Online Consultation: In the final stage, the online AI consultation module provides personalized mental health advice based on the analysis results of the user's psychological state. The system will simulate the role of a psychological counselor to provide the user with emotion regulation skills, cognitive behavioral therapy (CBT) advice, relaxation exercises, etc., to help the user further understand their psychological state and propose corresponding adjustment strategies. Through real-time online AI consultation, the user can obtain advice and guidance on mental health at any time, enhancing their emotion management and psychological regulation abilities.

[0188] Although embodiments of the present invention have been shown and described, it will be understood by those of ordinary skill in the art that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the present invention, and the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. An artificial intelligence-assisted psychological assessment system, characterized in that, It includes: A data collection module, which is used to collect the psychological assessment data of users and transmit the collected data to the data analysis module for processing; A data analysis module, which is used to analyze the collected data using machine learning algorithms and provide the analysis results to the scale generation module and the result presentation module; A scale generation module, which is used to automatically generate the scales, questions, dimensions and answers required for psychological assessment, and dynamically adjust the assessment content according to the results of the data analysis module; A result presentation module, which is used to present the analysis results to users in a graphical way and provide assessment feedback information to the user interaction module; A user interaction module, which is used to allow users to access the analysis results, assessment content, assessment feedback information and data to obtain more detailed psychological assessment information, and provide interaction data to the data collection module and the online AI consultation module; An online AI consultation module, which is used to provide online artificial intelligence consultation services, provide personalized suggestions according to user needs, and optimize the consultation feedback in combination with the data of the result presentation module; A modality selection and decision optimization module, which is used to calculate the contribution degrees of different data modalities, optimize the psychological assessment decision-making scheme, and provide the optimal modality combination information to the data analysis module.

2. The artificial intelligence-assisted psychological assessment system according to claim 1, wherein, The data collection module includes: A text acquisition unit, which is used to obtain the text information input by users and perform natural language processing to extract emotional features; A voice acquisition unit, which is used to extract the pitch, speech rate and pause pattern features of user voices; An expression acquisition unit, which is used to identify the changes in the facial expressions of users through computer vision technology and extract expression feature data; A painting acquisition unit, which is used to analyze the psychological patterns drawn by users and extract painting feature data.

3. An artificial intelligence-assisted psychological assessment system according to claim 1, wherein, The data analysis module includes: A psychological state modeling unit, which is used to construct a hierarchical representation of the psychological state based on a probabilistic graphical model; A machine learning analysis unit, which is used to perform classification and regression analysis on user data using neural network, support vector machine and random forest machine learning algorithms; A variational inference unit, which is used to perform variational inference optimization on psychological data to improve the computational efficiency of high-dimensional data modeling.

4. An artificial intelligence-assisted psychological assessment system according to claim 1, characterized in that, The scale generation module includes: A question bank management unit, which is used to store and manage the questions for psychological assessment and can dynamically adjust the assessment content based on user data; A personalized scale generation unit, which is used to generate assessment scales suitable for the personalized needs of users according to the user's historical assessment results; A dynamic adaptation unit, which is used to optimize the assessment questions based on reinforcement learning algorithms to make the assessment more in line with the user's current psychological state.

5. An artificial intelligence-assisted psychological assessment system according to claim 1, characterized in that, The result presentation module includes: A data visualization unit, which is used to present the psychological assessment results in the form of charts, curves and heat maps; A trend analysis unit, which is used to compare the psychological assessment data of users at different times and present the psychological change trend; A report generation unit, which is used to generate an assessment report according to the analysis results and support report export and online sharing.

6. An artificial intelligence-assisted psychological assessment system according to claim 1, characterized in that, The psychological state modeling of the data analysis module adopts a latent variable probabilistic graphical model and is defined by the following formula: P(Y|X) = ∑ Z P(Y|Z)P(Z|X); Among them, P(Y|X) represents the posterior probability of the user's mental state Y given the multimodal data X; Y is the user's mental state variable; X = {X t , X s , X e , X p} is multimodal psychological assessment data; Z is a latent variable representing the individual's potential psychological characteristics; P(Y|Z) represents the conditional probability distribution of the latent variable Z on the mental state Y; P(Z|X) represents the mapping probability of the multimodal data X to the latent variable Z.

7. An artificial intelligence-assisted psychological assessment system according to claim 1, wherein, The data analysis module uses variational inference to optimize the calculation of the posterior probability of the psychological state, and its formula is defined as follows: Where: P(Y|X) is the posterior probability of the user's mental state; P(Y, Z|X) is the joint probability distribution, representing the joint probability of the mental state Y and the latent variable Z under the input data X; Q(Z) is the approximate posterior distribution of the latent variable Z; D KL (Q(Z)||P(Z|X)) is the Kullback-Leibler divergence, which is used to measure the difference between the approximate distribution Q(Z) and the true posterior distribution P(Z|X); denotes the expectation calculation with respect to the distribution Q(Z).

8. An artificial intelligence-assisted psychological assessment system according to claim 1, wherein, The psychological state prediction of the data analysis module adopts a dynamic Bayesian network, and the calculation formula is as follows: Where: P(Y t |Y t-1 ,X t ) means at time t, given the psychological state Y at the previous moment t-1 and the current multimodal psychological assessment data X t After that, the user's psychological state Y t The probability of Z t is a hidden variable at time t, indicating the user's potential psychological characteristic state; P(Y t |Z t ) represents the latent variable Z t For psychological state Y t The probability of influence; P(Z t |Y t-1 ,X t ) represents the latent variable Z t From the previous moment's psychological state Y t-1 and the current input data X t Jointly determined conditional probability distribution.

9. An artificial intelligence-assisted psychological assessment system according to claim 1, wherein, The mode selection and decision optimization module calculates the contribution degree of each mode data using information entropy and optimizes the mode weights according to the information entropy. The calculation method is as follows: where: H(X i ) represents the information entropy of modality X i , which is used to measure the data complexity of this modality; represents the probability distribution of class j in modality X i , satisfying W i represents the optimized weight of modality X i , which is calculated by normalizing the information entropy; H(X i ) is the entropy value of the i-th modality; H(X j ) represents the sum of the information entropies of all modalities X j respectively.

10. An artificial intelligence-assisted psychological assessment method, applied to an artificial intelligence-assisted psychological assessment system according to any one of claims 1-9, characterized in that, It includes the following steps: S1. Data collection: Collect the text, voice, expression, and drawing data of the user, and preprocess the multi-modal data to provide preparation for subsequent psychological state modeling and data analysis; S2. Psychological state modeling: Use a probabilistic graphical model to construct a hierarchical representation of the psychological state, and map the collected multi-modal data to the potential psychological feature space; S3. Data analysis: Use machine learning algorithms to analyze the user's psychological data, and optimize the calculation efficiency through variational inference to obtain the user's psychological state distribution; S4. Scale generation: Dynamically generate a personalized psychological assessment scale according to the analysis results and the user's individual characteristics to meet the user's needs; S5. Result presentation: Display the analysis results of the psychological assessment through data visualization methods, and provide an analysis of the psychological change trend in combination with the user's historical data; S6. Mode selection and optimization: Calculate the contribution degree of multi-modal data based on information entropy and optimize the mode weights to ensure the accuracy and personalization of the decision-making process; S7. Online consultation: Based on the psychological state analysis results, provide personalized mental health advice for the user through the online AI consultation module to help the user further understand their psychological condition and make corresponding adjustments.

Images