CN-122000059-A - Mental health assessment method

CN122000059ACN 122000059 ACN122000059 ACN 122000059ACN-122000059-A

Abstract

The application discloses a psychological health assessment method, and belongs to the technical field of artificial intelligence and psychological health assessment. The method comprises the steps of obtaining campus behavior time sequence data of a target student, wherein the campus behavior time sequence data comprise a plurality of pieces of campus behavior data which are ordered according to time stamps, each piece of campus behavior data is used for describing behavior events triggered by the target student at corresponding time stamps, determining P initial behavior characteristics and Q initial space-time characterization characteristics based on the campus behavior time sequence data, constructing a behavior abnormal composition of the target student, determining the target behavior characteristics based on the behavior abnormal composition, carrying out characteristic fusion on the target behavior characteristics and the Q initial space-time characterization characteristics to obtain global behavior characteristics of the target student, and evaluating the psychological health state of the target student based on the global behavior characteristics to obtain a psychological health evaluation result. Therefore, the application can effectively evaluate the psychological health state of the student by analyzing the campus behavior time sequence data of the student.

Inventors

LI SHANXI
YANG MINQIANG
TAO YONGFENG
HE CHEN

Assignees

兰州大学

Dates

Publication Date: 20260508
Application Date: 20260126

Claims (10)

1. A method of mental health assessment, the method comprising: Acquiring campus behavior time sequence data of a target student, wherein the campus behavior time sequence data comprises a plurality of pieces of campus behavior data sequenced by time stamps, and each piece of campus behavior data is used for describing behavior events triggered by the target student at corresponding time stamps; Based on the campus behavior time sequence data, P initial behavior characteristics and Q initial space-time characterization characteristics are determined, wherein the P initial behavior characteristics are used for describing the campus network access behavior of the target student, the Q initial space-time characterization characteristics are used for describing the change condition of the campus network access behavior of the target student along with time, and P and Q are integers which are larger than or equal to 1; constructing a behavioral profile of the target student, and determining target behavioral characteristics based on the behavioral profile, wherein the behavioral profile is used for describing association relations among the P initial behavioral characteristics; Performing feature fusion on the target behavior feature and the Q initial space-time characterization features to obtain global behavior features of the target students; and evaluating the psychological health state of the target student based on the global behavioral characteristics to obtain a psychological health evaluation result.
2. The method of claim 1, wherein the P initial behavioral characteristics include a first class of characteristics for describing liveness of campus behavior of the target student, a second class of characteristics for describing behavior patterns of the target student, and a third class of characteristics for describing regularity of work and rest of the target student; the determining P initial behavioral characteristics based on the campus behavioral time sequence data includes: And extracting behavior characteristics from the campus behavior time sequence data to obtain the P initial behavior characteristics.
3. The method of claim 1, wherein the determining Q initial spatiotemporal characterization features based on the campus behavioral time series data comprises: Determining a plurality of discrete event sequences based on the campus behavior time sequence data, wherein each discrete event sequence describes the occurrence frequency of one type of campus behavior event in an evaluation period, and the occurrence frequencies of events in at least two discrete event sequences in the plurality of discrete event sequences are different; performing time sequence alignment processing on the discrete event sequences to obtain a plurality of aligned event sequences, wherein the sequence length of each aligned event sequence is the same, and the sequence length refers to the number of campus behavior events included in the corresponding event sequence; and extracting space-time characteristics from the plurality of aligned event sequences to obtain the Q initial space-time characterization characteristics.
4. The method according to claim 1 or 2, wherein the behavioral profile includes P nodes corresponding to the P initial behavioral characteristics one by one, a connecting edge exists between a first node and a second node, the first node is any one of the P nodes, and the second node is any one of the P nodes, which is different in type from the initial behavioral characteristic corresponding to the first node.
5. The method of claim 4, wherein the determining target behavioral characteristics based on the behavioral profile comprises: Determining weight values of all connecting edges in the behavior abnormal graph, wherein the weight values are used for describing association strength of initial behavior characteristics corresponding to two nodes connected by the corresponding connecting edges in the campus behavior time sequence data; Calculating the aggregate behavior characteristics corresponding to the P nodes respectively based on the weight values of the connecting edges in the behavior iso-graph, wherein the aggregate behavior characteristics are determined based on the initial behavior characteristics of the nodes connected with the corresponding nodes and the weight values of the connecting edges; And determining the target behavior characteristic based on the aggregation behavior characteristics respectively corresponding to the P nodes.
6. The method of claim 5, wherein calculating the aggregate behavior feature corresponding to each of the P nodes based on the weight values of each connection edge in the behavior profile comprises: constructing a plurality of edge weight matrixes based on the weight values of the connecting edges in the behavioral profile, wherein each edge weight matrix is used for describing the weight value of one type of connecting edge; Calculating an aggregate behavior characteristic corresponding to a target node based on an initial behavior characteristic corresponding to the target node, an initial behavior characteristic corresponding to a neighbor node and a neighbor edge weight matrix, wherein the target node is any node in the behavior iso-graph, the neighbor node comprises at least one node connected with the target node in the behavior iso-graph, and the neighbor edge weight matrix comprises at least one adjacent matrix of the edge weight matrices, wherein the at least one adjacent matrix describes weight values of connecting edges between the neighbor node and the target node.
7. The method according to any one of claims 1-6, wherein the feature fusion of the target behavioral feature and the Q initial space-time characterization features to obtain global behavioral features of the target student comprises: Determining an affinity matrix based on the target behavior feature and the Q initial space-time characterization features, the affinity matrix describing semantic relatedness between the target behavior feature and the Q initial space-time characterization features; Calibrating the Q initial space-time characterization features based on the affinity matrix to obtain Q target space-time characterization features; And carrying out feature fusion on the target behavior features and the Q target space-time characterization features to obtain the global behavior features.
8. The method of any one of claims 1-6, wherein the mental health assessment results comprise a mental health score indicating a risk of the target student having a mental problem; The step of evaluating the psychological health state of the target student based on the global behavioral characteristics to obtain a psychological health evaluation result comprises the following steps: And inputting the global behavior characteristics into an evaluation model to obtain the psychological health score output by the evaluation model.
9. The method of claim 8, wherein the method further comprises: determining scoring contribution degrees respectively corresponding to various campus behaviors based on the mental health scores, wherein the scoring contribution degrees are contribution degrees of corresponding campus behaviors to the mental health scores, and the campus behaviors are determined based on the campus behavior time sequence data; The mental health evaluation result further comprises scoring contribution degrees corresponding to the campus behaviors respectively.
10. The method of any one of claims 1-6, wherein the acquiring campus behavioral time series data of the target student over the assessment period comprises: acquiring original behavior time sequence data of the target students in an evaluation period; And desensitizing the data of the identity information of the target student in the original behavior time sequence data to obtain the campus behavior time sequence data.

Description

Mental health assessment method Technical Field The application relates to the technical field of artificial intelligence and mental health evaluation, in particular to a mental health evaluation method. Background Psychological health problems of students are increasingly receiving social attention, and particularly in college student groups, psychological health problems have a non-negligible effect on academic progress, daily life and social interactions. At present, common psychological health assessment and early warning means mainly depend on students to report themselves autonomously or are carried out by professionals in combination with physiological monitoring, clinical interviews and other means, but the methods have certain limitations in practical application. The self-reporting result is easily affected by subjective factors of students, for example, partial students may feel about privacy, worry about being discriminated and the like, and are unwilling to faithfully state the self-reporting result, and at the same time, the cognition deviation and recall error also influence the accuracy of the reporting result, so that the objectivity of the evaluation result is difficult to ensure. The manual evaluation mode requires a great deal of professional human resources, including professional psychological consultants, psychiatric doctors and the like, has relatively high evaluation cost, and the manual evaluation often cannot respond timely, usually requires a certain time arrangement and flow, and cannot realize normalization, instantaneity evaluation and intervention of the psychological health condition of the college students, and can not timely detect and take corresponding measures at the early stage of the occurrence of the problems. Along with the vigorous development of artificial intelligence technology, how to utilize college student behavior data in school to comprehensively and objectively evaluate the psychological health status of students becomes a technical problem to be solved. Disclosure of Invention The application provides a psychological health assessment method which can be used for objectively and effectively assessing the psychological health state of students by analyzing campus behavior time sequence data of the students. The technical proposal is as follows: On the one hand, the psychological health assessment method comprises the steps of obtaining campus behavior time sequence data of a target student, wherein the campus behavior time sequence data comprise a plurality of pieces of campus behavior data which are ordered according to time stamps, each piece of campus behavior data is used for describing behavior events triggered by the target student at the corresponding time stamp, determining P initial behavior characteristics and Q initial space-time characterization characteristics based on the campus behavior time sequence data, wherein the P initial behavior characteristics are used for describing campus network access behaviors of the target student, the Q initial space-time characterization characteristics are used for describing the change condition of the campus network access behaviors of the target student along with time, P and Q are integers which are larger than or equal to 1, constructing behavioral iso-patterns of the target student, determining target behavior characteristics based on the behavioral iso-patterns, the behavioral iso-patterns are used for describing association relations among the P initial behavior characteristics, carrying out characteristic fusion on the target behavior characteristics and the Q initial space-time characterization characteristics, and obtaining global behavior characteristics of the target student, and assessing psychological health states of the target student based on the global behavior characteristics, and obtaining psychological health assessment results. In another aspect, there is provided a mental health assessment apparatus, the apparatus comprising: The data acquisition module is used for acquiring campus behavior time sequence data of the target students, wherein the campus behavior time sequence data comprise a plurality of pieces of campus behavior data which are ordered according to time stamps, and each piece of campus behavior data is used for describing behavior events triggered by the target students at the corresponding time stamps; The first feature extraction module is used for determining P initial behavior features and Q initial space-time characterization features based on campus behavior time sequence data, wherein the P initial behavior features are used for describing campus network access behaviors of target students, the Q initial space-time characterization features are used for describing the change condition of the campus network access behaviors of the target students along with time, and P and Q are integers which are larger than or equal to 1; The second feature extraction module is