VR-based mental health assessment methods

Abstract

This invention relates to the field of psychological assessment technology, specifically a VR-based mental health assessment method, comprising the following steps: analyzing event types, interaction states, and priorities based on virtual scene event nodes; selecting suitable nodes; adjusting the acquisition window; judging changes in EEG amplitude; identifying segmented time points; combining extreme values ​​and response intervals; matching high-risk benchmarks; and classifying psychological risk indicators. In this invention, user interaction behavior and virtual situation events are dynamically mapped to achieve closed-loop management of the entire data flow. The acquisition window is adjusted in conjunction with multi-source parameters, signal segments are defined in real-time based on actual EEG fluctuations and scene cognitive states, EEG electrode values ​​and behavioral response timing are extracted synchronously, multi-dimensional features are coupled and analyzed, and the dynamic grouping of psychological risks is based on the joint setting of multiple indicator intervals. Data is guaranteed to be timely in high-frequency interaction scenarios, and the assessment results can reflect the continuous evolution of the user's psychological state in complex environments.

Description

Technical Field

[0001] This invention relates to the field of psychological assessment technology, and more particularly to a VR-based method for psychological health assessment. Background Technology

[0002] Psychological assessment techniques mainly include technologies related to the collection and analysis of an individual's mental health status. These techniques cover psychological state detection, collection of psychophysiological signals, psychological scale assessment, and psychological state discrimination based on physiological and behavioral data. Traditional mental health assessment methods involve collecting subjective information from individuals by filling out psychological scales or conducting interviews, or obtaining basic physiological signals through simple physiological indicator collection instruments. These signals are then analyzed in conjunction with standard psychological assessment tools. Typically, paper-and-pencil questionnaires, simple EEG acquisition devices, and behavioral observation records are used to complete the data collection and basic analysis of the assessed individual's mental state.

[0003] Traditional psychological assessment methods primarily rely on static data collection or subjective self-reporting. The on-site environment is limited, interactive feedback is delayed, and there is a lack of adaptability to the combined changes of multiple types of parameters. Physiological and psychological states are collected in isolation, making it difficult to effectively reflect the dynamic relationships during the interaction process. Group judgments are based on a single point in time or item-by-item evaluations, which are difficult to adapt to the complexity of tasks and the variability of situations. Risk identification tends to form static labels, information is not updated in a timely manner, and there are gaps between different assessment dimensions. It is difficult to reveal the psychological risks under complex emotional reactions and changing scenarios in a timely and layered manner. Summary of the Invention

[0004] The purpose of this invention is to address the shortcomings of existing technologies by proposing a VR-based mental health assessment method.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: a VR-based mental health assessment method, comprising the following steps:

[0006] S1: Based on the contextual event triggering nodes in the virtual scene, analyze whether the event type is consistent with the preset type, judge the user interaction status and specification requirements, and combine the event priority sorting to compare each node item by item, filter the nodes that meet the adaptation standard, and obtain the event adaptation judgment sequence.

[0007] S2: Based on the event adaptation judgment sequence, analyze the visual complexity of the event, and calculate the start and end intervals of the acquisition window of each node by comparing the correlation between the click frequency of interactive behavior and the event priority, so as to obtain the acquisition window adjustment segment.

[0008] S3: Based on the acquisition window adjustment segment, determine the instantaneous amplitude change of EEG in each channel, compare the average amplitude of the current interval with that of the previous interval, analyze the EEG fluctuation trend of the difference interval, identify the segmentation time, and obtain the EEG segmentation feature set.

[0009] S4: Based on the EEG segment feature set, determine the EEG signal interval for each segment, analyze the cognitive task nodes, construct an extreme value sequence by the maximum and minimum EEG amplitudes, compare it with the user behavior response time, calculate the correlation of the corresponding intervals, and obtain the extreme value response feature interval.

[0010] The present invention is improved in that the event adaptation judgment sequence includes node validity, node influence degree, and node hierarchical information; the acquisition window adjustment segment includes window division method, window identifier, and window associated events; the EEG segmentation feature set includes segmentation type, segmentation stability, and segmentation range index; and the extreme value response feature interval includes response coupling coefficient, extreme value interval distribution, and behavioral synchronization characteristics.

[0011] The present invention is improved in that the step of obtaining the event adaptation determination sequence is specifically as follows:

[0012] S111: Based on the contextual event triggering nodes in the virtual scene, analyze the types and preset standards of contextual events in the virtual scene, compare the matching relationship between each event type and the standard type, determine whether the user's behavioral parameters in the interaction process are consistent with the norm, filter out nodes that meet the standards, and obtain the node adaptability comparison sequence.

[0013] S112: Based on the node adaptability comparison sequence and combined with event priority, calculate the relative position difference of each node in the sorting process, filter the node content of key concern, and obtain the node filtering and sorting sequence.

[0014] S113: Based on the node filtering and sorting sequence, compare the effectiveness, influence degree and hierarchical attributes of each node, obtain the node adaptation differences, and determine the adaptation status of each node with the benchmark to obtain the event adaptation judgment sequence.

[0015] The present invention is improved in that the step of obtaining the acquisition window adjustment segment is specifically as follows:

[0016] S211: Based on the event adaptation judgment sequence, for the visual performance of each node, the color, shape, dynamic elements and visual hierarchy features in the scene are statistically analyzed, and combined with the degree of node influence and hierarchical relationship, the visual load of the node in the scene is judged to obtain the event visual load index.

[0017] S212: Based on the event visual load index, analyze the relationship between the click situation of interactive behavior in the event node and the node criticality, calculate the response distribution of interactive operation in different nodes, and filter the data segments with prominent correlation to obtain the interactive coupling distribution sequence.

[0018] S213: Based on the interactive coupling distribution sequence, combined with the visual load and interactive distribution changes of each node, determine the range in which the acquisition window of each event node needs to be extended or reduced, optimize the start and end limits of the acquisition time period, and obtain the acquisition window adjustment segment.

[0019] The present invention is improved in that the steps for obtaining the EEG segment feature set are specifically as follows:

[0020] S311: Based on the acquisition window adjustment segment, analyze the continuous change of instantaneous amplitude of multi-channel EEG, compare the amplitude changes of each channel in different segments, and determine the fluctuation between segments by constructing an amplitude change sequence between multi-channel data to obtain amplitude change data;

[0021] S312: Based on the amplitude change data, analyze the differences in continuously changing segments, obtain the change fluctuation sequence, determine the range of segments in the multi-channel data where abrupt changes occur, and obtain the amplitude abrupt change segments;

[0022] S313: Based on the amplitude mutation segment, filter its multi-channel data, optimize the amplitude extreme features within the segment, and statistically analyze the segmentation type and amplitude stability within each segment to obtain the EEG segmentation feature set.

[0023] The present invention is improved in that the step of obtaining the extreme value response characteristic interval is specifically as follows:

[0024] S411: Based on the EEG segment feature set, analyze the amplitude characteristics and fluctuation performance of each segment, compare the extreme amplitude with the minimum amplitude for each cognitive branch task stage, calculate the amplitude difference, and screen related segments to obtain the extreme amplitude difference segments.

[0025] S412: Based on the extreme value amplitude difference range, compare it with the time distribution of user behavior response time, determine the time sequence of extreme value fluctuation and response behavior through the correspondence of task stages, identify combinations with close correspondence, and obtain extreme value response coupling combination;

[0026] S413: Based on the extreme response coupling combination, analyze the coupling performance of amplitude change and behavior interval between each group, screen the key correlation segments, and obtain the extreme response characteristic interval.

[0027] The present invention is improved in that the steps further include:

[0028] S5: Based on the extreme value response characteristic interval, analyze the joint state of the extreme value sequence interval length and the behavioral response interval, compare the matching with the high-risk segment benchmark and the delay time benchmark, screen the high-risk group, summarize the rest as medium and low risk groups, and obtain the psychological risk classification index.

[0029] The psychological risk classification indicators include risk level, group label, and risk interval boundary.

[0030] The present invention is improved in that the steps for obtaining the psychological risk classification indicators are as follows:

[0031] S511: Based on the extreme value response characteristic interval, analyze the extreme value sequence duration interval and the user's behavior response time interval, compare the changing trends of the two at each corresponding node, sort out the differences in performance under all nodes, identify data nodes with consistent difference characteristics, and obtain extreme value behavior difference nodes.

[0032] S512: Based on the extreme value behavior difference nodes, determine the matching status of each node with the high-risk continuous segment benchmark and the delay time benchmark in the feature distribution, identify nodes that meet the high-risk judgment criteria, and obtain the high-risk distribution interval.

[0033] S513: Based on the high-risk distribution range, compare it with the risk classification criteria, analyze its distribution range and grouping conditions, filter the data that meets the high-risk group, and classify the remaining data into the medium and low-risk group to obtain the psychological risk classification index.

[0034] Compared with the prior art, the advantages and positive effects of the present invention are as follows:

[0035] In this invention, user interaction behavior and virtual situation events are dynamically mapped to achieve closed-loop management of the entire data flow. The acquisition window is adjusted in conjunction with multiple source parameters, signal segmentation is defined in real time based on actual EEG fluctuations and scene cognitive state, EEG electrode values ​​and behavioral response timing are extracted synchronously, multi-dimensional features are coupled and analyzed, and the dynamic grouping of psychological risks is based on the joint setting of multiple indicator intervals. Data is guaranteed to be timely in high-frequency interaction scenarios, and the evaluation results can reflect the continuous evolution of the user's psychological state in complex environments. The system supports asynchronous feedback and scenario customization, and is suitable for different users and various psychological state discrimination scenarios. Attached Figure Description

[0036] Figure 1 This is a flowchart of the main steps of the present invention;

[0037] Figure 2 This is a flowchart of the process for obtaining the event adaptation determination sequence in this invention;

[0038] Figure 3 This is a flowchart of the acquisition process for the adjustment section of the acquisition window in this invention;

[0039] Figure 4 This is a flowchart illustrating the acquisition of the EEG segment feature set in this invention.

[0040] Figure 5 This is a flowchart illustrating the process of obtaining the extreme response characteristic interval in this invention.

[0041] Figure 6 This is a flowchart illustrating the process of obtaining the psychological risk classification indicators in this invention. Detailed Implementation

[0042] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0043] In the description of this invention, it should be understood that the terms "length," "width," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," and "outer," etc., indicating orientation or positional relationships, are based on the orientation or positional relationships shown in the accompanying drawings and are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the invention. Furthermore, in the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0044] Example

[0045] Please see Figure 1 This invention provides a technical solution: a VR-based mental health assessment method, comprising the following steps:

[0046] S1: Based on the contextual event triggering nodes in the virtual scene, analyze whether the scene event type corresponds to the preset event type, determine whether the user interaction state parameters are consistent with the standard state requirements, combine the event priority, sort according to the order of event impact, compare the adaptability of each node item by item, filter the nodes that meet the judgment criteria, and obtain the event adaptability judgment sequence.

[0047] S2: Based on the event adaptation judgment sequence, analyze the visual complexity of the event, and by comparing the correlation between the click frequency of interactive behavior and the event priority, calculate the adjustment range of the start and end time of the collection window for each event node to obtain the collection window adjustment segment.

[0048] S3: Based on the acquisition window adjustment segment, determine the continuous change of the instantaneous amplitude parameter of the EEG in each channel, compare the statistical difference between the average amplitude of the current interval and the average amplitude of the previous interval, analyze the amplitude change trend of the EEG in the difference interval, identify the moment when the amplitude change meets the segmentation criterion, and obtain the EEG segmentation feature set.

[0049] S4: Based on the EEG segment feature set, determine the range of each EEG signal segment, analyze the cognitive branch task nodes, construct an extreme value sequence by the maximum and minimum amplitudes of EEG, compare it with the user's behavior response time, calculate the correlation between the corresponding intervals of the two, and obtain the extreme value response feature interval.

[0050] S5: Based on the extreme value response characteristic interval, analyze the joint state of the extreme value sequence duration interval length and the behavioral response time interval, compare the degree of matching between the joint state and the high-risk duration segment benchmark and the delay time benchmark, screen the group belonging to the high-risk standard, and classify the remaining data into medium and low-risk groups to obtain the psychological risk classification index.

[0051] The event adaptation judgment sequence includes node validity, node influence degree, and node hierarchical information. The acquisition window adjustment segment includes window division method, window identifier, and window associated events. The EEG segment feature set includes segment type, segment stability, and segment range index. The extreme value response feature interval includes response coupling coefficient, extreme value interval distribution, and behavioral synchronization characteristics. The psychological risk classification index includes risk level, group label, and risk interval boundary.

[0052] In S1, scene event type refers to a pre-defined category of interactive events in a VR virtual scene task. For example, a specific task or plot trigger node in modules such as "psychological testing," "relaxation training," or "attention training." Pre-defined event types refer to the event category standards defined in the psychological assessment plan or software system backend; that is, a list of event types built into the system and used as a reference for judgment. User interaction state parameters refer to the quantitative data of the user's state when operating or participating in the VR scene. For example, whether a certain interactive action such as clicking, dragging, selecting, moving, gazing, or responding to a voice has been completed, as well as the duration and frequency of the action. Standardized state refers to the qualified / target state specified by the system for each type of event node. For example, "must complete a certain interactive action," "maintain attention for the required duration," "operation process not interrupted," etc. Node adaptability refers to the degree to which the user's current state matches the system's pre-defined standards during actual user participation in each situational event node. It is used to determine whether the user's performance at that node meets the standard. Judgment criteria refer to the set of rules used to filter nodes, generally set by logically combining parameters such as event type, interaction state, and priority. It is the basis for judging whether a node "meets" the evaluation conditions.

[0053] In S2, visual complexity of an event refers to the richness of visual elements in the virtual scene corresponding to an event node (such as color, dynamics, number of elements, visual interference, etc.), which reflects the visual load and interference level of the task. Interactive behavior refers to the specific operations performed by the user in the virtual scene, such as button clicks, object dragging, eye focus, head rotation, etc. Correlation refers to comparing the frequency / pattern of interactive behavior with the event priority to analyze whether the interactive activity matches the importance of the event. Acquisition window refers to the specific time period (start and end time) used to acquire EEG signal data, that is, the time range for automatically locating EEG acquisition before and after the event node occurs. Adjustment interval refers to the specific time period length and position of the acquisition window dynamically adjusted through the above analysis, which is the result of optimization and positioning of the default acquisition window.

[0054] In S3, each channel refers to the signal pathway corresponding to different acquisition electrodes on the EEG device, such as multi-channel EEG acquisition of the forehead, parietal lobe, etc.; the current interval / previous interval refers to the time period of EEG data divided according to the acquisition window, "current interval" refers to the time period being analyzed, and "previous interval" refers to the previous time period that has been analyzed; statistical difference refers to comparing the numerical changes between two intervals of EEG signals (such as statistical measures such as mean, variance, extreme values, etc.) to judge the fluctuation of the signal; amplitude variation trend refers to the trend of EEG signal rising, falling or remaining stable over time, reflecting the dynamic changes of the subject's psychological or physiological state; the moment that meets the segmentation criteria refers to the time point when the signal shows obvious changes or abrupt changes by judging the amplitude variation trend, which is used as the starting point of the new data segmentation.

[0055] In S4, cognitive branch task nodes refer to specific cognitive tasks designed in psychology courses (such as attention tests, reaction time tests, and stress responses), corresponding to key task steps or nodes in the scenario; extreme value sequences refer to the ordered set of maximum and minimum amplitudes extracted from each segment of EEG signal, reflecting the main fluctuation range of the signal within the segment interval; user behavior response time refers to the time taken for a user to complete a specified operation (such as clicking, moving, etc.) in a virtual scenario task, reflecting the speed of cognitive and psychological reaction; corresponding interval correlation refers to the synchronicity or correlation analysis of EEG value fluctuations and behavior response time within the same time period, examining the degree of coupling between the changes of the two.

[0056] In S5, the joint state refers to the state presented by both the duration segment of the extreme value sequence and the behavioral response time interval, which serves as the basis for joint judgment; the high-risk duration segment benchmark and the delay time benchmark refer to the EEG value duration standard and behavioral response time delay standard used to define the "high risk" judgment. The benchmark can be set by historical data and industry standards; the high-risk standard group refers to the user group whose joint state is identified as high risk by the system; the medium and low risk group refers to other groups that do not meet the high-risk standard (generally ordinary or alert level).

[0057] By employing a data acquisition architecture based on the collaborative operation of an XR EEG feedback system kit and a VR all-in-one device, including:

[0058] When participating in the mental health assessment process, users wear portable EEG sensors (such as XR brain-computer interfaces with integrated TGAM chips) to collect multi-channel EEG signals from different brain regions (such as the frontal and parietal lobes) in real time; at the same time, they wear and operate VR all-in-one devices (such as PICO Neo3) to load preset virtual scene psychological course tasks, which include multiple modules such as attention tests, relaxation, stress simulation, and psychological catharsis.

[0059] The VR scene is pre-configured with multiple situational event trigger nodes. These nodes are configured via scripts to specify the conditions under which events are triggered at different times or when the user completes different interactive actions (such as gazing, clicking, moving, or responding with voice). When a user interacts with the scene in the VR virtual environment, for example, by staring at a rapidly appearing red sphere for more than 1.5 seconds in the "attention focus test," the node's validity is determined. Based on the node's determination result and the user's operation data, the collection of EEG signals is triggered within the corresponding time period, and the XR brain-computer interface continuously sends multi-channel electrical signal data streams from the user's brain.

[0060] Based on dynamic analysis of the priority of scene task nodes, visual load, user interaction density, and instantaneous amplitude changes of multi-channel EEG, acquisition window adjustment segments are generated for different nodes. These acquisition windows are used to dynamically determine how long before and after the node occurs to record EEG signals. For example, the default 10-second window can be extended to 15 seconds or shortened to 7 seconds to ensure that the acquired data accurately corresponds to the current psychological stimulus task.

[0061] The XR brain-computer interface device worn by the user samples brainwaves from different channels, such as the forehead and parietal lobe, within the aforementioned window, forming multi-channel parallel raw time-series data. After preliminary filtering on the XR brain-computer interface, this data is transmitted wirelessly to the VR all-in-one device via Bluetooth or WiFi, or directly to the backend server for real-time recording.

[0062] Synchronously record the timestamps generated by all behavioral operations (clicks, drags, gaze durations, voice inputs) performed by the user in the VR scenario. By comparing the user's behavioral time points with the occurrence times of the brain electrode value segments item by item in the subsequent process, precise alignment of the behavioral data and the electroencephalogram data on the time axis is achieved, which can be used to derive the psychological risk classification indicators.

[0063] Please refer to Figure 2 , and the steps for obtaining the event adaptation determination sequence are specifically as follows:

[0064] S111: Based on the situational event trigger nodes in the virtual scenario, analyze the types and preset standards of the situational events in the virtual scenario, compare the matching relationship between each event type and the standard type, determine whether the behavioral parameters of the user during the interaction conform to the specifications, screen the nodes that meet the standards, and obtain the node adaptability comparison sequence;

[0065] Call the scenario task script configuration list, sequentially read the type, number, position coordinates of each node record, and the expected standard values for the user's behavior. Subsequently, open the preset event type standard list in the psychological assessment background, and compare the type of each node item by item with the types listed in this list. If there is an identical item for the node type in the list, it is determined that the node type match is valid, and then this node is further analyzed; otherwise, it is directly skipped. Then, for the valid nodes, call the behavioral parameter values generated by the user during the interaction, such as reading the first gaze focus delay time, total gaze duration, and operation completion flag after the red sphere appears. Compare the user's first gaze delay with the maximum allowed delay of the node. If it is less than or equal to, it is determined to be qualified. Compare the user's total gaze duration with the minimum gaze time required by the node. If it is greater than or equal to, it is also considered qualified. Then, check whether the node operation step is finally completed. If all three conditions are met, this node is recorded as a specification matching node; otherwise, it is judged as unqualified. For example, node 1 requires a maximum allowed delay of 1 second and a minimum gaze of 1.5 seconds. The user's data on this node is 0.8 seconds of delay, 1.7 seconds of gaze, and the task is completed, all of which are qualified, so node 1 is selected for the next step of processing. If the user's delay in node 2 reaches 1.3 seconds, exceeding the standard, it is directly judged as unqualified, and the subsequent values are not compared. By reading, comparing, and judging the compliance of the behavior item by item for all nodes in this way, the node numbers of all nodes whose behavioral data meet the node standards are collected into the adaptability comparison sequence.

[0066] S112: According to the node adaptability comparison sequence, combined with the event priority, calculate the relative position differences of each node during the sorting process, screen the key node content, and obtain the node screening and sorting sequence;

[0067] Each node in the sequence is sequentially recorded by calling its priority value configured in the scenario task script. Priorities are typically integers from 1 to 5, representing the node's importance in the psychological assessment process. A pairwise comparison is used to directly compare the priority of each node with all other nodes in the sequence. If the current node's priority is higher than another node, a higher priority count is added. After all comparisons are completed, the number of times each node is higher than other nodes in the sequence is recorded. For example, after comparing node 1 with node 4, node 1's priority of 4 is higher than node 4's 3, so node 1 is recorded as higher. Then, when compared with node 6, node 6's priority of 5 is higher than node 1, so no increase is added. Through this process, node 1... Node 6 has a total count higher than the two nodes by two times, and Node 6 has a count higher than the opponent by five times in all comparisons. Then, the nodes are sorted from largest to smallest according to the higher count to form a priority list, which is then compared with a screening benchmark value. For example, the benchmark is set to 2. This benchmark is derived from the statistical analysis of the average priority distribution of nodes in similar virtual scene psychological training in the past, which is about 2.1. Adjusting it to 2 is used to screen more representative nodes. In this way, Node 6 is included in the key focus list because it has a total count of five times higher than the benchmark, Node 1 is also included because it has a total count of two times equal to the benchmark, and Node 4 is excluded because it has only a total count of one time lower than the benchmark. The resulting key focus node screening and sorting sequence includes Node 6 and Node 1, which are used for subsequent dynamic adjustment of the acquisition window and EEG segmentation calculation stage.

[0068] S113: Based on the node selection and sorting sequence, compare the effectiveness, influence, and hierarchical attributes of each node using the following formula:

[0069] ;

[0070] Obtain node adaptation differences It then determines the adaptation status of each node to the baseline, obtaining an event adaptation determination sequence, where, Indicates the total number of nodes. Indicates the first The validity parameters of each node, Indicates the first Hierarchical information parameters of each node, Indicates the first The parameter representing the degree of influence of each node. Indicates the first The corresponding parameters of each node in the node fitness comparison sequence. Indicates the first The corresponding parameters of each node in the node filtering and sorting sequence.

[0071] Node adaptation difference refers to the degree of difference between the current state of a node and the requirements of a scenario event node in a virtual scenario task, based on the node's validity, hierarchical attributes, and degree of influence, and compared with the system's preset standards. It is used to describe the degree of deviation of a single node from the adaptation specification requirements throughout the entire process, and is presented as a data-driven indicator that can be used for sorting, filtering, and further judgment.

[0072] Obtain the normalized node validity parameters They are respectively , , The corresponding node hierarchical information parameters They are respectively , , Node influence parameter They are respectively , , The parameters of the corresponding node in the node adaptability comparison sequence are called. They are respectively , , The parameters of the corresponding node in the sorted sequence are called. They are respectively , , The parameters mentioned above are calculated according to the formula, and the values ​​at each node in the molecule are calculated. Result:

[0073] Node 1 contains:

[0074] ;

[0075] In node 2:

[0076] ;

[0077] Node 3 contains:

[0078] ;

[0079] After adding the three terms together, the numerator is obtained as follows:

[0080] .

[0081] Calculate each node in the denominator The summation result:

[0082] Node 1 is:

[0083] ;

[0084] Node 2 is:

[0085] ;

[0086] Node 3 is:

[0087] ;

[0088] The sum of the three is:

[0089] ;

[0090] After calculating the square root, it becomes .

[0091] After substituting into the formula, we get

[0092] ;

[0093] Combined with node adaptation differences The calculation results are used to determine whether the preset adaptation reference range is [missing information]. to Between, the result obtained in this calculation If the result falls within this range, it indicates that the analyzed node has strong adaptability after comprehensively considering multiple parameters such as effectiveness, hierarchical information, and degree of impact. This result is also used as a component of the event adaptation judgment sequence and is passed to subsequent stages. The formula integrates three highly structured upper-level indicators into a unified differential data structure and uses the square root normalization process to uniformly measure the underlying sorting and comparison items, so that the entire event adaptation judgment sequence can achieve consistency measurement across task scenarios and difference comparison between nodes, thus completing the acquisition of the event adaptation judgment sequence.

[0094] Please see Figure 3 The specific steps for obtaining the adjustment section of the acquisition window are as follows:

[0095] S211: Based on the event adaptation judgment sequence, for the visual performance of each node, the color, shape, dynamic elements and visual hierarchy features in the scene are statistically analyzed, and combined with the degree of node influence and hierarchical relationship, the visual load of the node in the scene is judged to obtain the event visual load index.

[0096] For each node, its pre-configured visual parameters in the scene file are read. These parameters record the number of colors, the number of geometric shapes, whether it contains dynamic animation elements, and the node's layer depth number in the entire scene rendering sequence. For instance Node 1, the parameters are read as follows: 6 colors, 4 shapes, dynamic element flag (yes), and layer number 5. The number of colors is compared with the scene's general baseline intervals [0, 3], [4, 6], and [7, 10] to determine that the node's color count falls into the second tier, i.e., the medium range. The number of shapes is then compared with the intervals [0, 2], [3, 5], and [6, 8] to confirm it's also at a medium level. The dynamic element flag is directly checked for existence (a binary state; since Node 1 has dynamic elements, it's considered as 1). Finally, the layer number is compared with the preset intervals [1, 3] and [4, 6], indicating it's in the high-level segment. The numerical values ​​are summarized into node visual parameter states: medium color, medium shape, dynamic, and high level. Then, combined with the node influence degree field (e.g., node 1 is set to influence degree of 4 (integer value between 1 and 5, with 5 being the strongest) and level relationship of 5), the values ​​are added to obtain a temporary visual load index initial value of 19. Then, it is compared with the preset visual load division benchmark value of 20. Since 19 is slightly lower than the benchmark value of 20, it is judged as a medium load node. If the visual parameters of node 2 are statistically 9 colors, 7 shapes, dynamic, level 5, and influence degree of 5, the sum is 26, which is greater than the benchmark value of 20, so it is judged as a high load node. By reading the number of node graphic elements, judging whether there is animation, comparing the level number, adding the node preset influence value, and comparing it with the benchmark value, the visual load level of each node is formed. All nodes are gradually judged after such calculation to form a set of node visual load indices.

[0097] S212: Based on the visual load index of the event, analyze the relationship between the click situation of interactive behavior in the event node and the node criticality, calculate the response distribution of interactive operation in different nodes, and filter the data segments with prominent correlation to obtain the interactive coupling distribution sequence.

[0098] A correlation analysis is performed on the number of clicks recorded in the interaction behavior log for each node and the node's criticality value. First, the click count is read from the record table one by one. For example, node 1 records a total of 12 clicks and its criticality is configured as 4. Multiplying the click count and criticality values ​​gives 48. Next, node 2 records 22 clicks and its criticality is 5, multiplying the two values ​​gives 110. Then, the product values ​​of the nodes are compared one by one, and the nodes are divided according to the intervals [0, 50], [51, 100], and [101, 150]. Node 1 is classified as medium-range and node 2 as high-range. Then, it is checked whether the click distribution of node 1 in the interaction log is concentrated in a certain time period of the task. For example, it is counted that node 1 has 9 clicks in the 10th to 20th second period. A 75% share of total clicks is considered a concentrated distribution, while the average distribution of node 2 from the 5th to the 25th second is considered dispersed. By recording the interaction density and key value accumulation of each node within a time period, and combining this with a preset threshold of 80, prominent interaction density is determined. If the node product value exceeds 80 and the main concentrated distribution ratio exceeds 60%, it is considered a node data segment with prominent correlation. Node 2, with a product value of 110 and an interaction ratio of 65% in the main time period, meets the condition and is selected. Node 1, with a product value of 48, although concentrated in time, does not exceed the threshold and is not selected. Thus, by reading click records, performing direct numerical product, comparing distribution ratios, and comparing with a fixed threshold, all node interaction segments with significant correlation are obtained and formed into an interaction coupling distribution sequence.

[0099] S213: Based on the interactive coupling distribution sequence, combined with the visual load and interactive distribution changes of each node, determine the range in which the acquisition window of each event node needs to be extended or reduced, optimize the start and end limits of the acquisition time period, and obtain the acquisition window adjustment section.

[0100] The visual load level and interaction distribution records of each node are read again. In this process, it is first judged that if a node is judged to have high visual load and has a high interaction density in the interaction coupling sequence, the acquisition window time of the node needs to be extended. The extension ratio is based on a fixed window adjustment coefficient of 1.5, which extends the acquisition window from the default 10-second acquisition window to 15 seconds. Then, for a node with medium visual load but low interaction coupling value (i.e., the product is less than 50), the acquisition window is reduced to 0.7 times the original value, which is 7 seconds. In the judgment process, the node attributes are read one by one to determine whether the two conditions are simultaneously satisfied: high load and high coupling or medium-low load and low coupling. Then the acquisition window time adjustment is performed. By reading two types of index values, performing dual condition judgments, and directly modifying the start and end times of the acquisition window, the acquisition window time range of all nodes is gradually extended or reduced. After completion, the adjusted start time and end time of all nodes are recorded in the acquisition window adjustment segment result table.

[0101] Please see Figure 4 The specific steps for obtaining the EEG segment feature set are as follows:

[0102] S311: Based on the acquisition window adjustment segment, analyze the continuous changes in the instantaneous amplitude of multi-channel EEG, compare the amplitude changes of each channel in different segments, and determine the fluctuation between segments by constructing the amplitude change sequence between multi-channel data to obtain amplitude change data;

[0103] First, the raw EEG data from multiple channels within the acquisition window corresponding to each node is read sequentially. Each window is divided into multiple time segments, and the EEG sampling values ​​within each segment are arranged in millisecond sequences in millivolts. During reading, signal sequences are extracted separately for different channels. For example, the time series extracted from the frontal channel has 1200 points, and the time series extracted from the top of the head has 1180 points. Then, the instantaneous amplitude difference between two adjacent sampling points is calculated in each time series. By continuously calculating, a list of amplitude changes for each channel within the window is obtained. Then, the instantaneous amplitude differences are accumulated and averaged to form the average amplitude change rate for each channel. Finally, the average amplitude change rates of the frontal channel and the top of the head channel are directly compared. If the average amplitude of the frontal channel is 1.3 millivolts and that of the top of the head is 0.7 millivolts, then the frontal channel is considered to have a higher change in that window. Based on the preset amplitude variation range [0, 0.5] as stable, [0.6, 1.0] as moderate fluctuation, and [1.1, 1.5] as significant fluctuation, 1.3 is determined to be a significant fluctuation. After processing all channels sequentially, a multi-channel amplitude variation sequence is summarized in each window. This sequence is then compared with the average amplitude variation rate of the same channel in the previous acquisition window. If there is an interval where the amplitude variation increases by more than 0.5 mV, the window is marked as a significant difference segment in the data structure. For example, if the forehead channel in the second acquisition window of node 1 changes from 0.6 mV to 1.3 mV, it is recorded as a sudden change segment. By sequentially reading the original EEG sequence for each acquisition window, calculating the instantaneous amplitude difference point by point, and then comparing the average amplitude variation of the same channel across windows, the amplitude variation data of each segment of each node is obtained.

[0104] S312: Based on amplitude variation data, analyze the differences in continuously varying segments using the following formula:

[0105] ;

[0106] Obtain the changing fluctuation sequence Determine the range of abrupt changes in multi-channel data to identify amplitude abrupt change segments. This indicates the total number of data points within the amplitude variation data. Indicates the first Data items in the amplitude change data, Indicates the first [unit] in the current segment The data item representing the instantaneous amplitude of the channel. Indicates the first in the previous segment The data item representing the instantaneous amplitude of the channel. This represents the normalized denominator term. The number of segments indicating abrupt changes. Indicates the first Data items at the starting point of each mutation segment, Indicates the first Data items at the endpoint of each mutation segment;

[0107] A fluctuation sequence refers to an ordered sequence formed by statistically analyzing, aggregating, and calculating the amplitude changes of multi-channel EEG signals between adjacent segments. It is used to describe the fluctuation intensity and amplitude of EEG signals in each acquisition segment. It is a comprehensive measurement result of the fluctuation of multi-channel EEG signals between different acquisition windows and is one of the key data foundations in the analysis process.

[0108] The number of participating items that constituted the differences was counted, and the total number of samples was obtained by combining the data from the previous stage. The amplitude change terms involved in the calculation are extracted as follows: , , , After normalization, we obtain the following results: , , , ;

[0109] The instantaneous amplitude data for the current channel segment are as follows:

[0110] , , , ;

[0111] After normalization, they are respectively:

[0112] , , , ;

[0113] The instantaneous amplitude data of the channel in the previous time period is as follows:

[0114] , , , ;

[0115] After normalization, it becomes:

[0116] , , , ;

[0117] The normalization term of the sampling denominator is set to It is a constant term constructed based on the maximum amplitude range and standard deviation of the sample;

[0118] The collected mutation segments comprise two time periods, with the starting data items being respectively... , After normalization, they are respectively , The terminating data items are respectively , After normalization, it becomes , Therefore .

[0119] Substitute the above values ​​into the original formula:

[0120] ;

[0121] The first part, expanded sequentially into the square term and the product term, is as follows:

[0122] ;

[0123] ;

[0124] ;

[0125] ;

[0126] Sum of the above four items:

[0127] ;

[0128] Divide by Root of the square root:

[0129] ;

[0130] Part Two Calculations:

[0131] ;

[0132] Combinatorial operations:

[0133] ;

[0134] This result indicates that, within this acquisition window, the aggregate fluctuation value formed by multiple channels within the continuous sampling segment is... Therefore, in subsequent steps, the amplitude abrupt change segment can be extracted and identified based on this value, and the segment can be included in the data source item for subsequent stability and extreme value segmentation analysis. The larger the value, the stronger the channel aggregation fluctuation. The formula takes into account both the intensity and direction of change by combining the square and cross product, avoiding the situation of misjudging aggregation fluctuation due to the offset of a single channel.

[0135] S313: Based on the amplitude mutation segment, filter its multi-channel data, optimize the amplitude extreme features within the segment, and statistically analyze the segmentation type and amplitude stability within each segment to obtain the EEG segmentation feature set.

[0136] First, the multi-channel data segments marked with significant amplitude changes in the previous steps are selected. For example, node 1, where both the forehead and top of the head are recorded as abrupt changes in the second acquisition window, is selected for analysis. Then, the refined original sample value sequence is reread within this abrupt change segment, and the maximum and minimum amplitude values ​​are calculated. For example, if the maximum amplitude of the forehead channel is 2.1 mV and the minimum is 0.4 mV, the span is calculated to be 1.7 mV. Subsequently, based on the amplitude stability range [0, 0.5] as stable, [0.6, 1.2] as moderate fluctuation, and [1.3, 2.0] as high fluctuation, 1.7 mV is identified as a high fluctuation segment, and the number of samples in this segment is then counted. The duration of continuous increase or decrease is used to determine the segment type. If the continuous increase lasts for 150 milliseconds, which is greater than the stable baseline of 100 milliseconds, it is recorded as a long increasing segment and marked as a strong increasing segment type. In this way, the maximum and minimum value difference calculation and continuous trend length statistics are performed on the original data in each mutation segment to compare whether it exceeds the stability threshold and assign segment labels accordingly. Finally, all segments of the node are summarized to form the statistical results of segment type and amplitude stability. The second window of node 1 is recorded as a strong increasing high fluctuation segment, and if the amplitude span of the third window of node 1 is 0.9 millivolts, it is judged as a medium fluctuation short increasing segment. After all the results are sorted, an EEG segment feature set is generated.

[0137] Please see Figure 5 The specific steps for obtaining the extreme value response characteristic interval are as follows:

[0138] S411: Based on the EEG segment feature set, analyze the amplitude characteristics and fluctuation performance of each segment, compare the extreme amplitude with the minimum amplitude for each cognitive branch task stage, calculate the amplitude difference, and screen related segments to obtain the extreme amplitude difference segments.

[0139] For each segment in the set, its stored maximum and minimum amplitude values, along with the corresponding segment start and end times, are sequentially retrieved. The maximum and minimum values ​​within the same segment are then subtracted to obtain the amplitude difference. The difference is then categorized according to preset amplitude difference intervals: [0, 0.5], [0.6, 1.2], and [1.3, 2.0]. For example, if segment 1 has a maximum amplitude of 2.1 mV and a minimum amplitude of 0.7 mV, the difference is 1.4 mV, falling within the [1.3, 2.0] interval and thus considered a significant difference segment. The cognitive branch task stage identifier corresponding to the current segment is then read. If segment 1 belongs to stage 3 of the "Attention Test," this segment is recorded separately under the branch task. This calculation is repeated for all EEG segments, directly comparing the maximum and minimum values ​​of each segment to obtain the amplitude difference, which is then compared with the baseline values ​​for each interval. The range is determined by subtracting the maximum value of 0.8 mV from the maximum value of segment 2, resulting in a difference of only 0.2 mV. This falls within the range [0, 0.5] and is considered a slight difference. By reading the amplitude values, performing direct subtraction, and comparing with the preset difference range, the amplitude difference level of all segments is determined. All segment numbers in the medium and significant difference range are filtered in the segment registration table to form a preliminary list. Then, the task stage corresponding to the segment is checked, and segment numbers with multiple high amplitude differences within the same task stage are merged into the summary table of that stage. For example, if there are three segments with amplitude differences above 1.3 mV under "Stress Task Stage 2", they are all recorded together. By reading, calculating, classifying, and merging stage segments in this way, the summary table records the cognitive task stage and segment information corresponding to all segments with large amplitude differences, forming extreme amplitude difference ranges.

[0140] S412: Based on the extreme value amplitude difference range, compare it with the time distribution of user behavior response time. Through the correspondence of task stages, determine the time sequence of extreme value fluctuations and response behaviors, identify combinations with close correspondence, and obtain extreme value response coupling combinations.

[0141] First, the start and end times of each extreme difference segment are read sequentially. Then, the user's operation records for the same task are retrieved from the behavior data log file. A direct comparison is performed to determine whether the user's behavior response time falls within the time range of the EEG value difference segment. For example, if the extreme segment lasts from the 15th to the 20th second and the user clicks at the 17th second, it is recorded as an overlapping interval. Then, the time distribution of other behaviors under the same task is checked. If there are more than two user clicks or drags in the extreme difference segment, it is determined that there is a close correspondence. Then, the identification information of the task stage is read, such as "Psychological catharsis task stage 1". This combination is added to the coupling check result table. Subsequently, each extreme amplitude difference segment and all records in the task are traversed. The user behavior time of each phase is repeatedly compared. For example, if the extreme value range is from the 22nd to the 26th second, but the user only operates at the 28th second, it is not considered a close correspondence. By repeatedly reading the extreme value range time value and the user behavior time value, and comparing the time sequence and whether it is within the time range after each reading, the number of times the user operation occurs in each segment is counted. Then, segments with more than 2 times are considered to be closely corresponding segments. This threshold is based on the average number of behavior triggers of 20 similar psychological courses in the past, which is 2.4 times, rounded down to 2. It is used to determine that at least 2 overlaps are required to be considered as having obvious coupling. After repeating all the records, a coupling combination table is generated, summarizing all extreme value response coupling combinations that are judged to be closely corresponding to the extreme value fluctuation and the user behavior response in terms of time sequence.

[0142] S413: Based on the extreme response coupling combination, the coupling performance between amplitude changes and behavioral intervals among groups is analyzed, using the formula:

[0143] ;

[0144] By filtering out key correlated segments, we obtain the extreme value response characteristic intervals, where... Represents the correlation coefficient of the coupling interval. Indicates the first Group extreme value duration Indicates the first Group amplitude conversion rate, Indicates the first Group behavior response interval Indicates the first The difference in magnitude of the task phases. This indicates the total number of sets of coupled data pairs.

[0145] The coupling interval correlation coefficient refers to a quantitative coefficient that reflects the degree of coupling between extreme value changes and behavioral responses within the same task phase. It is calculated by analyzing multiple data points, such as the duration of extreme values, the rate of change of amplitude, and the interval of behavioral responses, for each group of extreme value coupling sets. It represents the overall correlation level between extreme value fluctuation characteristics and user behavior. It can be used to analyze and judge the synchronicity, correlation, or linkage between changes in EEG electrode values ​​and user behavioral responses in different segments, and further provide a basis for subsequent judgments such as screening extreme value response characteristic intervals and risk classification.

[0146] Extract the corresponding data from the selected task phases 3, 4, and 5, and obtain the extreme value duration for each combination. Amplitude conversion rate Behavioral response interval and the difference in the magnitude of the task phase In mission phase 3, the duration of the original extreme value is After normalization The amplitude transformation rate is synthesized from local fluctuation data, and the original calculation is as follows: After normalization The original behavioral response interval was... Normalization The amplitude difference at this stage was originally... After normalization Substitute into the formula to calculate:

[0147] In mission phase 3, the following is defined:

[0148] ;

[0149] Similarly, in task phase 4, , , , ,calculate:

[0150] ;

[0151] In mission phase 5 , , , ,calculate:

[0152] ;

[0153] Compare the above results with the benchmark coefficient. Comparison revealed that since all three sets of data were less than the baseline coefficient, no high-coupling stage was selected. The determination of the extreme response characteristic interval needs to be further calculated in other task stages or more combinations. The formula introduces the difference between the extreme value duration and the amplitude change rate and combines the behavioral response interval and the amplitude difference to form a composite ratio, which helps to more objectively reflect the overall coupling characteristics of EEG fluctuations and behavioral responses, thereby achieving effective quantification of group screening in multi-stage tasks.

[0154] Please see Figure 6 The specific steps for obtaining psychological risk classification indicators are as follows:

[0155] S511: Based on the extreme value response characteristic interval, analyze the duration interval of the extreme value sequence and the user's behavioral response time interval, compare the changing trends of the two at each corresponding node, sort out the differences in performance under all nodes, identify data nodes with consistent difference characteristics, and obtain extreme value behavior difference nodes.

[0156] The duration of extreme value sequences recorded within the feature interval is read sequentially, and the maximum and minimum amplitude durations corresponding to each node are extracted. Then, the list of user behavior response times in the virtual scene task under the same node is called, and the two sets of time periods are compared item by item. The difference between the duration interval length of the extreme value sequence and the behavior response time interval is directly calculated to obtain the change difference of each corresponding node. Then, according to the preset difference judgment interval [0, 0.5 seconds] is judged as slight difference, [0.6, 1.5 seconds] as moderate difference, and [1.6, 3 seconds] as significant difference, the interval is determined. For example, the maximum amplitude of node 1 is maintained for 2.1 seconds, the user click behavior interval is 1.0 seconds, and the difference of 1.1 seconds falls into the moderate difference zone. Node 2 maintains a maximum amplitude of 3.5 seconds, with a user response delay of 0.5 seconds. The difference of 3.0 seconds falls into the significant difference zone. The difference levels of each node are recorded one by one. Then, the data trends of the same node in different stages are compared. If node 1 obtains difference values ​​of 0.9 seconds, 1.2 seconds, and 1.3 seconds in the three stages respectively, showing an increasing trend, it is recorded as an increasing trend. If node 2 is 2.9 seconds, 3.1 seconds, and 3.0 seconds in the three stages, it is recorded as a stable high difference trend. By reading the time value of each node, performing direct difference calculation, judging the level by interval, and comparing trends in multiple stages, all nodes that show the same increasing or stable high value are picked out separately to form a consistent difference node list. The nodes are separately organized and marked as extreme value behavior difference nodes.

[0157] S512: Based on extreme value behavior difference nodes, determine the matching of each node with the high-risk continuous segment benchmark and the delay time benchmark in terms of feature distribution, identify nodes that meet the high-risk judgment criteria, and obtain the high-risk distribution interval;

[0158] The system sequentially reads the extreme value duration length and user response time delay of each node record. Then, it compares the static configuration values ​​of the high-risk duration benchmark and the delay time benchmark. The high-risk duration benchmark is 2.5 seconds, and the high-risk delay benchmark is 1.2 seconds. If the extreme value duration length of a node record is greater than 2.5 seconds and the behavior response delay time is greater than 1.2 seconds, it is determined that both meet the high-risk criteria. For example, if node 2 has an extreme value length of 3.0 seconds and a delay time of 1.4 seconds, both exceeding the corresponding benchmarks, node 2 is marked as a high-risk node. Then, node 1 is checked; its extreme value length is 1.3 seconds, less than the benchmark, so it is directly marked as a medium-low risk node, and the delay is no longer compared. This process is repeated, reading the two types of time values ​​of each node and comparing them with the benchmark values. The simultaneous fulfillment of both conditions is used as the basis for determining whether a node is high-risk. This process is repeated for all nodes, and the risk classification status of each node is recorded. All nodes determined to meet the high-risk conditions are added to the high-risk distribution interval list, forming the high-risk distribution interval corresponding to that node.

[0159] S513: Based on the high-risk distribution range, compare it with the risk classification criteria, analyze its distribution range and grouping conditions, screen the data that meets the high-risk group, and classify the remaining data into the medium and low-risk group to obtain the psychological risk classification index.

[0160] The system reads the start and end times of the distribution of each identified high-risk node on the timeline. Then, it calls the grouping threshold recorded in the risk classification criteria file. This threshold file sets the cumulative high-risk interval length greater than 5 seconds as the high-risk group threshold. Next, it reads the cumulative high-risk interval length of each node throughout the entire process. For example, node 2, with a cumulative time of 6.3 seconds, exceeds the 5-second benchmark and is directly classified into the high-risk group, while node 3, with a cumulative time of 4.2 seconds, does not exceed the 5-second benchmark and is classified into the medium-low-risk group. The cumulative time length of each node is read and directly compared with the fixed group benchmark value, and the results are recorded in sequence. All nodes exceeding the benchmark value are finally included in the high-risk group and a node identifier is generated. The remaining nodes that do not exceed the benchmark value are uniformly included in the medium-low-risk group, forming a psychological risk classification index. High-risk nodes and medium-low-risk nodes are registered in the output structure according to the group labels for subsequent data use.

[0161] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments that can be applied to other fields. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.

Claims

1. A VR-based mental health assessment method, characterized in that, The method comprises the following steps: S1: based on the context event trigger node in the virtual scene, analyze whether the scene event type is consistent with the preset event type, judge the user interaction state and the specification state requirement, combine the event priority, sort according to the event influence order, compare each node item by item, filter the nodes meeting the discrimination standard, and obtain the event adaptation judgment sequence; S2: based on the event adaptation judgment sequence, analyze the event visual complexity, compare the correlation between the interaction behavior click frequency and the event priority, calculate the adjustment start and end interval of each node collection window, and obtain the collection window adjustment section; S3: based on the collection window adjustment section, judge the instantaneous amplitude change of the brain waves of each channel, compare the average amplitudes of the current interval and the previous interval, analyze the brain wave trend of the difference interval, identify the time when the amplitude change meets the segmentation criterion, and obtain the brain wave segmentation feature set; S4: based on the brain wave segmentation feature set, judge each brain wave signal interval, analyze the cognitive branch task node, compare the maximum and minimum brain wave amplitudes to form an extreme value sequence, compare with the user behavior response time, calculate the corresponding interval correlation, and obtain the extreme value response feature interval; S5: based on the extreme value response feature interval, analyze the joint state of the extreme value sequence interval length and the behavior response interval, compare the matching with the high-risk section benchmark and the delay time benchmark, filter the high-risk group, and summarize the rest as a low-risk group, and obtain the psychological risk classification index; The psychological risk classification index includes risk level, grouping label, and risk interval boundary.

2. The VR-based mental health assessment method of claim 1, wherein, The event adaptation judgment sequence includes node effectiveness, node influence degree, and node hierarchical information, the collection window adjustment section includes window division method, window identifier, and window related event, the brain wave segmentation feature set includes segmentation type, segmentation stability, and segmentation range index, and the extreme value response feature interval includes response coupling coefficient, extreme value interval distribution, and behavior synchronization characteristics.

3. The VR-based mental health assessment method of claim 1, wherein, The acquisition step of the event adaptation judgment sequence is specifically: S111: based on the context event trigger node in the virtual scene, analyze the type and preset standard of the context event in the virtual scene, compare the matching relationship between each event type and the standard type, judge whether the behavior parameters of the user in the interaction process meet the standard, filter the nodes meeting the standard, and obtain the node adaptation comparison sequence; S112: according to the node adaptation comparison sequence, combine the event priority, calculate the relative position difference of each node in the sorting process, filter the node content of focus, and obtain the node filtering and sorting sequence; S113: according to the node filtering and sorting sequence, compare the effectiveness, influence degree and hierarchical attribute of each node, obtain the node adaptation difference, and judge the adaptation of each node to the benchmark, and obtain the event adaptation judgment sequence.

4. The VR-based mental health assessment method of claim 1, wherein, The acquisition step of the collection window adjustment section is specifically: S211: based on the event adaptation judgment sequence, for the visual performance of each node, count the color, shape, dynamic element and visual level features in the scene, combine the node influence degree and hierarchical relationship, judge the visual load of the node in the scene, and obtain the event visual load index; S212: Based on the event visual load index, the relationship between the click situation of the interaction behavior in the event node and the node criticality is analyzed, the response distribution of the interaction operation in the difference node is calculated, and the data segments with outstanding relevance are screened to obtain an interaction coupling distribution sequence; S213: According to the interaction coupling distribution sequence, in combination with the visual load and interaction distribution change of each node, the interval in which the collection window of each event node needs to be extended or reduced is judged, the start and end limits of the collection time period are optimized, and a collection window adjustment section is obtained.

5. The VR-based mental health assessment method of claim 1, wherein, The acquisition step of the EEG segmentation feature set is specifically: S311: Based on the collection window adjustment section, the continuous change of the instantaneous amplitude of the multi-channel EEG wave is analyzed, the amplitude change of each channel in the difference section is compared, the amplitude change sequence between the multi-channel data is constructed, the fluctuation between each section is judged, and amplitude change data is obtained; S312: Based on the amplitude change data, the difference of the continuous change section is analyzed, the change fluctuation sequence is obtained, the section range in which the mutation occurs in the multi-channel data is judged, and an amplitude mutation section is obtained; S313: According to the amplitude mutation section, the multi-channel data thereof is screened, the extreme amplitude feature in the section is optimized, the segmentation type and amplitude stability in each section are counted, and an EEG segmentation feature set is obtained.

6. The VR-based mental health assessment method of claim 1, wherein, The acquisition step of the extreme value response feature interval is specifically: S411: Based on the EEG segmentation feature set, the amplitude feature and fluctuation performance of each segmentation are analyzed, the extreme amplitude and minimum amplitude of each cognitive branch task stage are compared, the amplitude difference situation is calculated, and the associated section is screened to obtain an extreme amplitude difference section; S412: Based on the extreme amplitude difference section, the time distribution of the user behavior response time is compared, the time sequence of the extreme fluctuation and the response behavior is judged through the corresponding relationship of the task stage, the combination with a close corresponding relationship is identified, and an extreme response coupling combination is obtained; S413: Based on the extreme response coupling combination, the coupling performance of the amplitude change and the behavior interval in each group is analyzed, the section with key relevance is screened, and an extreme response feature interval is obtained.

7. The VR-based mental health assessment method of claim 1, wherein, The acquisition step of the psychological risk classification index is specifically: S511: Based on the extreme value response feature interval, the extreme sequence continuous interval and the user's behavior response time interval are analyzed, the change trend of each corresponding node is compared, the difference performance under all nodes is combed, the data node with consistent difference characteristics is identified, and an extreme behavior difference node is obtained; S512: Based on the extreme behavior difference node, the matching situation of each node and the high-risk continuous section benchmark and the delay time benchmark in the feature distribution is judged, the node meeting the high-risk judgment standard is identified, a high-risk distribution interval is obtained; S513: Based on the high-risk distribution interval, the distribution range and grouping condition are compared with the risk classification basis, the data meeting the high-risk grouping is screened, the remaining data is classified into a low-risk grouping, and a psychological risk classification index is obtained.

Images