Social gardening intervention and evaluation system based on VR technology

Abstract

The invention discloses a social gardening intervention and evaluation system based on a VR technology, and particularly relates to the field of auxiliary psychotherapy, and the system comprises a label combination space construction module which builds a three-dimensional coordinate space based on the cognition, emotion and behavior dimensions of a patient, and forms a psychological label combination code; the communication willingness evaluation module calculates a communication willingness score between two cooperative patients; a label atlas construction module constructs a label suitability atlas based on the communication intention score, and recognizes a conflict and complementary label relationship; a grouping reconstruction module executes patient isolation and cooperative group reconstruction according to the spatial distribution of the conflict labels; the task performance evaluation module is combined with different cooperation structures and psychological label input to construct a task performance evaluation model; and the intervention strategy configuration module selects an optimal intervention strategy configuration scheme through task response parameters output by the model based on a grouping structure and a task node candidate allocation strategy, and realizes task structure optimal matching under the guidance of individual labels.

Description

Technical Field

[0001] The present invention relates to the technical field of assisted psychotherapy, and more particularly, to a social gardening intervention and evaluation system based on VR technology. Background Art

[0002] In existing virtual social gardening intervention systems, users are typically grouped based on interests, age, frequency of use, or random system strategies to participate in social gardening activities such as shared plant care and collaborative landscape design. However, this grouping approach overlooks a key fact: users participating in gardening interventions often have different psychological issues, such as anxiety, depression, avoidant personality, and social phobia. These issues have significant psychological mechanisms that directly influence individual behavior during social collaboration and intervention responses. More seriously, existing systems fail to identify the potential "complementary" or "conflicting" relationships between these psychological differences, leading to coordination difficulties, psychological resistance, and ineffective intervention during actual collaboration. Furthermore, the systems lack continuous tracking and feedback mechanisms for collaborative effectiveness, making it impossible to dynamically optimize team structures based on recovery performance during the collaborative process. This results in low group intervention efficiency and a poor user experience. For users with more severe psychological issues or high social sensitivity, inappropriate collaboration combinations can even increase psychological burden. Summary of the Invention

[0003] In order to overcome the above-mentioned defects of the prior art, an embodiment of the present invention provides a social gardening intervention and evaluation system based on VR technology to solve the problems raised in the above-mentioned background technology.

[0004] To achieve the above object, the present invention provides the following technical solutions: A social gardening intervention and evaluation system based on VR technology includes a label combination space construction module, a communication willingness assessment module, a label map construction module, a grouping reconstruction module, a task performance assessment module, and an intervention strategy configuration module, wherein: The label combination space construction module maps the psychological problems of the patient group into a three-dimensional psychological label system composed of emotions, cognition and behavior, and constructs a psychological label combination space; The communication willingness assessment module monitors the virtual gardening collaborative task records, collects patients' interactive behavior data in the collaborative task context based on a sliding window, and calculates the communication willingness scores between patients with different psychological label combinations; The label map construction module maps the patient's communication willingness score to the psychological label combination space, constructs the label adaptability map, and distinguishes conflicting labels from complementary labels; The group reconstruction module isolates the individual patients corresponding to the conflicting labels and reconstructs the virtual gardening collaboration groups; The task performance evaluation module extracts task execution feedback data of individual patients corresponding to complementary labels under various gardening collaborative task structures and establishes a task performance evaluation model; The intervention strategy configuration module constructs a task execution strategy for patients in the virtual gardening collaboration group based on the output results of the task performance evaluation model and outputs the corresponding intervention strategy configuration.

[0005] In a preferred embodiment, the label combination space construction module maps the psychological problems of the patient group into a three-dimensional psychological label system consisting of emotions, cognition and behavior. The construction of the psychological label combination space specifically includes: Obtain psychological problem classification records from the medical files of the patient group and convert each patient's psychological problem classification record into cognitive scores, emotional scores, and behavioral scores; For each patient, a single-point psychological vector corresponding to the combination of emotion, cognition and behavior labels is established, and the label items corresponding to all single-point psychological vectors are embedded in the three-dimensional space structure to establish a psychological label combination space.

[0006] In a preferred embodiment, embedding the label items corresponding to the single-point psychological vector into a three-dimensional space structure to establish a psychological label combination space specifically includes: A three-dimensional coordinate system with cognitive dimension, emotional dimension and behavioral dimension as the axis is constructed, and the three types of label items in the single-point psychological vector are embedded in the corresponding axis and the polarity direction is fixed; Performing a three-axis mapping operation on the single-point mental vectors of all patients to generate the coordinates of the single-point mental vectors of the patients in a three-dimensional coordinate system; Based on the positive and negative combinations of the three-axis coordinates, the three-dimensional psychological space is divided into eight non-overlapping quadrants, and each quadrant is numbered. Get the quadrant number to which the patient's coordinate point belongs, and mark it as the encoding value of the patient's psychological label combination.

[0007] In a preferred embodiment, the communication willingness assessment module monitors the virtual gardening collaborative task records, collects the interactive behavior data of patients in the collaborative task context based on a sliding window, and calculates the communication willingness scores between patients with different psychological label combinations, specifically including: During the execution of the virtual gardening task, the interactive task nodes of two-person collaboration were identified, and the interaction behavior sequence was extracted according to the timestamps of the patient initiating and responding to the interaction in the task node. Define sliding time windows of equal width, extract the patient's interactive behavior segments window by window in the interactive behavior sequence, and count the cumulative number of initiating and responding interactions; In each sliding time window, the cumulative length of delayed response behaviors and the ratio of asynchronous response behaviors between patients were counted to establish an interactive response disruption index. A communication willingness scoring function based on empowerment fusion is constructed, and a communication willingness score is generated based on the cumulative number of interactions and the interaction response fracture index.

[0008] In a preferred embodiment, the label map construction module maps the patient's communication willingness score to the psychological label combination space, constructs the label adaptability map, and distinguishes conflicting labels from complementary labels, specifically including: Obtain the communication willingness scores of the two-person collaborative patients in each interactive task node and extract the coding values ​​of the two patients in the psychological label combination space; The communication willingness scores were grouped into quadrants based on the coding values, and the mean of the communication willingness scores within each quadrant group was calculated to construct a psychological label group compatibility score matrix. The psychological label combinations in the adaptability score matrix are used as nodes, and the relationships between the psychological label combinations are used as edges. A psychological label adaptability graph structure is established, and the adaptability scores are marked on the edges. The conflict and complementarity determination threshold segments are set, and different psychological label combinations are divided into conflict labels and complementary labels according to the threshold segments where the adaptability scores are located.

[0009] In a preferred embodiment, the group reconstruction module isolates individual patients corresponding to conflicting labels and reconstructs virtual gardening collaboration groups, specifically including: Retrieve the marked conflicting label pairs in the psychological label adaptability map and locate the corresponding patient psychological label combination code; Extract the individual patients in the current collaborative group that contain conflicting label combination codes, and remove the corresponding individual patient task participation history and interaction trajectory; Perform conflict minimization sorting based on the distribution of conflicting labels in each collaborative group to determine the priority isolation sequence of individual patients; Separate the priority isolation patients from the original collaborative grouping and construct a new virtual gardening collaborative group combination list.

[0010] In a preferred embodiment, the task performance evaluation module extracts task execution feedback data of individual patients corresponding to complementary tags under various gardening collaborative task structures, and establishes a task performance evaluation model, specifically including: The complete operation chain of the completed historical virtual gardening collaborative tasks is split into process nodes according to the task stages and interaction types, and a set of numbered task structure units is constructed; In the psychological label combination space, a single-point psychological vector consisting of the cognitive score, emotional score, and behavioral score of each patient in each reconstructed virtual gardening collaboration group is extracted; Collect feedback data from the execution logs of each task node in the task structure unit set, obtain the executor of the task node, and integrate the executor's single-point psychological vector and the response parameters of the task node into a task behavior sample set; The response parameters of the task node are obtained from the feedback data in the execution log, including the task progress rate, operation accuracy and collaboration synchronization rate; The task behavior sample set is used as the training sample set, the patient's single-point psychological vector and task node number are used as input items, and the response parameters of the task node are used as output items to train the task performance evaluation model.

[0011] In a preferred embodiment, the intervention strategy configuration module constructs a task execution strategy for patients in the virtual gardening collaboration group based on the output of the task performance evaluation model, and the output of the corresponding intervention strategy configuration specifically includes: Regularly update the medical records of the patient group. When the number of patients increases or the coordinates of the single-point psychological vector in the psychological label combination space change, update the label adaptability map and reconstruct the virtual gardening collaboration grouping; In each virtual gardening collaboration group, a task node allocation strategy table for all patients performing tasks is pre-generated, and the patient's single-point mental vector and task node allocation strategy are input into the task performance evaluation model; The output results of the task performance evaluation model are normalized and converted into task performance scores through weighted fusion. The task node allocation strategy corresponding to the highest task performance score is selected as the intervention strategy configuration.

[0012] The technical effects and advantages of the VR-based socialized gardening intervention and evaluation system of the present invention are as follows: By constructing a three-dimensional psychological label combination space consisting of emotion, cognition and behavior labels, we can achieve structured modeling of the patient's psychological state and enhance the accuracy of subsequent individual difference identification. By introducing an interactive behavior data collection mechanism in the gardening collaboration task and combining it with the sliding window method to achieve dynamic evaluation, we can effectively improve the objectivity and timeliness of the communication willingness evaluation without relying on traditional questionnaires. The constructed label adaptability map can accurately divide conflicting labels from complementary labels, providing a stable data foundation for the optimization of collaborative groups. By isolating and reconstructing the groups of patients with conflicting labels, we can avoid behavioral disorders caused by the superposition of psychological conflicts during the collaboration process. Furthermore, through the evaluation model established by task performance feedback data, the task execution results can be quantified and comparable, which improves the pertinence and interpretability of the intervention strategy generation. Finally, the model output results are applied to task strategy allocation, effectively constructing the correspondence between individual labels and task structure, thereby achieving the optimal configuration of collaborative task matching between patients and improving the flexibility of the overall intervention. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] Figure 1 This is a structural diagram of a socialized gardening intervention and evaluation system based on VR technology in the present invention. DETAILED DESCRIPTION

[0014] The following will provide a clear and complete description of the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0015] Example 1 Figure 1 The present invention provides a social gardening intervention and evaluation system based on VR technology, including a label combination space construction module, a communication willingness evaluation module, a label map construction module, a grouping reconstruction module, a task performance evaluation module, and an intervention strategy configuration module, wherein: The label combination space construction module maps the psychological problems of the patient group into a three-dimensional psychological label system composed of emotions, cognition and behavior, and constructs a psychological label combination space; The communication willingness assessment module monitors the virtual gardening collaborative task records, collects patients' interactive behavior data in the collaborative task context based on a sliding window, and calculates the communication willingness scores between patients with different psychological label combinations; The label map construction module maps the patient's communication willingness score to the psychological label combination space, constructs the label adaptability map, and distinguishes conflicting labels from complementary labels; The group reconstruction module isolates the individual patients corresponding to the conflicting labels and reconstructs the virtual gardening collaboration groups; The task performance evaluation module extracts task execution feedback data of individual patients corresponding to complementary labels under various gardening collaborative task structures and establishes a task performance evaluation model; The intervention strategy configuration module constructs a task execution strategy for patients in the virtual gardening collaboration group based on the output results of the task performance evaluation model and outputs the corresponding intervention strategy configuration.

[0016] The label combination space construction module maps the psychological problems of the patient group into a three-dimensional psychological label system composed of emotions, cognition and behavior, and constructs a psychological label combination space.

[0017] Historically registered psychological problem classification records are obtained from the patient's existing medical records. This record is derived from psychiatric clinical diagnoses, rehabilitation plan tracking forms, and disease labels generated by auxiliary psychological assessment systems. This information covers psychological problem categories that have been explicitly labeled or identified through long-term observation, such as anxiety disorders, depressive disorders, impulse control disorders, cognitive disorders, avoidant personality, and social disorders. To achieve unified calculation and evaluation, this step introduces a structured transformation rule library that establishes qualitative mappings between various psychological problem labels and three psychological dimensions: cognition, emotion, and behavior. This transformation rule library is developed by mental health practitioners and categorizes and models the common impacts of various common psychological problems in clinical manifestations and responses to behavioral interventions. For example, anxiety disorders are labeled as having a significant impact on the emotional dimension and also manifest as withdrawal behaviors in the behavioral dimension; impulse control disorders primarily affect the behavioral dimension and may also cause decision delays in the cognitive dimension; and cognitive disorders are directly mapped to the cognitive dimension while potentially having a weak impact on the emotional dimension. For each psychological problem label, the system assigns quantitative scores for three dimensions, ranging from -5 to +5. Positive values ​​indicate an enhancing or activating effect on the dimension, negative values ​​indicate an inhibitory or weakening effect, and 0 indicates no significant effect. For anxiety disorders, for example, the typical conversion is +4 for the emotional dimension, -2 for the behavioral dimension, and +1 for the cognitive dimension. After completing the dimensional conversion for all of a patient's psychological problem labels, the mapped values ​​for each label are summed within the dimension to generate a three-dimensional psychological value composite for each patient. This composite is represented as a standard vector consisting of cognitive, emotional, and behavioral scores, which serves as the patient's input identifier in subsequent task structure matching and group analysis. To prevent extreme data distribution, the system sets clipping boundaries for each dimension after conversion. Dimension values ​​exceeding ±5 are truncated to ±5, ensuring that all data fall within a reasonable statistical range.

[0018] Using three-dimensional psychological score vectors as the basic data input, a three-dimensional spatial coordinate system was constructed with the three dimensions of cognition, emotion, and behavior as the axes. The spatial structure of the coordinate system uses a rectangular Cartesian coordinate system, with the X-axis corresponding to the cognitive dimension, the Y-axis corresponding to the emotional dimension, and the Z-axis corresponding to the behavioral dimension. Each coordinate axis is centered at zero, and positive and negative directions represent positive and negative effects on that psychological dimension. Specific directions are as follows: positive directions in the cognitive dimension represent cognitive activity and clear logic, while negative directions represent cognitive retardation and difficulty in judgment. Positive directions in the emotional dimension represent elevated or stable emotions, while negative directions represent depression and anxiety. Positive directions in the behavioral dimension represent overt and positive behaviors, while negative directions represent withdrawal and apathy. All patients' three-dimensional psychological score vectors are mapped to this three-dimensional coordinate system in a single-point manner. To ensure comparability between dimension values, the system first performs linear normalization on the raw dimension values, mapping the range of -5 to +5 to a standardized interval of -1 to +1, ensuring that each dimension has an equal weight on the coordinate projection. For example, a patient's original psychological score is cognition +3, emotion -2, behavior +1. After normalization, the cognition is +0.6, emotion is -0.4, and behavior is +0.2. Then the coordinates of the single-point psychological vector in the three-dimensional psychological space are (0.6, -0.4, 0.2).

[0019] The space is divided into 8 non-overlapping quadrants. The division method is determined by the positive and negative value combination of each coordinate axis. For example, when the cognitive value is positive, the emotional value is negative, and the behavioral value is positive, the coordinate point belongs to the "positive-negative-positive" quadrant. To unify the numbering system, this embodiment adopts a three-digit binary numbering method, defining the positive direction as 1 and the negative direction as 0. The positive and negative combinations of each dimension are converted into three binary bits, forming a total of 8 categories of numbers from 000 to 111. For example, "negative-negative-positive" corresponds to the number 010, which is the second category of quadrants; "positive-positive-positive" corresponds to the number 111, which is the eighth category of quadrants. This quadrant number becomes the psychological label combination code value for each patient. The system binds each patient's psychological vector coordinate point to its quadrant number to generate a coding index and establishes a psychological label combination space, which includes the patient identification code, three-dimensional coordinates, quadrant number and label combination code.

[0020] The communication willingness evaluation module monitors the virtual gardening collaborative task records, collects the interactive behavior data of patients in the collaborative task context based on a sliding window, and calculates the communication willingness scores between patients with different psychological label combinations.

[0021] Collaborative task monitoring is performed within the virtual gardening system. The virtual gardening platform pre-configures various task structures, some of which include explicit two-person collaborative subtasks, such as alternating planting, synchronized pruning, and counter-moving. These task nodes have identifiable collaborative behavior entry points, which the system defines as two-person interactive task nodes. Identification criteria include the presence of an interactive action response between the two participants, and the actions must have temporal dependencies and completion interlocks within the task module. The virtual gardening platform's operation log data is used to extract behavioral records during collaborative tasks. This information includes the operation object ID, execution timestamp, behavior type label (e.g., tool call, planting confirmation, task progress), and operator ID. Within identified two-person interactive task nodes, the system selects the first interactive action initiated by each patient within the task node as an "initiate interaction" event and marks the first valid feedback action by the responder as a "respond interaction" event. An interactive behavior sequence is defined as a chain of initiation and response events between a pair of patients within the same task node. The system assembles all initiation-response behavior pairs into interaction behavior sequences in timestamp order, and assigns a patient pair number and task node identifier to each sequence.

[0022] A sliding time window mechanism is introduced to the constructed interaction sequence. The sliding time window is defined based on the average task cadence period. In this embodiment, referring to the median operation time of three common collaborative tasks, the length of each time window is set to 30 seconds, and the window sliding step is set to 15 seconds. That is, a 30-second time segment is extracted every 15 seconds in the sequence, forming a continuous sliding window sequence of equal width. This setting covers the behavioral distribution characteristics of the entire task process while balancing computational efficiency and data integrity. Interaction segment extraction is performed for each sliding time window, and the initiator and responder interaction records completed within that time segment are retrieved. Within each time window, the system counts the number of interactions initiated and responded by the patient within the current window and records the time interval between interaction pairs. For behavior pairs where the response delay exceeds the task-specified threshold (for example, a normal task cadence response should be completed within 3 seconds), the delay in seconds is recorded as the length of the delayed response behavior. If an interaction does not receive a response within the same time window, the behavior is marked as asynchronous response behavior.

[0023] Within a sliding time window, the following metrics are calculated for each patient: first, the cumulative number of completed interactions (including both initiations and responses); second, the cumulative duration of delayed responses; and third, the proportion of asynchronous responses in all interactions. For example, within a certain time window, Patients A and B completed eight interactions, three of which were delayed, totaling nine seconds, and two of which were out of order. Therefore, the delay duration within that window was calculated to be nine seconds, and the asynchronous response ratio was 25%.

[0024] After completing the behavioral metric statistics at the sliding time window level, an interaction response fracture index was constructed to characterize interaction quality, and a communication willingness scoring function was further constructed. The interaction response fracture index uses two core inputs: the cumulative duration of delayed responses and the proportion of asynchronous responses. To unify the weights of these two dimensions, the system normalizes these two values ​​separately, setting a maximum delay of 15 seconds and a maximum asynchronous response ratio of 1.0. After normalization, a 9-second delayed response becomes 0.6, while the asynchronous response ratio of 0.25 remains unchanged. The interaction response fracture index is defined as the weighted sum of these two normalized values. To prevent a single abnormal event from over-amplifying the fracture level, a response elasticity factor is introduced to exponentially adjust the asynchronous response term, with an exponent of 1.2. The final fracture index is expressed as the weighted sum of the two values. The weighting coefficient can be customized based on the task type and the patient's label combination quadrant, for example, giving higher weight to asynchronous behaviors in two-person collaborative tasks. We further constructed a communication willingness scoring function, which uses the cumulative number of interactions as a positive factor for the degree of active communication and the response interruption index as a negative factor. The overall function structure is: Communication willingness score = Basic activity score × (1 - Response interruption factor). The basic activity score is determined by the cumulative number of interactions per unit time and can be set to a maximum of 10, while the interruption factor has a maximum value of no more than 1. For example, if the number of interactions within a certain time window is 8 (activity score of 8 / 10) and the response interruption index is 0.62, the communication willingness score is 8 × (1 - 0.62) = 3.04.

[0025] The label map construction module maps the patient's communication willingness score to the psychological label combination space, constructs a label adaptability map and distinguishes conflicting labels from complementary labels.

[0026] Extract the coding values ​​of the two patients in the collaborative pair's psychological label combination space. This label combination space is constructed based on a three-dimensional label system, consisting of cognitive, emotional, and behavioral dimensions. Within this three-axis coordinate system, the space is divided into eight non-overlapping quadrants based on the positive and negative attributes of the scores. Each patient's psychological label combination can be uniquely mapped to a coding value in a particular quadrant. The communication willingness scores of each collaborative pair are categorized and organized according to their corresponding two sets of coding value combinations (e.g., 001-010, 001-111, etc.). That is, all scores with label combinations of 001-010 are grouped together, scores with 001-111 are grouped together, and so on, forming a score grouping table for each label combination. Statistical analysis is performed on each set of scores, and the average communication willingness score for that label combination is calculated as the compatibility indicator for that combination. If multiple scores exist for a given combination, the arithmetic mean of all scores is used. Simple quantile filtering can be used to remove outliers. In this way, a psychological label combination compatibility score matrix is ​​constructed. Each entry in the matrix represents the average fit score between a set of psychological label encoding combinations.

[0027] The resulting matrix represents the initiator label combination, with the columns representing the responder label combination. Each cross-term in the two-dimensional matrix represents the fit score for the corresponding label combination. For example, the second row and third column of the matrix (converted to binary notation and subtracted by 1) might indicate that the average communication willingness score for the 001-010 combination is 0.61, reflecting the degree of fit between the label combinations of patients coded 001 and 010.

[0028] Assuming actual collaboration between psychological label combinations, edges are constructed for all combinations with scores that appear in the scoring matrix. Each edge connects a pair of distinct label-encoded nodes, and the corresponding score values ​​serve as the edge weight information embedded in the graph structure. Edge directionality can be omitted, resulting in an undirected graph, as the compatibility of label combinations can be assumed to be symmetrical. Compatibility scoring thresholds are set, and scores are normalized to a range of 0–1. The conflict and complementarity thresholds are set at 0.4 and 0.7, respectively. Scores below 0.4 are considered conflicting label combinations, above 0.7 are considered complementary, and combinations between these thresholds are considered neutral. The thresholds are derived from prior statistical experience, such as the 25th and 75th percentiles of the overall score distribution, or can be manually pre-set based on the distribution of patients' psychological problems. The score on each edge is compared with the thresholds, and the edges are classified as conflicting, complementary, or neutral, based on the interval. These edges are then marked with different symbols or colors for subsequent analysis. In the graph, label combination pairs pointed by conflicting edges will be marked as not suitable for co-grouping, while label combinations pointed by complementary edges will be marked as collaborative recommendations.

[0029] The group reconstruction module isolates individual patients corresponding to conflicting labels and reconstructs virtual gardening collaboration groups.

[0030] In the constructed psychological label adaptability graph, the system first traverses all connected edges and retrieves label combination edges marked as conflicting relationships in the graph. Each edge marked as a conflicting relationship is marked with a unique bidirectional psychological label combination code. Based on the node codes of the conflicting label pairs, the psychological label combination codes of the individual patients currently enrolled in all virtual gardening collaboration groups are matched in sequence, and group records containing both conflicting combination codes are screened out. The individual patient numbers corresponding to the above conflicting combinations are located. To prevent the intervention analysis from being affected by historical behavior, the system retrieves the task execution records, two-person collaboration node interaction data, and collaborative operation logs of such patients in the current and previous gardening task cycles, and performs a clearing operation on their interaction behavior trajectories and task participation history. This process is completed by a unified trajectory and record stripping engine to prevent the conflicting patients' participation history from interfering with the subsequent policy training model.

[0031] On the basis of confirming the conflicting patients, in order to avoid unnecessary large-scale group reconstruction, the system further calculates the conflict label distribution density. The conflict distribution density here refers to the coverage breadth and node aggregation degree of a specific conflict label pair in the current grouping system. Taking the grouping unit as the granularity, the frequency of co-occurrence of conflict label pairs in each collaborative group is evaluated, and the frequency is sorted in descending order as a reference for isolation priority. Adopting the conflict minimization principle, a priority isolation sorting queue is constructed based on the conflict distribution density and the proportion of task participation in the collaborative group. This sorting queue prioritizes eliminating individuals who have conflicting interactions in multiple collaborative tasks at the same time and whose personal task participation is relatively low, so as to reduce the damage to the task chain after the group splits. The task participation calculation uses the product of the number of task nodes completed by the patient in the previous cycle and the proportion of task contribution as a reference indicator, and the numerical scoring result is provided by the task performance evaluation model.

[0032] After determining the priority for isolation, these patients are systematically separated from the original collaborative group and marked as unassigned. Next, tasks are restructured based on complementary tag groups that do not conflict with each other in the current system. Unassigned patients are matched to available psychological tag combinations (if no complementary tag matches are found, the patient is placed in a separate group). Reference is made to the operation records and interaction rhythms within the original task group to generate several new collaborative group combination lists.

[0033] The task performance evaluation module extracts task execution feedback data of individual patients corresponding to complementary tags under various gardening collaborative task structures to establish a task performance evaluation model.

[0034] Completed virtual gardening collaborative task execution data is structured. This task data is derived from the system's operational logs and interaction records, covering multiple stages, including task initiation, execution, interaction, and completion. To ensure indexability and assignability of the task structure, the entire operation chain is segmented by task phase (e.g., planting preparation, plant placement, maintenance interaction, and outcome verification) based on the organizational logic of the task content. Task nodes are further categorized into "independent execution nodes" and "collaborative interaction nodes" based on whether they require explicit interaction. Each segmented node is assigned a unique number, forming a standardized "task structure unit set."

[0035] After the task structure units are numbered, execution feedback data for each task node is extracted from the task logs, focusing on the following three parameters: First, task progress rate, which is the duration from task initiation to task node completion, obtained by comparing the timestamps in the task node logs; second, operation accuracy, which is the system's statistical analysis of the error rate or deviation rate of each operation, obtained by comparing the operation record with the expected operation; and third, collaborative synchronization rate, which applies only to collaborative interaction nodes and is defined as the ratio of the average time difference between each participant completing their respective tasks in that node to the total duration of the collaborative node, used to measure the degree of synchronization. Each piece of feedback data is associated with the identity of the performer. Thus, within the same data unit, the patient's single-point psychological vector is combined with the corresponding task node number and feedback parameters to form a structured "task behavior sample." To ensure sample diversity and representativeness, execution records of at least 80% of all historical node types should be collected, and the sample set should be grouped by task node number. All sample data undergoes data integrity verification, and records with missing feedback items or incomplete patient labels are eliminated to ensure a consistent and accurate data foundation for model training.

[0036] The constructed task behavior sample set is input into the task performance evaluation model training process. This model is designed as a multi-input, multi-output mapping model. Its input consists of the patient's single-point mental vector's three-dimensional coordinates (cognitive, emotional, behavioral) and the task node number. The node number and mental vector are jointly encoded using an embedded vector. The output is three response parameters: task progress rate, operation accuracy, and collaborative synchronization rate. Model training can be implemented using a multi-layer perceptron (MLP) network architecture or a gradient boosted tree (GBDT)-based model. The specific selection should be based on the sample size, dimensionality, and degree of nonlinearity. Five-fold cross-validation should be performed during model training, and an early stopping strategy should be implemented to prevent overfitting. Regarding parameter setting, an initial learning rate of 0.01 and a maximum number of training epochs of 200 should be used. Training should be terminated if the validation set loss does not decrease after 10 consecutive epochs. For model accuracy verification, target error thresholds should be set for each response parameter, for example, task progress rate error should not exceed ±10%, operation accuracy error should not exceed ±5%, and collaborative synchronization rate deviation should not exceed ±0.1. After training, the model is saved in a deployable format.

[0037] The intervention strategy configuration module constructs a task execution strategy for patients in the virtual gardening collaboration group based on the output results of the task performance evaluation model and outputs a corresponding intervention strategy configuration.

[0038] The label space and label map are updated based on the dynamic changes in the patient population's medical records. Patient record update monitoring is established in the system database. Registered medical records are periodically traversed, new patients are counted, and existing patients' cognitive, emotional, and behavioral scores are resampled and compared. If any score dimension of a patient changes beyond a set drift threshold (for example, a change in the emotional score exceeds ±1.5 points), the coordinates of that patient's single-point psychological vector are considered to have shifted. The updated single-point psychological vectors of all patients are then remapped to a three-dimensional psychological label combination space. Combinatorial logic (positive and negative polarity combinations) is constructed using the signs of each vector on the three axes, and the corresponding regions are divided according to the established eight-quadrant numbering system, for example, positive-negative-positive corresponds to quadrant 3. If a patient's coordinate change causes their quadrant to shift, it is marked as a "label quadrant shift," triggering a local reconstruction of the label adaptability map. When reconstructing the graph, according to the defined communication willingness score mapping logic, the graph structure with psychological label combinations as nodes and adaptability scores as connecting edge weights is rebuilt, and the decision boundaries of conflicting labels and complementary labels are refreshed.

[0039] After completing the collaborative grouping graph update, a pre-generation operation is performed on the task allocation strategies for the groups belonging to complementary label combinations in the current label adaptability graph. The specific method is as follows: First, the task nodes in the historical task structure unit set are counted to determine the total number of task nodes for the gardening task in the current cycle (e.g., 15 nodes). Then, based on the number of patients in the current group (e.g., 5), different task node allocation schemes are generated using permutations and combinations. For each matching combination of patient and task node, the mapping between task number and performer number is recorded, and strategy table entries are generated by numbering. Subsequently, each task node allocation strategy and the standardized psychological vectors of each patient in the group are used as joint inputs to the task performance evaluation model. This model, trained in the previous step using historical execution feedback data, is capable of predicting a patient's completion speed, accuracy, and collaborative performance at a specific task node.

[0040] The task performance evaluation model generates a corresponding predicted output for each input combination of a patient's mental vector and a task node assignment. The output includes three response parameters for each task node: task progress rate, operation accuracy, and collaborative synchronization rate. To enhance consistency in comparison, the response parameters of all strategy combinations are normalized. Using the maximum-minimum normalization method as an example, the output values ​​of each strategy under different parameters are compressed to the range [0, 1]. A weighting function is then constructed based on the system's preset parameter weights. The three parameters of each strategy are fused to generate a task performance score for that strategy. Weights are set based on task characteristics. For example, in a task that emphasizes collaboration, the collaborative synchronization rate weight can be set to 0.5, and the operation accuracy and progress rate weights can be set to 0.25, respectively. Finally, the highest-scoring candidate strategy is selected, and its matching relationship between task nodes and patients is extracted as the final intervention strategy configuration for this round of virtual gardening collaboration.

[0041] In this embodiment, the virtual gardening task environment is fully implemented using virtual reality (VR) technology. Patients participate in gardening tasks using a wearable head-mounted display (HMD) and motion-captured controllers. The system constructs an immersive virtual greenhouse space with several interactive gardening work units embedded within it, such as virtual planters, soil manipulation platforms, watering troughs, and pruning areas. These units are generated using a 3D modeling system (such as Unity3D or UnrealEngine), and an event response mechanism is implemented to record operation trajectories and timestamps. Specific interactive tasks, such as sowing, watering, fertilizing, pruning, and collaborative planting, are presented as node-based task flows. The system dynamically adjusts the task progress map based on the completion of each task node. The spatial positioning module integrated into the VR system captures the patient's hand movements and posture changes during operation. Combined with the HMD's eye tracking function, this module records the patient's response time during the task, enabling accurate collection of task node behavior data. All patient tasks are mapped in real time as system log data for access by other system modules.

[0042] The above formulas are all dimensionless and numerical calculations. The formulas are obtained by collecting a large amount of data and performing software simulation to obtain the most recent real situation. The preset parameters and thresholds in the formulas are set by technicians in this field according to actual conditions.

[0043] The above embodiments can be implemented in whole or in part via software, hardware, firmware, or any other combination. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product comprises one or more computer instructions or computer programs. When loaded or executed on a computer, the processes or functions described in the embodiments of this application are fully or partially performed. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired means (e.g., infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium accessible by a computer or a data storage device such as a server or data center that contains a collection of one or more available media. The available medium can be magnetic media (e.g., floppy disks, hard disks, tapes), optical media (e.g., DVDs), or semiconductor media. The semiconductor media can be a solid-state drive.

[0044] Those skilled in the art will appreciate that the modules and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0045] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and modules described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0046] In the several embodiments provided in this application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are merely schematic. For example, the division of the modules is only a logical function division. In actual implementation, there may be other division methods, such as multiple modules or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or modules, which can be electrical, mechanical or other forms.

[0047] The modules described as separate components may or may not be physically separate, and the components shown as modules may or may not be physical modules, and may be located in one place or distributed across multiple network modules. Some or all of the modules may be selected to achieve the purpose of this embodiment according to actual needs.

[0048] In addition, each functional module in each embodiment of the present application may be integrated into one processing module, or each module may exist physically separately, or two or more modules may be integrated into one module.

[0049] If the functions are implemented in the form of software function modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0050] The above description is merely a specific embodiment of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of this application. Therefore, the scope of protection of this application should be based on the scope of protection of the claims.

[0051] Finally: The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A social gardening intervention and evaluation system based on VR technology, characterized by: It includes a label combination space construction module, a communication willingness assessment module, a label map construction module, a grouping reconstruction module, a task performance assessment module, and an intervention strategy configuration module, among which: The label combination space construction module maps the psychological problems of the patient group into a three-dimensional psychological label system composed of emotions, cognition and behavior, and constructs a psychological label combination space; The communication willingness assessment module monitors the virtual gardening collaborative task records, collects patients' interactive behavior data in the collaborative task context based on a sliding window, and calculates the communication willingness scores between patients with different psychological label combinations; The label map construction module maps the patient's communication willingness score to the psychological label combination space, constructs the label adaptability map, and distinguishes conflicting labels from complementary labels; The group reconstruction module isolates the individual patients corresponding to the conflicting labels and reconstructs the virtual gardening collaboration groups; The task performance evaluation module extracts task execution feedback data of individual patients corresponding to complementary labels under various gardening collaborative task structures and establishes a task performance evaluation model; The intervention strategy configuration module constructs a task execution strategy for patients in the virtual gardening collaboration group based on the output results of the task performance evaluation model and outputs the corresponding intervention strategy configuration.

2. The VR-based social gardening intervention and evaluation system according to claim 1, characterized in that: The label combination space construction module maps the psychological problems of the patient group into a three-dimensional psychological label system consisting of emotions, cognition and behavior. The construction of the psychological label combination space specifically includes: Obtain psychological problem classification records from the medical files of the patient group and convert each patient's psychological problem classification record into cognitive scores, emotional scores, and behavioral scores; For each patient, a single-point psychological vector corresponding to the combination of emotion, cognition and behavior labels is established, and the label items corresponding to all single-point psychological vectors are embedded in the three-dimensional space structure to establish a psychological label combination space.

3. The VR-based social gardening intervention and evaluation system according to claim 2, characterized in that: The step of embedding the label items corresponding to the single-point psychological vector into a three-dimensional space structure and establishing a psychological label combination space specifically includes: A three-dimensional coordinate system with cognitive dimension, emotional dimension and behavioral dimension as the axis is constructed, and the three types of label items in the single-point psychological vector are embedded in the corresponding axis and the polarity direction is fixed; Performing a three-axis mapping operation on the single-point mental vectors of all patients to generate the coordinates of the single-point mental vectors of the patients in a three-dimensional coordinate system; Based on the positive and negative combinations of the three-axis coordinates, the three-dimensional psychological space is divided into eight non-overlapping quadrants, and each quadrant is numbered. Get the quadrant number to which the patient's coordinate point belongs, and mark it as the encoding value of the patient's psychological label combination.

4. The VR-based social gardening intervention and evaluation system according to claim 1, characterized in that: The communication willingness evaluation module monitors the virtual gardening collaborative task records, collects the patient's interactive behavior data in the collaborative task context based on a sliding window, and calculates the communication willingness scores between patients with different psychological label combinations. Specifically, the module includes: During the execution of the virtual gardening task, the interactive task nodes of two-person collaboration were identified, and the interaction behavior sequence was extracted according to the timestamps of the patient initiating and responding to the interaction in the task node. Define sliding time windows of equal width, extract the patient's interactive behavior segments window by window in the interactive behavior sequence, and count the cumulative number of initiating and responding interactions; In each sliding time window, the cumulative length of delayed response behaviors and the ratio of asynchronous response behaviors between patients were counted to establish an interactive response disruption index. A communication willingness scoring function based on empowerment fusion is constructed, and a communication willingness score is generated based on the cumulative number of interactions and the interaction response fracture index.

5. The VR-based social gardening intervention and evaluation system according to claim 1, characterized in that: The label map construction module maps the patient's communication willingness score to the psychological label combination space, constructs a label adaptability map, and distinguishes conflicting labels from complementary labels. Specifically, it includes: Obtain the communication willingness scores of the two-person collaborative patients in each interactive task node and extract the coding values ​​of the two patients in the psychological label combination space; The communication willingness scores were grouped into quadrants based on the coding values, and the mean of the communication willingness scores within each quadrant group was calculated to construct a psychological label group compatibility score matrix. The psychological label combinations in the adaptability score matrix are used as nodes, and the relationships between the psychological label combinations are used as edges. A psychological label adaptability graph structure is established, and the adaptability scores are marked on the edges. The conflict and complementarity determination threshold segments are set, and different psychological label combinations are divided into conflict labels and complementary labels according to the threshold segments where the adaptability scores are located.

6. The VR-based social gardening intervention and evaluation system according to claim 1, characterized in that: The group reconstruction module isolates individual patients corresponding to conflicting labels and reconstructs virtual gardening collaboration groups, specifically including: Retrieve the marked conflicting label pairs in the psychological label adaptability map and locate the corresponding patient psychological label combination code; Extract the individual patients in the current collaborative group that contain conflicting label combination codes, and remove the corresponding individual patient task participation history and interaction trajectory; Perform conflict minimization sorting based on the distribution of conflicting labels in each collaborative group to determine the priority isolation sequence of individual patients; Separate the priority isolation patients from the original collaborative grouping and construct a new virtual gardening collaborative group combination list.

7. The VR-based social gardening intervention and evaluation system according to claim 1, characterized in that: The task performance evaluation module extracts task execution feedback data of individual patients corresponding to complementary labels under various gardening collaborative task structures, and establishes a task performance evaluation model, specifically including: The complete operation chain of the completed historical virtual gardening collaborative tasks is split into process nodes according to the task stages and interaction types, and a set of numbered task structure units is constructed; In the psychological label combination space, a single-point psychological vector consisting of the cognitive score, emotional score, and behavioral score of each patient in each reconstructed virtual gardening collaboration group is extracted; Collect feedback data from the execution logs of each task node in the task structure unit set, obtain the executor of the task node, and integrate the executor's single-point psychological vector and the response parameters of the task node into a task behavior sample set; The response parameters of the task node are obtained from the feedback data in the execution log, including the task progress rate, operation accuracy and collaboration synchronization rate; The task behavior sample set is used as the training sample set, the patient's single-point psychological vector and task node number are used as input items, and the response parameters of the task node are used as output items to train the task performance evaluation model.

8. The VR-based social gardening intervention and evaluation system according to claim 1, characterized in that: The intervention strategy configuration module constructs a task execution strategy for patients in the virtual gardening collaboration group based on the output of the task performance evaluation model, and the output corresponding intervention strategy configuration specifically includes: Regularly update the medical records of the patient group. When the number of patients increases or the coordinates of the single-point psychological vector in the psychological label combination space change, update the label adaptability map and reconstruct the virtual gardening collaboration grouping; In each virtual gardening collaboration group, a task node allocation strategy table for all patients performing tasks is pre-generated, and the patient's single-point mental vector and task node allocation strategy are input into the task performance evaluation model; The output results of the task performance evaluation model are normalized and converted into task performance scores through weighted fusion. The task node allocation strategy corresponding to the highest task performance score is selected as the intervention strategy configuration.

Images