Mrep pattern-based autism child rehabilitation evaluation method, system and device

Abstract

The application belongs to the technical field of autism children rehabilitation evaluation, and discloses an autism children rehabilitation evaluation method, system and device based on an MREP mode, which converts the performance of children on target evaluation projects into quantitative feature data through structured behavior quantitative coding, calculates the abnormality of children in four dimensions of medical treatment, rehabilitation, education and family based on the data, screens out a primary intervention target set in combination with development lag judgment, automatically matches a structured intervention scheme for each target and sorts and executes the structured intervention scheme by querying a preset evidence-based intervention mapping matrix, collects process data of institutions and families within an intervention period, and calculates a progress index of each target according to the process data, and obtains a comprehensive evaluation value by weighting the average mastery degree of each dimension target and the overall target achievement rate. The application realizes the full-process data of collection, intervention and effect evaluation, and improves the accuracy, scientificity and efficiency of rehabilitation services.

Description

Technical Field

[0001] This invention belongs to the field of rehabilitation assessment technology for children with autism, specifically involving a method, system, and device for rehabilitation assessment of children with autism based on the MREP model. Background Technology

[0002] Rehabilitation intervention for children with autism is a complex system engineering project that requires the integration of medical treatment, rehabilitation, education, and family care (i.e., the MREP model).

[0003] Current problems in practice include: the assessment process relies heavily on the therapist's subjective experience, resulting in insufficient quantification of children's behavior and a blurred baseline competency profile; when developing intervention plans based on assessment results, there is a lack of standardized tools that automatically and accurately match specific competency gaps with evidence-based intervention methods, making it difficult to guarantee the personalization and scientific rigor of intervention strategies; and the evaluation of intervention effects often remains at a macro-level description or single dimension, lacking a quantitative indicator that can comprehensively reflect children's competency progress in multiple dimensions and scenarios, making it difficult to accurately monitor and dynamically optimize the rehabilitation process. Summary of the Invention

[0004] The purpose of this invention is to provide a method, system, and device for rehabilitation assessment of children with autism based on the MREP model.

[0005] The rehabilitation assessment method for children with autism based on the MREP model includes the following steps: S1. Obtain video streams and labeled events of interactive activities of children with autism. For each target assessment item in the interactive activities, perform structured behavior quantitative coding based on the associated video streams and labeled events to generate quantitative feature data for each target assessment item. S2. Based on the quantitative feature data, calculate the dimensional anomaly scores of each of the four dimensions of medical care, rehabilitation, education, and family, and assign weight coefficients to the four dimensions. Based on the dimensional anomaly scores and weight coefficients, perform a weighted calculation to obtain the comprehensive urgency index of each quantitative feature data. Compare the comprehensive urgency index with a preset threshold, and select the quantitative feature data corresponding to the comprehensive urgency index that is greater than the preset threshold to form a primary target set. At the same time, mark the target evaluation items in the primary target set that meet the development lag conditions as development lag. S3. For each quantitative feature data in the primary target set, match and generate the corresponding structured intervention tuple by querying the preset evidence-based intervention mapping matrix. Sort all structured intervention tuples according to the comprehensive urgency index and development lag marker to generate an intervention execution sequence. S4. Based on the intervention execution sequence, recommend intervention training and corresponding intervention cycles. In each intervention training, record the corresponding quantitative feature data as process quantitative data. At the same time, collect family scenario generalization data corresponding to the process quantitative data. S5. At the end of the intervention training, based on all process quantitative data and family scenario generalization data, calculate the individual estimated indicators of process quantitative data and family scenario generalization data. The proportion of individual estimated indicators greater than the indicator threshold in all intervention cycles is used as the overall goal achievement rate. The average cycle mastery of quantitative characteristic data belonging to the four dimensions of medical care, rehabilitation, education, and family ecology is aggregated and calculated separately. S6. Based on the overall goal achievement rate and the average cycle mastery, a comprehensive evaluation value is generated by weighted summation.

[0006] In S2, target evaluation items in the primary target set that meet the development lag condition are marked as development lag items, specifically as follows: Based on the quantitative characteristic data, the reference developmental age is obtained, and the actual physiological age of the child to be evaluated is obtained. Development lag condition 1: Based on quantitative characteristic data, the target evaluation project is marked as passing independently; Condition 2 for developmental lag: Reference developmental age < actual physiological age; If both development lag condition 1 and development lag condition 2 are true, the number of months of development lag is calculated based on the difference between the actual physiological age and the reference development age. This number is then compared with the preset development lag threshold. If the number of months of development lag is greater than or equal to the development lag threshold, a development lag mark is made for the corresponding target evaluation item.

[0007] In S2, based on quantified feature data, the dimensional anomaly scores are calculated for each of the four dimensions: medical care, rehabilitation, education, and family. Specifically: Medical dimension: Obtain the norm mean and standard deviation of key indicators from quantitative feature data, calculate the standardized score, and calculate the absolute deviation as the medical dimension anomaly score; Rehabilitation dimension: Compare quantitative feature data with behavioral standards to determine its status, and perform dimension mapping based on the status to obtain the abnormality score of rehabilitation dimension; Education dimension: The sitting time feature is obtained from the quantitative feature data, normalized, and the abnormality score of the education dimension is obtained; Family dimension: Generalized data of family scenarios are obtained from quantitative feature data, mapped, and the family dimension anomaly score is obtained.

[0008] In S3, all structured intervention tuples are sorted according to the comprehensive urgency index and development lag markers to generate an intervention execution sequence, specifically: Structured intervention tuples with developmental lag markers are designated as the first group, and structured intervention tuples without developmental lag markers are designated as the second group. Within each group, they are sorted from high to low according to the comprehensive urgency index. All structured intervention tuples after sorting are used as the intervention execution sequence with the first group first and the second group last.

[0009] In S3, each quantitative feature data in the primary target set is matched and generated into a corresponding structured intervention tuple by querying a preset evidence-based intervention mapping matrix. Specifically, the quantitative feature data is used as the key to query the evidence-based intervention mapping matrix, obtain all field data of the corresponding row in the evidence-based intervention mapping matrix, and assemble these field data into a structured intervention tuple according to a preset format.

[0010] In S1, an event is a key action start and end point or state determination marked on the video timeline through the system interface during the viewing of the video stream. An event annotation must include at least: event type, occurrence time, duration, and action description.

[0011] The structured behavior quantification coding in S1 includes: response latency, behavior frequency, proportion of correct responses, behavior duration, cue level, and status determination.

[0012] S5 calculates separate estimated indicators for the process quantitative data and the family scenario generalized data based on all process quantitative data and family scenario generalized data. Specifically, the ratio of the number of correct responses to the total number of attempts in the process quantitative data is used as the first indicator, the ratio of the number of complete check-ins to the number of check-ins that should be checked in in the family scenario generalized data is used as the second indicator, and the ratio of the number of times the child to be assessed independently completes the target assessment item to the total number of times in the process quantitative data is used as the third indicator. The first, second, and third indicators are weighted and summed to obtain the separate estimated indicator.

[0013] A rehabilitation assessment system for children with autism based on the MREP model is used to implement a rehabilitation assessment method for children with autism based on the MREP model, including: The data acquisition module acquires video streams and labeled events of interactive activities of children with autism. For each target assessment item in the interactive activities, it performs structured behavior quantitative coding based on the associated video streams and labeled events to generate quantitative feature data for each target assessment item. The labeling module calculates the dimensional anomaly scores of quantitative feature data in four dimensions: medical care, rehabilitation, education, and family, and assigns weight coefficients to the four dimensions. Based on the dimensional anomaly scores and weight coefficients, a weighted calculation is performed to obtain the comprehensive urgency index of each quantitative feature data. The comprehensive urgency index is compared with a preset threshold, and the quantitative feature data corresponding to the comprehensive urgency index that is greater than the preset threshold are selected to form a primary target set. At the same time, the target evaluation items in the primary target set that meet the development lag conditions are marked as development lag. The intervention sequence generation module takes each quantitative feature data in the primary target set, queries a preset evidence-based intervention mapping matrix, matches and generates a corresponding structured intervention tuple, sorts all structured intervention tuples according to the comprehensive urgency index and development lag marker, and generates an intervention execution sequence. The process data acquisition module recommends intervention training and corresponding intervention cycles based on the intervention execution sequence. In each intervention training, it records the corresponding quantitative feature data as process quantitative data. At the same time, it collects family scenario generalization data corresponding to the process quantitative data. The average cycle mastery acquisition module calculates individual estimated indicators for the process quantitative data and family scenario generalized data based on all process quantitative data and family scenario generalized data at the end of the intervention training. The proportion of individual estimated indicators greater than the indicator threshold in all intervention cycles is used as the overall goal achievement rate. The module also aggregates and calculates the average cycle mastery of quantitative characteristic data belonging to the four dimensions of medical care, rehabilitation, education, and family ecology. The evaluation module generates a comprehensive evaluation value by weighted summation based on the overall target achievement rate and average periodic mastery.

[0014] A rehabilitation assessment device for autism children based on the MREP model includes a processor and a memory, wherein the processor executes a computer program stored in the memory to implement a rehabilitation assessment method for autism children based on the MREP model.

[0015] Compared with the prior art, the technical solution provided in this application has at least the following beneficial effects: By transforming traditional assessment items into computable structured behavioral codes and quantitative feature data, the problem of assessment relying on subjective experience is solved, and a reliable data foundation is established for subsequent accurate analysis.

[0016] Deterministic target selection and prioritization can automatically identify core intervention targets from assessment data and match them with structured intervention plans generated from evidence-based practices, significantly improving the scientific rigor and efficiency of intervention strategy development.

[0017] This study achieves a multi-dimensional, comprehensive, and quantitative assessment of the rehabilitation process. It proposes and calculates separate estimation indicators for the effectiveness of integrated training, family generalization, and skill independence. These indicators are then aggregated to generate average periodic mastery and overall goal achievement rates, reflecting progress across four dimensions: medical care, rehabilitation, education, and family environment. The final comprehensive assessment value is obtained. This comprehensive and quantitative assessment reflects the overall rehabilitation effectiveness of children, providing clear and objective data for adjusting rehabilitation strategies. Detailed Implementation

[0018] To further understand the content of this invention, the invention will be described in detail with reference to the embodiments.

[0019] This invention relates to a rehabilitation assessment method for children with autism based on the MREP model, which includes the following steps: S1. Obtain video streams and labeled events of interactive activities of children with autism. For each target assessment item in the interactive activities, perform structured behavior quantification coding based on the associated video streams and labeled events to generate quantitative feature data for each target assessment item.

[0020] Specifically, a video stream refers to a video recording made by a camera device that contains continuous behavioral performances of a child with autism or their parents during standardized interactive activities. This video stream records the complete time sequence of the interactive activities, including all audiovisual information such as the child's facial expressions, eye gaze direction, body movements, speech, and the stimuli presented by the assessor. The video stream serves as the raw data foundation for subsequent quantitative coding of behavior.

[0021] Annotated events refer to the start and end points of key behaviors or state determinations marked on the video timeline through the system interface during video streaming. Annotated events must include at least the following elements: event type (e.g., command issuance, eye contact, correct reaction), occurrence time, duration, and behavior description. Annotated events represent a preliminary structured processing of the raw video stream, transforming continuous unstructured video into a discrete, computable sequence of events.

[0022] A target assessment item is the smallest, indivisible behavioral observation unit in the rehabilitation assessment system. Each item corresponds to a specific, observable, and measurable behavioral skill or developmental indicator. Each target assessment item is predefined with the following attributes: item ID, behavioral description, passing criteria, reference age, domain, and MREP dimension.

[0023] Furthermore, the relevant domains include, for example, perception, gross motor skills, fine motor skills, language and communication, cognition, social interaction, self-care, and emotions and behavior. The entire evaluation process consists of a standardized evaluation item library comprised of several such target evaluation items.

[0024] Structured behavior quantification coding refers to the process of analyzing, measuring, and judging video streams and labeled events based on predefined behavioral definitions and success criteria for each target assessment item. This process includes the following sub-operations: Event extraction, which involves extracting all timestamps and behavioral records related to the current target assessment item from labeled events; Parameter calculation, which calculates specified quantification parameters according to the item type, including: Response latency: the time difference between stimulus presentation and the occurrence of the target behavior; Behavior frequency: the number of times the target behavior occurs per unit time; Correct response ratio: the ratio of the number of behaviors that meet the success criteria to the total number of attempts; Behavior duration: the total duration of a single target behavior; Cue level: the minimum level of assistance required for the child to achieve a correct response; Status determination: comparing the calculated quantification parameters with the predefined pass criteria for the item to determine the current status of the item as pass, intermediate response, or fail.

[0025] Furthermore, intermediate response refers to a transitional state in which a child, although failing to pass the structured behavior quantification coding process, demonstrates a clear intention to perform the target behavior or is able to partially complete it under assisted conditions. This includes clear intention but failure to execute, partial completion under assisted conditions, and quantification data falling into the intermediate threshold range.

[0026] Quantitative feature data is the output of structured behavioral quantification coding. It is multi-field, fully numerically structured data for a single target evaluation item. For example, for a target evaluation item "Understanding common item names: apples", its quantitative feature data might be: {Project ID: 25, Approval Status: Passed, Hint Level: Verbal Hint, Associated Video Clip: 5 minutes 34 seconds - 6 minutes 28 seconds....}

[0027] Video streams are unstructured continuous signals. Although event labeling divides the video stream into discrete time-point markers, these are all event sequences rather than state metrics. Subsequent steps require comparable, aggregateable, weighted, and thresholdable numerical data, which cannot be performed on binary states or video segments. Therefore, it is necessary to transform human-understandable observational phenomena into machine-computable mathematical objects and normalize heterogeneous data into quantifiable feature data.

[0028] S2. Based on the quantitative feature data, calculate the dimensional anomaly scores for each of the four dimensions: medical care, rehabilitation, education, and family. Assign weight coefficients to the four dimensions. Based on the dimensional anomaly scores and weight coefficients, perform a weighted calculation to obtain the comprehensive urgency index for each quantitative feature data. Compare the comprehensive urgency index with a preset threshold and select the quantitative feature data corresponding to the comprehensive urgency index that is greater than the preset threshold to form a primary target set. At the same time, mark the target evaluation items in the primary target set that meet the development lag conditions as development lag.

[0029] Based on the quantitative feature data, the dimensional anomaly scores were calculated for each of the four dimensions: medical care, rehabilitation, education, and family. Medical dimension: Obtain the norm mean and standard deviation of key indicators from quantitative feature data, calculate the standardized score, and calculate the absolute deviation as the medical dimension anomaly score; Furthermore, in this dimension, key indicators refer to core observations derived from internationally recognized diagnostic criteria (such as DSM-5) and diagnostic tools (such as ADOS and ADI-R). In this application, these key indicators correspond to quantitative characteristic data.

[0030] The norm mean and standard deviation are derived from a pre-stored external norm database. This database was established through standardized behavioral observation and data collection from a large-scale group of typical developmental children, followed by statistical analysis.

[0031] The standardized score is the difference between a data point in a quantitative feature dataset and the mean of a norm, divided by the standard deviation. The ratio of the standardized score to the standardized score threshold is the absolute deviation.

[0032] Rehabilitation dimension: Compare quantitative feature data with behavioral standards to determine its status, and perform dimension mapping based on the status to obtain the abnormality score of rehabilitation dimension; Education dimension: The sitting time feature is obtained from the quantitative feature data, normalized, and the abnormality score of the education dimension is obtained; Family dimension: Generalized data of family scenarios are obtained from quantitative feature data, mapped, and the family dimension anomaly score is obtained.

[0033] In traditional practice, the MREP model has remained at the level of theoretical advocacy, lacking operational tools. This application, by assigning weight coefficients to four dimensions for each assessment item, mathematizes this multidimensional intervention concept for the first time, moving beyond general discussions to explicit expression through weight coefficients: weight coefficients are assigned to the abnormality scores of each dimension, and a weighted calculation is performed based on the abnormality scores and weight coefficients to obtain the comprehensive urgency index of each quantitative feature data. The comprehensive urgency index is compared with a preset threshold, and the quantitative feature data corresponding to the comprehensive urgency index greater than the preset threshold are selected to form a primary target set. This is because the fundamental contradiction in autism rehabilitation is that there are too many intervention targets, while the daily intervention time is limited. Therefore, by using threshold screening to form a primary target set, scarce rehabilitation resources are allocated to the most critical intervention points to maximize intervention benefits.

[0034] At the same time, target assessment items in the primary target set that meet the development lag conditions are marked as development lag items, specifically: Based on the quantitative characteristic data, the reference developmental age is obtained, and the actual physiological age of the child to be evaluated is obtained. Development lag condition 1: Based on quantitative characteristic data, the target evaluation project is marked as passing independently; Condition 2 for developmental lag: Reference developmental age < actual physiological age; If both developmental lag condition 1 and developmental lag condition 2 are true, the number of months of developmental lag is calculated based on the difference between the actual physiological age and the reference developmental age. This number is then compared with the preset developmental lag threshold. If the number of months of developmental lag is greater than or equal to the developmental lag threshold, the corresponding target assessment item is marked as developmentally lagable to identify hidden lag targets and fill the blind spots of traditional assessment.

[0035] S3. For each quantitative feature data in the primary target set, match and generate the corresponding structured intervention tuple by querying the preset evidence-based intervention mapping matrix. Sort all structured intervention tuples according to the comprehensive urgency index and development lag marker to generate an intervention execution sequence.

[0036] Specifically, using quantitative feature data as the key, a query is performed in the evidence-based intervention mapping matrix to obtain all field data of the corresponding row in the evidence-based intervention mapping matrix, and these field data are assembled into a structured intervention tuple according to a preset format.

[0037] Structured intervention tuples with developmental lag labels are designated as the first group, and structured intervention tuples without developmental lag labels are designated as the second group. Within each group, they are sorted from high to low according to the comprehensive urgency index. All sorted structured intervention tuples are then connected in the order of the first group first and the second group last to form the intervention execution sequence. Tuples ranked earlier receive more training opportunities in the daily intervention time allocation, while tuples ranked later can have their training density appropriately reduced or be delayed until the next cycle.

[0038] S4. Based on the intervention execution sequence, recommend intervention training and corresponding intervention cycles. In each intervention training, record the corresponding quantitative feature data as process quantitative data. At the same time, collect the family scenario generalization data corresponding to the process quantitative data.

[0039] Intervention training recommendations include the following, which should be analyzed in conjunction with specific medical advice. This application is for reference only: Applied behavior analysis intervention, driven by data, employing strategies such as positive reinforcement, task analysis, and desensitization training, primarily used to cultivate basic learning abilities, social skills, and self-care abilities; Speech therapy, targeting language expression and comprehension disorders, utilizing gamification, scenario simulation, and technology to enhance language skills; Sensory integration training, targeting deficiencies in touch, vestibular sense, and proprioception, improving children's concentration and motor coordination through multi-sensory training; Social skills training, using group interaction, role-playing, and social stories to enhance children's ability to identify emotions, respond, and solve problems in context; Gamified interactive intervention, promoting self-expression of language and social communication abilities through engaging scenarios in natural game interactions; Natural developmental behavior intervention, combining developmental psychology with behavioral intervention methods, conducting teaching in natural environments, exemplified by the "Early Start Denver Model," enhancing children's social interaction, language, and cognitive abilities through games, daily interactions, and scenario tasks, achieving skill generalization and autonomous application.

[0040] In addition, it includes training to assist in communication interventions, such as picture-exchange communication, which is suitable for children with limited language expression and helps them communicate through pictures; visual support and structured environments, which use schedules, work systems and visual cues to enhance children's understanding and adaptability to daily activities; peer intervention, which promotes social interaction and skills learning through demonstration and collaboration with peers; and parental intervention, which trains parents to implement scientific intervention strategies in daily family life to ensure the consistency and continuity of intervention measures.

[0041] S5. At the end of the intervention training, based on all process quantitative data and family scenario generalization data, calculate the individual estimated indicators of process quantitative data and family scenario generalization data. The proportion of individual estimated indicators greater than the indicator threshold in all intervention cycles is used as the overall goal achievement rate. The average cycle mastery of quantitative characteristic data belonging to the four dimensions of medical care, rehabilitation, education, and family ecology is calculated separately.

[0042] S6. Based on the overall target achievement rate and average cycle mastery, a comprehensive evaluation value is generated by weighted summation.

[0043] Based on all process-quantified data and family-scenario generalized data, separate estimated indicators for process-quantified data and family-scenario generalized data are calculated. Specifically, the ratio of the number of correct responses to the total number of attempts in the process-quantified data is used as the first indicator; the ratio of the number of complete check-ins to the number of check-ins that should have been completed in the family-scenario generalized data is used as the second indicator; and the ratio of the number of times the child to be assessed independently completed the target assessment item to the total number of times in the process-quantified data is used as the third indicator. The first, second, and third indicators are weighted and summed to obtain the separate estimated indicator.

[0044] The percentage of individual estimated indicators exceeding the indicator threshold in all intervention cycles is used as the overall target achievement rate.

[0045] The average periodic mastery of quantitative feature data belonging to the four dimensions of medical care, rehabilitation, education, and family ecology is calculated separately. The aggregation process is as follows: Before aggregation, it is ensured that the quantitative data of each process is clearly associated with a target assessment item. In step S2, each target assessment item has been identified as belonging to the MREP dimension (medical care, rehabilitation, education, family ecology) in the form of quantitative feature data. In this application, a target assessment item may belong to one or more dimensions.

[0046] For each dimension, select all target evaluation items marked as belonging to the dimension from the primary target set to form a subset. Determine whether the dimension label set of each target evaluation item contains the dimension. If it does, the target evaluation item belongs to the dimension. Multiply all target evaluation items of each dimension by their corresponding individual estimation indicators, sum them, and divide by the number of target evaluation items in the dimension to obtain the average periodic mastery.

[0047] Projects scattered across different fields and of different types will be regrouped and aggregated according to the MREP four-dimensional framework to form rehabilitation assessment data for children with autism.

[0048] Based on the overall goal achievement rate and average periodic mastery, a weighted sum is obtained to obtain a comprehensive assessment value, which is then used to conduct rehabilitation assessments for children with autism.

[0049] This application, through S1 to S6, completes the entire process from raw observation to quantified decision-making and evaluation. It transforms incalculable behavioral videos into calculable structured data, realizing the datafication of observation; S2 integrates multi-dimensional capability deficiencies into a single urgency index, completing the transformation from description to decision-making; S3 maps and prioritizes the intervention targets into evidence-based intervention prescriptions, realizing the transformation from decision-making to action; S4 continuously collects process data during implementation, constructing the recordability of actions; S5 refines fragmented process data into individual project progress and multi-dimensional aggregated indicators, completing the sublimation from recording to evidence; and S6 compresses multi-dimensional effectiveness evidence into a comprehensive evaluation value, achieving the final closed loop.

[0050] A rehabilitation assessment system for autism children based on the MREP model is used to implement the aforementioned rehabilitation assessment method for autism children based on the MREP model, including: The data acquisition module acquires video streams and labeled events of interactive activities of children with autism. For each target assessment item in the interactive activities, it performs structured behavior quantitative coding based on the associated video streams and labeled events to generate quantitative feature data for each target assessment item. The labeling module calculates the dimensional anomaly scores of quantitative feature data in four dimensions: medical care, rehabilitation, education, and family, and assigns weight coefficients to the four dimensions. Based on the dimensional anomaly scores and weight coefficients, a weighted calculation is performed to obtain the comprehensive urgency index of each quantitative feature data. The comprehensive urgency index is compared with a preset threshold, and the quantitative feature data corresponding to the comprehensive urgency index that is greater than the preset threshold are selected to form a primary target set. At the same time, the target evaluation items in the primary target set that meet the development lag conditions are marked as development lag. The intervention sequence generation module takes each quantitative feature data in the primary target set, queries a preset evidence-based intervention mapping matrix, matches and generates a corresponding structured intervention tuple, sorts all structured intervention tuples according to the comprehensive urgency index and development lag marker, and generates an intervention execution sequence. The process data acquisition module recommends intervention training and corresponding intervention cycles based on the intervention execution sequence. In each intervention training, it records the corresponding quantitative feature data as process quantitative data. At the same time, it collects family scenario generalization data corresponding to the process quantitative data. The average cycle mastery acquisition module calculates individual estimated indicators for the process quantitative data and family scenario generalized data based on all process quantitative data and family scenario generalized data at the end of the intervention training. The proportion of individual estimated indicators greater than the indicator threshold in all intervention cycles is used as the overall goal achievement rate. The module also aggregates and calculates the average cycle mastery of quantitative characteristic data belonging to the four dimensions of medical care, rehabilitation, education, and family ecology. The evaluation module generates a comprehensive evaluation value by weighted summation based on the overall target achievement rate and average periodic mastery.

[0051] A rehabilitation assessment device for autism children based on the MREP model includes a processor and a memory, wherein the processor executes a computer program stored in the memory to implement a rehabilitation assessment method for autism children based on the MREP model.

Claims

1. A rehabilitation assessment method for children with autism based on the MREP model, characterized in that, Includes the following steps: S1. Obtain video streams and labeled events of interactive activities of children with autism. For each target assessment item in the interactive activities, perform structured behavior quantitative coding based on the associated video streams and labeled events to generate quantitative feature data for each target assessment item. S2. Based on the quantitative feature data, calculate the dimensional anomaly scores for each of the four dimensions: medical care, rehabilitation, education, and family. Assign weight coefficients to the four dimensions. Based on the dimensional anomaly scores and weight coefficients, perform a weighted calculation to obtain the comprehensive urgency index for each quantitative feature data. Compare the comprehensive urgency index with a preset threshold and select the quantitative feature data corresponding to the comprehensive urgency index that is greater than the preset threshold to form a primary target set. At the same time, mark the target assessment items in the primary target set that meet the development lag condition as development lag. S3. For each quantitative feature data in the primary target set, match and generate the corresponding structured intervention tuple by querying the preset evidence-based intervention mapping matrix. Sort all structured intervention tuples according to the comprehensive urgency index and development lag mark to generate an intervention execution sequence. S4. Based on the intervention execution sequence, recommend intervention training and the corresponding intervention cycle. In each intervention training, record the corresponding quantitative feature data as process quantitative data. At the same time, collect family scenario generalization data corresponding to the process quantitative data. S5. At the end of the intervention training, based on all process quantitative data and family scenario generalization data, calculate the individual estimated indicators of process quantitative data and family scenario generalization data. The proportion of individual estimated indicators greater than the indicator threshold in all intervention cycles is used as the overall goal achievement rate. The average cycle mastery of quantitative characteristic data belonging to the four dimensions of medical care, rehabilitation, education, and family ecology is aggregated and calculated separately. S6. Based on the overall goal achievement rate and the average cycle mastery, a comprehensive evaluation value is generated by weighted summation.

2. The rehabilitation assessment method for autistic children based on the MREP model according to claim 1, characterized in that, In S2, target evaluation items in the primary target set that meet the development lag condition are marked as development lag items, specifically as follows: Based on the quantitative characteristic data, the reference developmental age is obtained, and the actual physiological age of the child to be evaluated is obtained. Development lag condition 1: Based on quantitative characteristic data, the target evaluation project is marked as passing independently; Condition 2 for developmental lag: Reference developmental age < actual physiological age; If both development lag condition 1 and development lag condition 2 are true, the number of months of development lag is calculated based on the difference between the actual physiological age and the reference development age. This number is then compared with the preset development lag threshold. If the number of months of development lag is greater than or equal to the development lag threshold, a development lag mark is made for the corresponding target evaluation item.

3. The rehabilitation assessment method for autistic children based on the MREP model according to claim 1, characterized in that, In S2, based on quantified feature data, the dimensional anomaly scores are calculated for each of the four dimensions: medical care, rehabilitation, education, and family. Specifically: Medical dimension: Obtain the norm mean and standard deviation of key indicators from quantitative feature data, calculate the standardized score, and calculate the absolute deviation as the medical dimension anomaly score; Rehabilitation dimension: Compare quantitative feature data with behavioral standards to determine its status, and perform dimension mapping based on the status to obtain the abnormality score of rehabilitation dimension; Education dimension: The sitting time feature is obtained from the quantitative feature data, normalized, and the abnormality score of the education dimension is obtained; Family dimension: Generalized data of family scenarios are obtained from quantitative feature data, mapped, and the family dimension anomaly score is obtained.

4. The method for rehabilitation assessment of autistic children based on the MREP model according to claim 1, characterized in that, In S3, all structured intervention tuples are sorted according to the comprehensive urgency index and development lag markers to generate an intervention execution sequence, specifically: Structured intervention tuples with developmental lag markers are designated as the first group, and structured intervention tuples without developmental lag markers are designated as the second group. Within each group, they are sorted from high to low according to the comprehensive urgency index. All sorted structured intervention tuples are then connected in the order of the first group first and the second group last to form the intervention execution sequence.

5. The rehabilitation assessment method for autistic children based on the MREP model according to claim 1, characterized in that, In S3, each quantitative feature data in the primary target set is matched and generated into a corresponding structured intervention tuple by querying a preset evidence-based intervention mapping matrix. Specifically, the quantitative feature data is used as the key to query the evidence-based intervention mapping matrix, obtain all field data of the corresponding row in the evidence-based intervention mapping matrix, and assemble these field data into a structured intervention tuple according to a preset format.

6. The method for rehabilitation assessment of autistic children based on the MREP model according to claim 1, characterized in that, In S1, an event is a key action start and end point or state determination marked on the video timeline through the system interface during the viewing of the video stream. An event annotation must include at least: event type, occurrence time, duration, and action description.

7. The method for rehabilitation assessment of autistic children based on the MREP model according to claim 1, characterized in that, The structured behavior quantification coding in S1 includes: response latency, behavior frequency, proportion of correct responses, behavior duration, cue level, and status determination.

8. The rehabilitation assessment method for autistic children based on the MREP model according to claim 1, characterized in that, S5 calculates separate estimated indicators for the process quantitative data and the family scenario generalized data based on all process quantitative data and family scenario generalized data. Specifically, the ratio of the number of correct responses to the total number of attempts in the process quantitative data is used as the first indicator, the ratio of the number of complete check-ins to the number of check-ins that should be checked in in the family scenario generalized data is used as the second indicator, and the ratio of the number of times the child to be assessed independently completes the target assessment item to the total number of times in the process quantitative data is used as the third indicator. The first, second, and third indicators are weighted and summed to obtain the separate estimated indicator.

9. A rehabilitation assessment system for autistic children based on the MREP model, used to implement the rehabilitation assessment method for autistic children based on the MREP model as described in any one of claims 1-8, characterized in that, include: The data acquisition module acquires video streams and labeled events of interactive activities of children with autism. For each target assessment item in the interactive activities, it performs structured behavior quantitative coding based on the associated video streams and labeled events to generate quantitative feature data for each target assessment item. The labeling module calculates the dimensional anomaly scores of quantitative feature data in four dimensions: medical care, rehabilitation, education, and family, and assigns weight coefficients to the four dimensions. Based on the dimensional anomaly scores and weight coefficients, a weighted calculation is performed to obtain the comprehensive urgency index of each quantitative feature data. The comprehensive urgency index is compared with a preset threshold, and the quantitative feature data corresponding to the comprehensive urgency index that is greater than the preset threshold are selected to form a primary target set. At the same time, the target evaluation items in the primary target set that meet the development lag conditions are marked as development lag. The intervention sequence generation module takes each quantitative feature data in the primary target set, queries a preset evidence-based intervention mapping matrix, matches and generates a corresponding structured intervention tuple, sorts all structured intervention tuples according to the comprehensive urgency index and development lag marker, and generates an intervention execution sequence. The process data acquisition module recommends intervention training and corresponding intervention cycles based on the intervention execution sequence. In each intervention training, it records the corresponding quantitative feature data as process quantitative data. At the same time, it collects family scenario generalization data corresponding to the process quantitative data. The average cycle mastery acquisition module calculates individual estimated indicators for the process quantitative data and family scenario generalized data based on all process quantitative data and family scenario generalized data at the end of the intervention training. The proportion of individual estimated indicators greater than the indicator threshold in all intervention cycles is used as the overall goal achievement rate. The module also aggregates and calculates the average cycle mastery of quantitative characteristic data belonging to the four dimensions of medical care, rehabilitation, education, and family ecology. The evaluation module generates a comprehensive evaluation value by weighted summation based on the overall target achievement rate and average periodic mastery.

10. A rehabilitation assessment device for autistic children based on the MREP model, characterized in that, It includes a processor and a memory, wherein the processor executes a computer program stored in the memory to implement the rehabilitation assessment method for autistic children based on the MREP model as described in any one of claims 1-8.