Design and development guidance method for teaching toys for children with mental disorder based on big data

Abstract

The invention relates to the technical field of big data analysis, and provides a mental disorder child teaching toy design and development guidance method based on big data, and the method comprises the steps: carrying out the quantification of multi-dimensional demand indexes of a child, calculating the weight of each demand index through an analytic hierarchy process, setting a dynamic intervention target vector, and collecting the behavior data of the child in real time. Meanwhile, past learning records and diagnosis report information of children are collected, key feature data are extracted in a targeted mode according to demand weights, a combined state evaluation model is constructed, a comprehensive state score is calculated based on state distribution output by the model, and then the comprehensive state score is compared with a dynamic intervention target for analysis; dynamically generating an adjustment strategy according to the output comprehensive state score and the determined dynamic intervention target; according to the method, the evaluation model is constructed by combining the acquired data with the analytic hierarchy process, the strategy is dynamically adjusted, social guidance and multi-party feedback are considered at the same time, and dynamic optimization and man-machine interaction balance of the teaching toy are realized.

Description

Technical Field

[0001] The present invention relates to the technical field of big data analysis, and more specifically, to a method for guiding the design and development of teaching toys for children with mental disorders based on big data. Background Art

[0002] With the rapid development of big data, artificial intelligence (AI), and Internet of Things technologies, the field of special education is undergoing a transformation from traditional teaching aids to intelligent and personalized intervention tools; the maturity of time series modeling and multi-modal data fusion in big data analysis technology makes it possible to capture children's behavioral characteristics in real time, and the combination of embedded sensors and lightweight algorithms provides a hardware foundation for low-latency and high-precision interactive feedback.

[0003] As the proportion of children with mental disorders increases globally year by year (statistics from the WHO show that the incidence of autism reaches 1 / 54), traditional "one-size-fits-all" teaching aids are difficult to meet the needs of differentiated intervention. Cognitive psychology research indicates that early dynamic adaptive training can significantly improve the social intention recognition ability of children with autism. Therefore, despite significant technological progress, current teaching toys for children with mental disorders still have the following deficiencies: 1. The disconnection between dynamic adaptability and data closed-loop: Existing products mostly rely on static preset algorithms such as jigsaw puzzles with fixed difficulty levels, lacking the ability to dynamically optimize based on real-time behavior data, resulting in limited intervention effects. For example, children with autism may need to reduce the task complexity and enhance tactile feedback in an anxious state, but static systems cannot respond to such needs in real time, which instead exacerbates frustration.

[0004] 2. The paradox of human-computer interaction and social deprivation: Over-reliance on teaching toys may reduce the interaction opportunities between children and parents or teachers, and social skills training is exactly one of the core goals of mental disorder intervention. For children who use teaching toys for a long time, their joint attention ability may decrease. To avoid social deprivation and reduce the participation in teaching toys, the situation of reverting to the traditional inefficient teaching aid mode may occur. Summary of the Invention

[0005] In order to solve the above technical problems, the present invention provides a method for guiding the design and development of teaching toys for children with mental disorders based on big data. By collecting data, combining the analytic hierarchy process to construct an evaluation model and dynamically adjusting strategies, while taking into account social guidance and multi-party feedback, it realizes dynamic optimization and the balance of human-machine and interpersonal interactions to solve the problems in the prior art.

[0006] A method for guiding the design and development of teaching toys for children with mental disorders based on big data includes: Step 1, Requirement Analysis and Goal Definition: Quantify the requirement indicators in the three dimensions of children's cognition, social interaction, and emotion, calculate the weights of each requirement indicator using the Analytic Hierarchy Process, and set the dynamic intervention goal vector simultaneously. Step 2, Multimodal Data Collection: Deploy multiple sensors on teaching toys according to the quantified requirement indicators and dynamic intervention goals, collect children's action, expression, and operation frequency behavior data in real time, record historical information through a memory, and extract key feature data according to the requirement weights. Step 3, Construct a Joint State Evaluation Model: Combine the requirement goals and the processed feature data to construct a joint state evaluation model, calculate the comprehensive state score based on the output of the model, and then compare and analyze it with the dynamic intervention goal. Step 4, Dynamic Strategy Adjustment: Generate an adjustment strategy for dynamic adjustment according to the output comprehensive state score and the determined dynamic intervention goal. The adjustment strategy includes adjusting the difficulty of teaching content and social guidance strategies. Step 5, Feedback Optimization Testing and Mass Production: Collect user feedback to continuously optimize the joint state evaluation adjustment strategy and model. After the strategy is adjusted, conduct tests again, and after meeting the standards, carry out mass production.

[0007] Preferably, the specific steps of Step 1 are as follows: Step 1.1, Comprehensively obtain children's needs from different dimensions by observing children's performance in the classroom and social activities: Cognitive dimension, including the needs of language comprehension and mathematical operation abilities; Social dimension, including the needs of cooperation ability and communication ability; Emotional dimension, including the needs of emotion management and self-confidence; Step 1.2, Convert the collected needs into quantified indicators; construct an Analytic Hierarchy Model, where the criterion layer includes the cognitive, social, and emotional dimensions, and the scheme layer is the specific quantified requirement indicators under each dimension; Step 1.3, Through education experts and psychological experts, based on experience and professional knowledge, make pairwise comparisons of the quantified needs to determine the relative importance, form a judgment matrix with requirement indicators as elements, and the judgment matrix is expressed as ; where n is the order of the judgment matrix, is the total number of elements in the matrix, represents the importance of requirement indicator element i relative to requirement indicator element j, The value follows the 1-9 scale value method. 1-9 indicates that the importance of the two elements increases gradually. 1 means the two elements are equally important, 9 means element i is extremely more important than element j, and , ; Step 1.4: Use the power method numerical calculation method to solve the maximum eigenvalue in the judgment matrix and its corresponding eigenvector . After obtaining the eigenvector, perform normalization, that is: where T represents the transpose operation of a vector or matrix. After obtaining the weights of each requirement index , calculate the consistency index CI: Step 1.5: Then, obtain the average random consistency index RI corresponding to different orders n by referring to the standard table of the analytic hierarchy process, and calculate the consistency ratio: where when CR < 0.1, it is considered that the judgment matrix has satisfactory consistency and the weight calculation result is reliable; After the above calculations and tests, when the judgment matrix has satisfactory consistency, output the requirement index weight vector .

[0008] Preferably, after step 1.5, a standardized test is also conducted on the child, and the child's current ability baseline vector on each requirement index is statistically obtained , as well as the dynamic intervention target vector set for each requirement index of the child according to the educational goal and the child's development stage ; where represents the current level of the child on the i-th requirement index, represents the expected target value of the i-th requirement index.

[0009] Preferably, in step 2, the specific process of extracting key feature data according to the requirement weights is as follows: Based on the requirement index weights , screen out the features that have the most influence on the comprehensive status assessment of the child; Suppose there are a total of ' features, and each feature x is associated with n requirement indexes. The degree of association is represented by the Pearson correlation coefficient calculated by statistical analysis methods. Then the comprehensive importance score of the feature is: Sort the calculated from largest to smallest, and select the top m features as the features that have an impact on the comprehensive status assessment of the child.

[0010] Preferably, the specific process of constructing the joint state evaluation model in step 3 is as follows: The joint state evaluation model is jointly composed of a Gaussian mixture model and a Bayesian network. The Gaussian mixture model is used to model the data distribution in the feature data space, while the Bayesian network is used to describe the causal relationship between features. Let the feature data in the feature data space follow a Gaussian mixture distribution, and its probability density function is: where L is the number of Gaussian components, is the weight of the k-th Gaussian component and satisfies; where, is a multivariate Gaussian distribution with a mean of and a covariance matrix of , and its expression is: where m is the dimension of the feature data, that is, the total number of selected feature data, and the parameters of the Gaussian mixture model are calculated by the expectation-maximization algorithm.

[0011] Preferably, the Bayesian network in the joint state evaluation model is represented by a directed acyclic graph , where is the set of nodes, that is, each feature data corresponding to the demand index i, and E is the set of directed edges, representing the causal relationship between features; For each demand index , the feature data , and each node has a conditional probability distribution , where is the set of parent nodes of node , that is, the set of all nodes with a direct directed edge pointing to node ; The joint conditional probability of the feature given the state of its parent nodes is calculated through the chain rule of the Bayesian network ; Then, the K2 algorithm based on Bayesian network structure learning is used to determine the causal relationship structure between features.

[0012] Preferably, the specific process of calculating the comprehensive state score in step 3 is as follows: According to the trained joint state evaluation model, the feature data x of each child is calculated to obtain its probability distribution belonging to different state categories. Then, combined with the demand index weight , the comprehensive state score S is calculated; For the calculation of the comprehensive status score, first, calculate the posterior probability that the child status vector x belongs to the k-th Gaussian component: where, represents the latent variable that the feature data x belongs to the k-th Gaussian component. Then, combined with the demand index weight , calculate the comprehensive status score S: where the scoring function is constructed based on the features of the k-th Gaussian component and the i-th demand index, and its expression is: where, is a function based on the feature data x, which can perform non-linear transformation through the linear combination of eigenvalues. The specific form is determined according to the actual needs and data characteristics, and is used to further quantify the contribution of these features to the score.

[0013] Preferably, the specific process of dynamic policy adjustment in step 4 is as follows: Calculate the score gap and the adjustment coefficient o based on the comprehensively calculated status score S. Compare the comprehensive status score S with the dynamically intervened target score calculated based on the target vector and its corresponding weight vector to obtain the score gap ; The dynamically intervened target score is: The score gap: Then, calculate the adjustment coefficient o based on the score gap : Determine the adjustment strategy. According to the score gap and the demand index weight , for the demand index with higher weight and larger score gap, the adjustment intensity is greater, and dynamically adjust the difficulty of teaching content and social guidance strategy; Implement the adjustment strategy, apply the adjusted teaching content and social guidance strategy to the actual teaching, observe the reactions and performances of children. At the same time, according to the actual performances of children, give feedback and optimization to the adjustment strategy, and continuously adjust the teaching content and social guidance frequency to achieve better intervention effects; For the adjustment amount of the teaching content difficulty or social guidance frequency , assuming the current value is C, then the adjustment amount: The adjusted value is: Among them, is the demand index weight related to the difficulty of teaching content or the frequency of social guidance.

[0014] Preferably, in step 5: The collection of user feedback specifically includes comprehensively collecting the experience feelings, problem feedback, and improvement suggestions of parents, teachers, and children after using the teaching toys through the methods of online questionnaires, offline interviews, and setting up suggestion collection boxes in pilot sites; To evaluate whether the effect of the adjustment strategy meets the dynamic intervention goal, it is judged by comparing the changes in the scores of demand indicators of children in the cognitive, social, and emotional dimensions before and after using the teaching toys, as well as the changes in the behavioral performance of children in the actual scenario.

[0015] Compared with the prior art, the present invention has the following beneficial effects: 1. The present invention deploys sensors on the teaching toys to collect real-time behavioral data such as children's actions, expressions, and operation frequencies, constructs a joint state evaluation model to evaluate the comprehensive state of children in combination with the demand weights determined by the analytic hierarchy process, and dynamically generates an adjustment strategy according to the evaluation results, realizing dynamic optimization based on real-time behavioral data, achieving the beneficial effect of overcoming the problem that existing products rely on static preset algorithms and lack dynamic optimization capabilities, and improving the intervention effect.

[0016] 2. The present invention sets a reasonable social guidance strategy in the dynamic strategy adjustment, and collects the feedback of parents, teachers, and children in the feedback optimization test stage, comprehensively considering the balance of human-computer interaction and interpersonal interaction, realizing the interaction opportunity between children and real parents or teachers while using the teaching toys for intervention, achieving the beneficial effect of avoiding the problem of social deprivation and better realizing the social skill training goal. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 is a schematic flowchart of the method steps of the present invention; Figure 2 is a schematic flowchart of the specific process for establishing the joint state evaluation model of the present invention; Figure 3 is a schematic flowchart of the calculation of the adjustment strategy and optimization of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0018] The following further describes the implementation mode of the present invention in conjunction with the accompanying drawings and embodiments. The following embodiments are used to illustrate the present invention, but cannot be used to limit the scope of the present invention.

[0019] The present invention provides a method for guiding the design and development of teaching toys for children with mental disorders based on big data, including: Step 1, requirement analysis and goal definition, quantify the requirement indicators in the three dimensions of children's cognition, social interaction, and emotion, calculate the weights of each requirement indicator using the analytic hierarchy process, and simultaneously set a dynamic intervention target vector; Step 2, multi-modal data collection, deploy multiple sensors on the teaching toys according to the quantified requirement indicators and dynamic intervention goals, collect children's action, expression, and operation frequency behavior data in real time, record historical information through a memory, and extract key feature data according to the requirement weights; Step 3, construct a joint state evaluation model, combine the requirement goals and the processed feature data to construct a joint state evaluation model, calculate the comprehensive state score based on the output of the model, and then compare and analyze it with the dynamic intervention goal; Step 4, dynamic strategy adjustment, generate an adjustment strategy for dynamic adjustment according to the output comprehensive state score and the determined dynamic intervention goal, and the adjustment strategy includes adjusting the difficulty of teaching content and social guidance strategies; Step 5, feedback optimization testing and mass production, collect user feedback to continue to optimize the joint state evaluation adjustment strategy and model, after the strategy adjustment, conduct testing again, and conduct mass production after meeting the standards.

[0020] Embodiment 1: As Figures 1 - 3 shown, in this embodiment, a group of children with mental disorders newly enrolled in a special education school, covering different degrees of cognitive, social, and emotional development problems. Most of the existing teaching toys in the school are in a traditional fixed mode and cannot be adjusted according to the specific situation of each child, resulting in uneven teaching effects. Some children show obvious frustration or inattention when using the existing toys, and it is also difficult for teachers to accurately grasp the needs and progress of each child. Therefore, the school introduced a method for guiding the design and development of teaching toys for children with mental disorders based on big data to meet the special needs of these children with mental disorders.

[0021] The institution organized a team of teachers and experts to comprehensively collect the needs of children in aspects such as cognition (such as language understanding, mathematical operation ability), social interaction (such as cooperation ability, communication ability), and emotion (such as emotion management, self-confidence) by observing the performance of children in the classroom and social activities. For example, observe the reaction speed and accuracy of children's mathematical operations in the classroom, and record the situation of children cooperating with their peers to complete tasks in social activities, etc.

[0022] Convert the collected requirements into quantitative index data. For example, cognitive ability is quantified as the knowledge mastery rate and the learning progress rate; social ability is quantified as the social interaction frequency, the cooperation success rate, etc.

[0023] Through education experts and psychological experts, pairwise comparison of the quantified requirements is carried out according to experience and professional knowledge to determine the relative importance, and a judgment matrix with requirement indicators as elements is formed. The judgment matrix is expressed as ; where n is the order of the judgment matrix, is the total number of elements in the matrix, represents the importance of requirement indicator element i relative to requirement indicator element j, takes values following the 1-9 scale value method. 1-9 indicates that the importance of the two elements increases gradually. 1 means the two elements are equally important, 9 means element i is extremely more important than element j, and , ; Use the power method numerical calculation method to solve the maximum eigenvalue and its corresponding eigenvector in the judgment matrix. After obtaining the eigenvector, perform normalization processing, that is: where T represents the transpose operation on vectors or matrices. After obtaining the weights of each requirement indicator, calculate the consistency index CI: Then, obtain the average random consistency index RI corresponding to different orders n by referring to the standard table of the analytic hierarchy process, and calculate the consistency ratio: where when CR < 0.1, it is considered that the judgment matrix has satisfactory consistency and the weight calculation result is reliable; After the above calculations and tests, when the judgment matrix has satisfactory consistency, output the requirement indicator weight vector .

[0024] At the same time, standardized tests are also conducted on children, and the baseline vector of children's abilities at each requirement indicator is statistically obtained, as well as the dynamic intervention target vector set for each requirement indicator of children according to educational goals and children's development stages; where, represents the current level of children on the i-th requirement indicator, represents the expected target value of the i-th requirement indicator.

[0025] The use of big data technology to capture children's behavioral characteristics in real time provides rich data support for subsequent analysis and intervention, solving the problem of the lack of real-time data in existing products.

[0026] Then, according to the determined requirements and dynamic intervention goals, the institution deployed a variety of sensors on teaching toys, such as motion sensors, facial expression recognition sensors, etc., to collect children's behavioral data such as actions, facial expressions, and operation frequencies in real time. At the same time, it collected children's past learning records and diagnostic report information, and extracted key feature data according to the demand weights. The specific process is as follows: Based on the demand index weights , screen out the features that have the most influence on the comprehensive status assessment of children; Suppose there are a total of ' features, each feature x is associated with n demand indicators, and the degree of association is represented by the Pearson correlation coefficient calculated by statistical analysis methods. Then the comprehensive importance score of feature is: Sort the calculated from largest to smallest, and select the top m features as the features that have an impact on the comprehensive status assessment of children. This link effectively processes the collected data, extracts key information, improves the accuracy of subsequent model analysis, and lays a foundation for establishing a more accurate joint status assessment model.

[0027] Then, based on the extracted feature data, a joint status assessment model is constructed. The specific process is as follows: The joint status assessment model is jointly composed of a Gaussian mixture model and a Bayesian network. The Gaussian mixture model is used to model the data distribution in the feature data space, while the Bayesian network is used to describe the causal relationship between features; Suppose the feature data in the feature data space follows a Gaussian mixture distribution, and its probability density function is: Among them, L is the number of Gaussian components, is the weight of the kth Gaussian component, and satisfies; Among them, is a multivariate Gaussian distribution with a mean of and a covariance matrix of , and its expression is: Among them, m is the dimension of the feature data, that is, the total number of selected feature data, and the parameters of the Gaussian mixture model are calculated by the expectation-maximization algorithm.

[0028] Preferably, in the joint state evaluation model, the Bayesian network is represented by a directed acyclic graph where is the set of nodes, that is, each feature data in the corresponding demand index i, and E is the set of directed edges, representing the causal relationship between features; For each demand index , the feature data therein , for each node there is a conditional probability distribution where is the set of parent nodes of node , that is, the set of all nodes with a direct directed edge pointing to node ; The joint conditional probability of a feature given the states of its parent nodes is calculated by the chain rule of the Bayesian network ; then, the K2 algorithm based on Bayesian network structure learning is used to determine the causal relationship structure between features.

[0029] By deploying sensors on teaching toys to collect real-time behavioral data such as children's actions, expressions, and operation frequencies, and combining the demand weights determined by the analytic hierarchy process, a joint state evaluation model is constructed to evaluate the comprehensive state of children, realizing dynamic optimization based on real-time behavioral data, achieving the beneficial effect of overcoming the problem that existing products rely on static preset algorithms and lack dynamic optimization capabilities, and improving the intervention effect.

[0030] Example 2: As Figures 1 - 3 shown, in this embodiment, after the school has successfully established a joint state evaluation model in Example 1 and carried out teaching intervention on some children with mental disabilities for a period of time according to the model. However, in the actual application process, teachers found that although the personalization degree of teaching toys has been improved, there are still some problems. For example, the progress of some children in some aspects is not obvious, and the adjustment of teaching content sometimes fails to keep up with the development and changes of children in a timely manner. At the same time, parents also feedback that they are worried that children's excessive dependence on teaching toys will affect their normal communication with others.

[0031] So the school decided to further optimize the teaching strategy according to the comprehensive score output by the model, and carry out mass production and continuous improvement of teaching toys to ensure the maximization of teaching effects and the all-round development of children.

[0032] The specific process of calculating the comprehensive state score is as follows: According to the trained joint state evaluation model, the characteristic data x of each child is calculated to obtain the probability distribution of belonging to different state categories. Then, combined with the demand index weights , the comprehensive state score S is calculated; For the calculation of the comprehensive state score, first, calculate the posterior probability that the child state vector x belongs to the k-th Gaussian component: where represents the latent variable that the characteristic data x belongs to the k-th Gaussian component. Then, combined with the demand index weights , the comprehensive state score S is calculated: where the scoring function is constructed based on the characteristics of the k-th Gaussian component and the i-th demand index, and its expression is: where is a function based on the characteristic data x, which can perform non-linear transformation through the linear combination of characteristic values. The specific form is determined according to actual needs and data characteristics, and is used to further quantify the contribution of these characteristics to the score. Through the above calculation method, the comprehensive state of children is accurately quantified, providing a clear reference for subsequent dynamic policy adjustment and making the intervention more accurate and effective.

[0033] The specific process of dynamic policy adjustment is as follows: Calculate the score gap and the adjustment coefficient o based on the comprehensively calculated state score S. Compare the comprehensive state score S with the dynamically intervened target score calculated based on the target vector and its corresponding weight vector to obtain the score gap ; The dynamically intervened target score is: Score gap: Then, calculate the adjustment coefficient o based on the score gap : Determine the adjustment strategy. According to the score gap and the demand index weights , for the demand index with higher weight and larger score gap, the adjustment intensity is greater, and the difficulty of teaching content and social guidance strategy are dynamically adjusted; Implement an adjustment strategy, apply the adjusted teaching content and social guidance strategy to actual teaching, observe the children's reactions and performances. At the same time, based on the children's actual performances, provide feedback on and optimize the adjustment strategy, and continuously adjust the teaching content and social guidance frequency to achieve better intervention effects; For the difficulty of teaching content or the adjustment amount of social guidance frequency , let the current value be C, then the adjustment amount: The adjusted value is: where is the weight of the demand index related to the difficulty of teaching content or the social guidance frequency.

[0034] After a small batch of tests are carried out, specifically collect the experience feelings, problem feedback and improvement suggestions of parents, teachers and children after using the teaching toys through online questionnaires, offline interviews and setting up suggestion collection boxes at the pilot sites; Then judge whether the effect of the adjustment strategy meets the dynamic intervention goal by comparing the changes in the scores of the demand indicators of children in the cognitive, social and emotional dimensions before and after using the teaching toys, as well as the changes in the children's behavioral performances in the actual scenarios.

[0035] Through dynamically calculating the adjustment strategy, the dynamic adaptive optimization of the teaching toys is realized. The intervention strategy is adjusted in a timely manner according to the real-time status and needs of children, overcoming the problems that existing products rely on static preset algorithms and lack the ability of dynamic optimization, and improving the intervention effect.

[0036] The embodiments of the present invention are given for the purpose of illustration and description. Although the embodiments of the present invention have been shown and described above, it can be understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those of ordinary skill in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of the present invention.

Claims

1. A guiding method for the design and development of teaching toys for children with mental disorders based on big data, characterized in that, Including: Step 1, requirement analysis and goal definition: Quantify the requirement indicators in the three dimensions of children's cognition, social interaction, and emotion, calculate the weights of each requirement indicator using the analytic hierarchy process, and simultaneously set a dynamic intervention goal vector. Step 2, multi-modal data collection: Based on the quantified requirement indicators and dynamic intervention goals, deploy multiple sensors on teaching toys to collect children's action, expression, and operation frequency behavior data in real time. At the same time, record historical information through a memory, and extract key feature data according to the requirement weights. Step 3, construct a joint state evaluation model: Combine the requirement goals and the processed feature data to construct a joint state evaluation model, calculate the comprehensive state score based on the output of the model, and then compare and analyze it with the dynamic intervention goal. Step 4, dynamic strategy adjustment: Generate an adjustment strategy for dynamic adjustment according to the output comprehensive state score and the determined dynamic intervention goal. The adjustment strategy includes adjusting the difficulty of teaching content and social guidance strategies. Step 5, feedback optimization testing and mass production: Collect user feedback to continuously optimize the joint state evaluation adjustment strategy and model. After the strategy is adjusted, conduct tests again, and conduct mass production after meeting the standards.

2. The guiding method for the design and development of teaching toys for children with mental disorders based on big data according to claim 1, characterized in that: The specific steps in Step 1 are as follows: Step 1.1, comprehensively obtain children's needs from different dimensions by observing children's performance in the classroom and social activities: Cognitive dimension, including the needs of language comprehension and mathematical operation abilities; Social dimension, including the needs of cooperation ability and communication ability; Emotional dimension, including the needs of emotion management and self-confidence; Step 1.2, convert the collected needs into quantified indicators; construct an analytic hierarchy model, where the criterion layer includes the cognitive, social, and emotional dimensions, and the scheme layer is the specific quantified requirement indicators under each dimension. Step 1.

3. According to experience and professional knowledge, educational experts and psychological experts conduct pairwise comparisons on the quantified requirements to determine the relative importance, form a judgment matrix with demand indicators as elements, and the judgment matrix is expressed as ; where n is the order of the judgment matrix, is the total number of elements in the matrix, represents the importance of demand index element i relative to demand index element j, The value follows the 1-9 scale value method. 1-9 indicates that the importance of the two elements increases gradually. 1 means the two elements are equally important, and 9 means element i is extremely important compared to element j, and , ; Step 1.4, use the power method numerical calculation method to solve the maximum eigenvalue in the judgment matrix and its corresponding eigenvector , and after obtaining the eigenvector, perform normalization processing, that is: Among them, T represents the transpose operation on vectors or matrices. After obtaining the weights of various requirement indicators calculate the consistency index CI: Step 1.5, then, obtain the average random consistency index RI corresponding to different orders n by referring to the standard table of the analytic hierarchy process, and calculate the consistency ratio: Among them, when CR < 0.1, it is considered that the judgment matrix has satisfactory consistency and the weight calculation result is reliable; After the above calculations and tests, when the judgment matrix has satisfactory consistency, output the weight vector of the demand indicators .

3. The guiding method for the design and development of teaching toys for children with mental disorders based on big data according to claim 2, characterized in that: After step 1.5, a standardized test is also conducted on the child, and the baseline vector of the child's capabilities on each demand index is statistically obtained , and a dynamic intervention target vector for setting each demand index of the child is determined according to the educational goal and the child's development stage ; Among them, represents the current level of the child on the i-th demand index, represents the expected target value of the i-th demand index.

4. The guiding method for the design and development of teaching toys for children with mental disorders based on big data according to claim 1, characterized in that: In Step 2, the specific process of extracting key feature data according to the requirement weights is as follows: Based on the weights of requirement indicators , the features that have the most influence on the comprehensive status assessment of children are screened out; Suppose there are in total ' features, and each feature x is associated with n demand indicators. The degree of association is represented by the Pearson correlation coefficient calculated by statistical analysis methods. Then the comprehensive importance score of feature is: The calculated Sort them in descending order and select the top m features as the features that have an impact on the comprehensive state assessment of children.

5. The guiding method for the design and development of teaching toys for children with mental disorders based on big data according to claim 4, characterized in that: The specific process of constructing the joint state evaluation model in Step 3 is as follows: The joint state evaluation model is jointly composed of a Gaussian mixture model and a Bayesian network. The Gaussian mixture model is used to model the data distribution in the feature data space, and the Bayesian network is used to describe the causal relationship between features; Suppose the feature data in the feature data space obeys a Gaussian mixture distribution, and its probability density function is as follows: where L is the number of Gaussian components, is the weight of the k-th Gaussian component and satisfies; Among them, is a multivariate Gaussian distribution with a mean of and a covariance matrix of The expression is as follows: Among them, m is the dimension of the feature data, that is, the total number of selected feature data, and the parameters of the Gaussian mixture model are calculated by the expectation-maximization algorithm.

6. The guiding method for the design and development of teaching toys for children with mental disorders based on big data according to claim 5, characterized in that: In the joint state evaluation model, the Bayesian network is represented by a directed acyclic graph where is a set of nodes, that is, each feature data in the corresponding requirement index i, and E is a set of directed edges, representing the causal relationship between features; For each requirement metric , the feature data , each node has a conditional probability distribution , where is the set of parent nodes of node , that is, the set of all nodes with a direct directed edge pointing to node ; Calculate the joint conditional probability of features given the states of their parent nodes through the chain rule of Bayesian networks ; Then, use the K2 algorithm based on Bayesian network structure learning to determine the causal relationship structure between features.

7. The guiding method for the design and development of teaching toys for children with mental disorders based on big data according to claim 6, characterized in that: In Step 3, the specific process of calculating the comprehensive state score is as follows: According to the trained joint state evaluation model, calculate the feature data x of each child to obtain the probability distribution of different state categories, and then, combined with the demand index weights , calculate the comprehensive state score S; Comprehensive state score calculation: First, calculate the posterior probability that the child state vector x belongs to the k-th Gaussian component: Among them, represents the latent variable that the feature data x belongs to the k-th Gaussian component. Then, combining the demand index weights , calculate the comprehensive status score S: Among them, the scoring function is constructed based on the characteristics of the k-th Gaussian component and the i-th demand index, and its expression is: Among them, is a function based on the feature data x, which can perform non-linear transformation through linear combination of eigenvalues. Its specific form is determined according to actual requirements and data characteristics, and is used to further quantify the contribution of these features to the score.

8. The guiding method for the design and development of teaching toys for children with mental disorders based on big data according to claim 1, characterized in that: The specific process of dynamic strategy adjustment in Step 4 is as follows: Calculate the score gap based on the comprehensive status score S calculated by real-time computing and the adjustment coefficient o, compare the comprehensive status score S with the dynamic intervention target score calculated based on the target vector and its corresponding weight vector to obtain the score gap ; Dynamic intervention target score is as follows: Score gap: Based on the score difference again Calculate the adjustment coefficient o: Determine the adjustment strategy based on the score gap and the weight of the requirement indicators For requirement indicators with higher weights and larger score gaps, the adjustment intensity is greater, and the difficulty of teaching content and social guidance strategies are dynamically adjusted; Implement the adjustment strategy, apply the adjusted teaching content and social guidance strategy to actual teaching, observe the reactions and performances of children. At the same time, based on the actual performances of children, provide feedback on and optimize the adjustment strategy, and continuously adjust the teaching content and social guidance frequency to achieve better intervention effects; For the teaching content difficulty or the adjustment amount of the social guidance frequency , let the current value be C, then the adjustment amount is: Adjusted value is Among them, is the demand index weight related to the difficulty of teaching content or the frequency of social guidance.

9. The guiding method for the design and development of teaching toys for children with mental disorders based on big data according to claim 1, characterized in that: In step 5: The collection of user feedback specifically includes comprehensively collecting the experience feelings, problem feedback and improvement suggestions of parents, teachers and children after using the teaching toys through online questionnaires, offline interviews and setting up suggestion boxes at the pilot sites; Whether the effect of the evaluation adjustment strategy meets the dynamic intervention goal is judged by comparing the score changes of the demand indicators of children in the cognitive, social and emotional dimensions before and after using the teaching toys, as well as the behavior performance changes of children in the actual scenarios.

Images