A basic motor skill quantitative test method and intelligent evaluation device for child development evaluation

Abstract

The application discloses a basic motor skill quantitative test method and an intelligent evaluation device for child development evaluation, and relates to the technical field of child development evaluation. The specific steps of the method are as follows: first, collecting multi-source data of child movement through a device and standardizing and integrating the data; then, dividing sequence stages and constructing a standard action feature library; then, dynamically matching to determine specific stages and collecting exclusive parameters; then, performing multi-modal fusion evaluation, generating a report and early warning information; finally, generating a personalized intervention scheme according to the evaluation and early warning, and optimizing the model according to the effect to form a closed loop system. The application breaks the limitation of single data evaluation, collects and integrates data in multiple dimensions, divides sequence stages, constructs a standard feature library, and makes up for the deficiency of traditional evaluation. In addition, the application accurately judges and early warns risks through collaborative matching and fusion evaluation, generates differentiated intervention content, records data optimization models, forms a closed loop, solves the disconnection problem of traditional evaluation and intervention, and helps to improve child development.

Description

Technical Field

[0001] This invention relates to the field of child development assessment technology, specifically to a quantitative testing method and intelligent assessment device for basic motor skills in child development assessment. Background Technology

[0002] The development of children's basic motor skills is the core foundation for their physical and mental health and comprehensive ability improvement, directly affecting multiple dimensions such as growth and development, social interaction, and cognitive development. The period from 3 to 12 years old is a critical period for the rapid development of children's motor skills, making scientific and effective developmental assessment of children in this stage of significant practical importance. With the continuous development of the field of child development assessment, traditional experience-based assessment models are no longer sufficient to meet the needs of refinement and objectivity. The industry is gradually shifting towards data-driven and intelligent assessment. Currently, the maturity of visual capture and wearable sensing technologies has made multi-dimensional data collection possible, while the application of algorithms such as multi-source data fusion and deep learning provides technical support for the scientific validity of assessment results. Against this backdrop, there is an urgent need for a quantitative assessment method that can systematically integrate information from multiple aspects such as motor performance, physiological signals, and cognitive linkages, and fully adapt to the differences in children's age and gender, in order to achieve accurate capture and scientific analysis of children's basic motor skill development status and provide a reliable basis for children's developmental guidance.

[0003] Traditional methods for assessing children's motor skills have many limitations and cannot meet the current needs for refined assessment. Most traditional methods rely on the manual observation and subjective scoring of assessors, and the assessment results are easily affected by factors such as personal experience and observation angle, resulting in insufficient objectivity and consistency. They cannot accurately reflect the true developmental level of children. Some assessment methods focus only on single-dimensional data collection, or only on limb motor performance, or only monitor some physiological indicators, lacking a comprehensive capture of multi-dimensional information such as motor, physiological, and cognitive aspects. This leads to one-sided assessment results and makes it difficult to fully present the overall state of children's motor skill development. At the same time, traditional methods lack a unified standardized data integration mechanism. Data of different types and sources are difficult to effectively integrate and compare due to differences in dimensionality, and the impact of individual differences of children on the assessment results is not fully considered, further reducing the accuracy of the assessment. In addition, in the traditional assessment process, the assessment and intervention stages are independent of each other. The assessment results are difficult to directly translate into actionable intervention plans, and there is a lack of dynamic tracking and feedback optimization of intervention effects, which makes it impossible to form an effective closed-loop management and achieve timely identification and targeted intervention of early developmental risks. Summary of the Invention

[0004] The purpose of this invention is to overcome the shortcomings of existing technologies and provide a quantitative testing method and intelligent assessment device for basic motor skills in children's developmental assessment. This method simultaneously collects multi-dimensional data related to macroscopic movement, microscopic physiology, and cognition through visual capture and wearable devices. After standardized algorithm integration and processing, it divides the data into sequential stages based on children's motor development patterns and constructs a standard feature library. A collaborative matching algorithm is used to accurately locate the child's developmental stage. A multi-task deep learning model is then used to calculate skill scores and provide multi-dimensional risk warnings. Finally, based on the assessment results, a personalized intervention plan covering movement correction, physical training, and cognitive linkage training is generated. The assessment model is optimized using intervention data to form a complete closed loop. This method achieves comprehensiveness, accuracy, and targeted intervention in children's motor skills assessment, providing scientific support for children's developmental guidance.

[0005] To solve the above-mentioned technical problems, the present invention provides the following technical solution: On the one hand, a quantitative testing method for basic motor skills in children's developmental assessment, the specific steps of which are as follows:

[0006] S1. Full-dimensional data collection and integration: Through a visual capture system and wearable sensing devices, macroscopic motion parameters, multimodal data and microscopic physiological signals are collected simultaneously when children perform basic motor skills; the original multi-source data is fused and processed using a stage-adaptive multi-source data standardization algorithm to obtain a full-dimensional standardized integration score;

[0007] S2. Sequence Stage Division and Feature Library Construction: Based on the comprehensive standardized integrated score, the basic motor skills of children aged 3-12 years are divided into sequence stages; a standard movement feature library corresponding to each sequence stage is constructed to initially determine the sequence stage range to which the child belongs;

[0008] S3. Dynamic Sequence Matching and Parameter Acquisition: Using the action-physiology co-sequence matching algorithm, the full-dimensional standardized integrated score obtained in step S1 is decomposed into the child's current action feature vector; it is matched and calculated with the standard action feature library constructed in step S2 to determine the specific sequence stage to which the child belongs; for the determined sequence stage, the corresponding exclusive quantitative parameters are collected, and the action sequence lag is marked.

[0009] S4. Multimodal Fusion Assessment and Risk Warning: Deeply fuse the multi-dimensional data collected in step S1 and the sequence matching results in step S3; calculate the child's motor skill score based on a multi-task deep learning model; and generate a comprehensive developmental assessment report and early risk warning information using a multi-dimensional motor development risk warning algorithm.

[0010] S5. Personalized Intervention and Closed-Loop Optimization: Based on the assessment report and early warning information generated in step S4, generate a personalized intervention plan; based on the effect data during the intervention process, optimize the assessment model parameters to form a closed-loop system of assessment-intervention-optimization.

[0011] Furthermore, the specific implementation steps for synchronously collecting data through the visual capture system and wearable sensing devices are as follows: deploying the visual capture system to capture the limb movement trajectory of children when performing basic motor skills, recording joint rotation angles, movement completion time, movement amplitude, and movement transition time to form macroscopic motion parameters; equipping children with wearable sensing devices to collect children's muscle group electrical signals, heart rate, heart rate variability fluctuations, and muscle coordination patterns to form microscopic physiological signals; embedding lightweight cognitive tasks during the process of children performing basic motor skills, collecting the accuracy rate and reaction time of children completing cognitive tasks to form multimodal data; aligning the collected macroscopic motion parameters, microscopic physiological signals, and multimodal data with timestamps, and synchronously transmitting them to the data processing terminal for storage to complete the synchronous collection of multi-source data.

[0012] Furthermore, the mathematical expression for the stage-adapted multi-source data standardization algorithm is: ;in To standardize and integrate scores across all dimensions, For the i-th type of raw collected data, , For the localized norm extreme values ​​of the same age and gender for the i-th type of data, For the dynamic weights of the i-th class of data, The baseline adaptation coefficient is based on the child's age. In the final stage of the full-dimensional data collection and integration in step S1, the algorithm integrates macroscopic motion parameters, multimodal data, and microscopic physiological signal data to eliminate differences in the dimensions of various data. It combines dynamic weight allocation to adapt different data credibility, embeds sequence stage adaptation coefficients to calibrate the matching degree between the data and the corresponding sequence stage, and outputs a unified scale full-dimensional standardized integration score, providing standardized data support for the sequence stage division and feature library construction in step S2.

[0013] Furthermore, the specific content of the sequence stage division is as follows: Based on the theory of children's overall motor development sequence, the complete development trajectory of basic motor skills is decomposed; combined with the theory of decomposed motor sequence, the motor development patterns of the upper limbs, trunk, and lower limbs are broken down; the full-dimensional standardized integration score obtained in step S1 is correlated; the motor characteristics of children aged 3 to 12 years of different ages and genders are statistically analyzed; and basic motor skills are divided into three categories: movement skills, stability skills, and object manipulation skills, and then divided into 8 sequence stages. The 8 sequence stages are as follows: Stage 1 is the basic movement perception stage, suitable for children aged 3-4 years, focusing on basic limb movement and body perception abilities; Stage 2 is the basic stability establishment stage, suitable for children aged 4-5 years, focusing on static and simple dynamic body stability abilities; Stage 3 is the basic object contact stage, suitable for children aged 4-5 years, focusing on basic object contact abilities; Stage 4 is the basic object contact stage, suitable for children aged 5-6 years, focusing on basic limb movement and body perception abilities; Stage 5 is the basic stability establishment stage, suitable for children aged 4-5 years, focusing on static and simple dynamic body stability abilities; Stage 6 is the basic object contact stage, suitable for children aged 3-4 years, focusing on basic limb movement and body perception abilities; Stage 7 is the basic stability establishment stage, suitable for children aged 4-5 years, focusing on static and simple dynamic body stability abilities; Stage 8 is the basic object contact stage, suitable for children aged 3-4 years, focusing on basic limb movement and body perception abilities; Stage 9 is the basic object contact stage, suitable for children aged 3-4 years, focusing on basic limb movement and body perception abilities; Stage 1 ... For children aged 5-6, the focus is on basic hand-eye coordination and grasping abilities. Stage 4 is the comprehensive mobility coordination stage, suitable for children aged 6-7, emphasizing the ability to complete multi-limb coordinated movement movements. Stage 5 is the dynamic stability control stage, suitable for children aged 7-8, emphasizing the ability to adjust dynamic body stability in complex scenarios. Stage 6 is the simple object manipulation stage, suitable for children aged 8-9, emphasizing the ability to manipulate objects in simple ways such as pushing, throwing, and catching. Stage 7 is the advanced mobility application stage, suitable for children aged 9-10, emphasizing the ability to flexibly apply advanced mobility skills in different scenarios. Stage 8 is the comprehensive object manipulation stage, suitable for children aged 10-12, emphasizing the ability to manipulate objects accurately and in a complex manner. The age range, gender compatibility differences, and specific movement patterns for each stage are clearly defined.

[0014] Furthermore, the specific content of constructing the standard action feature library corresponding to each sequence stage is as follows: the feature library contains feature parameters of action amplitude, joint coordination state, and action completion rhythm for each stage. Key features of children's actions are extracted and compared with the standard action feature library. The comparison results are calibrated by combining the sequence stage adaptation coefficient. Abnormal action data generated during the collection process are eliminated, positioning deviations are corrected, and the range of sequence stages to which the child belongs is initially determined.

[0015] Furthermore, the mathematical expression for the action-physiological co-sequence matching algorithm is: ;in: For sequence matching accuracy, This is the feature vector of the child's current action. This represents the standard action feature vector corresponding to the sequence stage. The standard deviation of HRV fluctuation is in the millisecond range. The standard deviation of a specific muscle synergy pattern. This refers to the action-physiology weighting coefficient. The age and gender correction coefficient is used. In step S3, the dynamic sequence matching and parameter acquisition stage, the algorithm compares the child's current action feature vector with the standard action feature vector, embeds the age and gender correction coefficient to calibrate the matching result, and determines the core sequence stage to which the child belongs when the matching accuracy is ≥85%. When the matching accuracy is <85%, the action features are re-extracted and compared again. After accurate positioning is completed, the specific quantitative parameters for this stage are collected.

[0016] Furthermore, the multi-task deep learning model includes an input layer, a multimodal feature extraction layer, a feature fusion layer, and a multi-task output layer. The input layer receives the multi-dimensional data collected in step S1 and the sequence matching results from step S3, and completes data alignment. The multimodal feature extraction layer has three parallel branches that extract features from macroscopic motion parameters, microscopic physiological signals, and cognitive data, respectively. Each branch extracts deep features of the data through convolutional and pooling layers, and maps the features into fixed-dimensional feature vectors through fully connected layers. The feature fusion layer uses an attention mechanism to weight and fuse the feature vectors output from the three branches, strengthening the feature weights related to children's motor development and weakening the influence of redundant features. The multi-task output layer includes a motor skill score calculation branch, which processes the fused feature vectors through fully connected layers and normalization layers to calculate the children's motor skill score.

[0017] Furthermore, the mathematical expression of the multi-dimensional motor development risk early warning algorithm is: ;in: As an early warning index, For sequence matching accuracy, This represents the lag coefficient for the sequence stage. , The cognitive-motor correlation coefficient, The cognitive adaptation constant is used. In step S4, the multimodal fusion assessment and risk warning stage, the sequence matching accuracy, sequence stage lag coefficient, physiological-motor correlation factor and cognitive-motor correlation coefficient are input into the algorithm to quantify and calculate the early warning index. The numerical range of the early warning index is set as follows: an early warning index ≥4 is a high-risk level, an early warning index between 2 and 3.9 is a medium-risk level, and an early warning index ≤1.9 is a low-risk level. For children with disordered microphysiological signals and low sequence matching accuracy, early risk warnings related to motor development are issued.

[0018] Furthermore, the specific implementation steps of the personalized intervention program are as follows: Based on the assessment report and early warning information generated in step S4, differentiate the child's motor development deficiencies, physical fitness, and motor-cognitive linkage performance, and generate corresponding personalized intervention content; for motor development deficiencies, design motor correction training using interactive demonstration methods, specifying the training movements, training duration, and repetition count; for physical fitness, formulate targeted physical training content, specifying the training intensity, training frequency, and phased goals; for motor-cognitive linkage performance, design lightweight cognitive and motor-integrated training tasks, specifying the task content and completion requirements; and record the training completion status, motor performance data, and phased feedback data during the intervention process.

[0019] On the other hand, an intelligent assessment device for child development includes:

[0020] Data acquisition unit: including visual capture system, wearable sensing device and cognitive task embedding module, used to simultaneously collect children's macroscopic motion parameters, microscopic physiological signals and multimodal data, and complete data timestamp alignment and transmission storage;

[0021] Data processing unit: Built-in stage-adaptive multi-source data standardization algorithm processing module, used to fuse and process the raw multi-source data collected by the data acquisition unit, and output a full-dimensional standardized integrated score;

[0022] Sequence Processing Unit: Used to combine the full-dimensional standardized integrated score to divide the basic motor skills of children aged 3-12 into 8 sequence stages, build and store the standard movement feature library corresponding to each sequence stage, and initially determine the range of the sequence stage to which the child belongs;

[0023] Matching Calculation Unit: Built-in action-physiology coordinated sequence matching algorithm processing module, used to decompose the full-dimensional standardized integrated score into the current action feature vector, perform matching calculation with the standard action feature library, determine the specific sequence stage of the child and collect exclusive quantitative parameters;

[0024] Fusion Assessment and Early Warning Unit: Built-in multi-task deep learning model and multi-dimensional motor development risk early warning algorithm processing module, used to deeply fuse multi-dimensional data and sequence matching results, calculate motor skill scores, and generate comprehensive development assessment reports and early risk warning information;

[0025] Intervention Optimization Unit: Used to generate personalized intervention plans based on comprehensive developmental assessment reports and early risk warning information, record intervention process data, and optimize assessment model parameters to form a closed-loop optimization.

[0026] Compared with existing technologies, this quantitative testing method and intelligent assessment device for basic motor skills in children's developmental assessment has the following beneficial effects:

[0027] I. This invention breaks through the limitations of single-data assessment by synchronously collecting and standardizing multi-dimensional data, achieving comprehensive capture and scientific integration of information related to children's motor skills. Utilizing visual capture and wearable device linkage, it covers macroscopic motor performance, microscopic physiological state, and cognitive linkage data. Combined with a standardized algorithm adapted to different stages, it eliminates the dimensional differences between different types of data. Through dynamic weight allocation and age baseline adaptation, the integrated results are more closely aligned with the individual characteristics of children. Simultaneously, based on the developmental patterns of children's movements, it divides the sequence into stages, constructing a standard feature library adapted to different developmental stages. Precise matching clarifies the developmental stage to which a child belongs, avoiding generalization and one-sidedness in assessment. This provides solid standardized data support for subsequent assessments, making the presentation of children's motor skill development status more objective and targeted, effectively compensating for the problems of fragmented data and insufficient adaptability in traditional assessments.

[0028] Second, this invention achieves accurate judgment and risk warning of children's motor skill development through action-physiological coordination matching and multimodal fusion assessment. Relying on a multi-task deep learning model to deeply mine the value of data correlation, strengthen the weight of key features, and weaken the interference of redundant information, the calculation of skill scores is more scientific. Combined with a multi-dimensional risk warning mechanism, it accurately identifies potential risks in motor development and provides a clear direction for early intervention. On this basis, it generates differentiated intervention content based on children's developmental shortcomings, physical characteristics, and cognitive linkage performance, covering motor correction, physical training, and cognitive-motor integration tasks. At the same time, it continuously optimizes the assessment model by recording intervention process data, forming a complete closed loop. This coherent system from assessment to intervention to optimization not only ensures the pertinence and suitability of intervention measures, but also dynamically tracks children's developmental changes, helping children's motor skills and overall development level to steadily improve, and solving the problems of disconnect between traditional assessment and intervention and lack of dynamic adjustment.

[0029] Other advantages, objectives and features of the invention will be set forth in part in the description which follows, and in part will be apparent to those skilled in the art from the following examination or study, or may be learned from the practice of the invention. Attached Figure Description

[0030] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without any creative effort.

[0031] Figure 1A flowchart illustrating the steps of a quantitative testing method for basic motor skills in children's developmental assessment.

[0032] Figure 2 This is a working logic diagram of each unit of an intelligent assessment device for child development.

[0033] Figure 3 A flowchart for the closed-loop process of assessment and intervention of children's motor skills. Detailed Implementation

[0034] To further illustrate the technical means and effects of the present invention in achieving its intended purpose, the following detailed description of the specific implementation methods, structures, features, and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided below.

[0035] Example 1:

[0036] Specific implementation scenarios for quantitative testing methods of basic motor skills for children's developmental assessment.

[0037] In a children's motor development assessment laboratory, basic motor skills quantitative tests are conducted on children aged 3-12 years. The implementation steps are as follows: Figure 1 As shown:

[0038] S1. Comprehensive Data Acquisition and Integration: Deploy a visual capture system to capture children's limb movement trajectories in real time when performing basic motor skills such as running, jumping, throwing, and balancing. Record joint rotation angles, movement completion time, movement amplitude, and movement transition time to form macroscopic motion parameters. Have children wear dedicated wearable sensing devices to continuously collect children's muscle group electrical signals, heart rate, heart rate variability fluctuations, and muscle coordination patterns to form microscopic physiological signals. During children's motor skill testing, embed simple, lightweight cognitive tasks such as number recognition and color matching, simultaneously collecting the accuracy and reaction time of children completing these cognitive tasks to form multimodal data. Align the three types of data with timestamps and transmit them synchronously to the data processing terminal for storage, ensuring a high temporal correlation between the different dimensions of data, laying the foundation for subsequent comprehensive analysis. After collection, use a stage-adaptive multi-source data standardization algorithm to fuse the three types of data. The mathematical expression of the stage-adaptive multi-source data standardization algorithm is: ;in To standardize and integrate scores across all dimensions, For the i-th type of raw collected data, , For the localized norm extreme values ​​of the same age and gender for the i-th type of data, For the dynamic weights of the i-th class of data, To eliminate dimensional differences among various data types and ensure a unified analytical scale for different data types, baseline fit coefficients based on children's age are used. This is combined with dynamic weight allocation to assess the reliability of different data types. Embedded sequence stage fit coefficients are used to calibrate the matching degree between the data and the corresponding sequence stages. Finally, a unified, multi-dimensional standardized integrated score is output, providing standardized and reliable data support for subsequent sequence stage segmentation. Figure 3 As shown.

[0039] S2. Sequence Stage Division and Feature Database Construction: Based on the theory of children's overall motor development sequence, the complete developmental trajectory of basic motor skills such as running, jumping, throwing, and balancing is decomposed. Combined with the theory of decomposed movement sequence, the developmental patterns of children's upper limbs, trunk, and lower limbs are broken down. This is then linked to the comprehensive standardized integrated score obtained in the previous step. Statistical analysis is performed on the movement characteristics of children aged 3-12 of different ages and genders in these motor skills. Basic motor skills are divided into three categories: movement skills, stability skills, and object manipulation skills, and further divided into 8 sequence stages. The 8 sequence stages are as follows: Stage 1 is the basic movement perception stage, suitable for children aged 3-4, focusing on basic limb movement and body perception abilities; Stage 2 is the basic stability establishment stage, suitable for children aged 4-5, focusing on static and simple dynamic body stability abilities; Stage 3 is the basic object contact stage, suitable for children aged 5-6, focusing on basic hand-object contact and grasping abilities; Stage 4 is the comprehensive movement coordination stage, suitable for children aged 6-7, focusing on the ability to complete multi-limb coordinated movement movements; Stage 5 is the dynamic stability control stage, suitable for children aged 7-12, focusing on dynamic stability control... For 8-year-old children, the focus is on dynamic body stability adjustment in complex scenarios; Stage 6 is the simple object manipulation stage, suitable for 8-9-year-old children, focusing on simple pushing, throwing, and catching of objects; Stage 7 is the advanced mobility application stage, suitable for 9-10-year-old children, focusing on the flexible application of advanced mobility skills in different scenarios; Stage 8 is the comprehensive object manipulation stage, suitable for 10-12-year-old children, focusing on precise and complex comprehensive manipulation of objects. The system clearly defines the age range, gender compatibility differences, and specific movement patterns for each stage, allowing children at different developmental levels to find corresponding developmental reference standards. Simultaneously, a standard movement feature library corresponding to each sequence stage is constructed. This feature library contains feature parameters of movement amplitude, joint coordination state, and movement completion rhythm for each stage. Key features of children's movements are extracted and compared with the standard movement feature library. Combined with the sequence stage adaptation coefficient, the comparison results are calibrated, abnormal movement data generated during the collection process is eliminated, positioning deviations are corrected, and the sequence stage range to which the child belongs is quickly and initially determined, improving the efficiency of subsequent accurate matching.

[0040] S3. Dynamic Sequence Matching and Parameter Acquisition: An action-physiology co-sequence matching algorithm is employed. The standardized integrated score across all dimensions is decomposed into the child's current action feature vector, which is then matched against the corresponding feature vector in the constructed standard action feature library. An age and gender correction coefficient is embedded to calibrate the matching results, ensuring the matching process fully considers individual differences among children and improving matching accuracy. The mathematical expression for the action-physiology co-sequence matching algorithm is: ;in: For sequence matching accuracy, This is the feature vector of the child's current action. This represents the standard action feature vector corresponding to the sequence stage. The standard deviation of HRV fluctuation is in the millisecond range. The standard deviation of a specific muscle synergy pattern. This refers to the action-physiology weighting coefficient. An age and gender correction coefficient is used; when the matching accuracy is ≥85%, the core sequence stage to which the child belongs is determined; when the matching accuracy is <85%, the child's action features are re-extracted and compared again to ensure that the sequence stage is accurately located; after the accurate location is completed, the specific quantitative parameters corresponding to the sequence stage are collected to obtain key data that can reflect the child's developmental level at that stage, and the lag in the child's action sequence during the action execution process is marked to provide a reference for a comprehensive assessment of the child's developmental pace.

[0041] S4. Multimodal Fusion Assessment and Risk Warning: The multimodal data of macroscopic motion parameters and microscopic physiological signals collected in the first step, along with the sequence matching results from the third step, are input into a multi-task deep learning model. The input layer of this model first aligns the data. The multimodal feature extraction layer uses three parallel branches to extract deep features from the three types of data. Each branch extracts features through convolutional and pooling layers, and then maps the features to fixed-dimensional feature vectors through a fully connected layer, ensuring that the deep value of each type of data can be extracted. The feature fusion layer uses an attention mechanism to weight and fuse the feature vectors output from the three branches, strengthening the weights of features related to children's motor development and weakening the influence of redundant features, allowing subsequent calculations to focus more on key information. The motor skill score calculation branch of the multi-task output layer processes the fused feature vectors through a fully connected layer and a normalization layer to calculate a motor skill score that accurately reflects the child's actual ability. Subsequently, a multi-dimensional motor development risk warning algorithm is used, inputting sequence matching accuracy, sequence stage lag coefficient, physiological-motor correlation factor, and cognitive-motor correlation coefficient, to quantitatively calculate an early warning index. The mathematical expression of the multi-dimensional motor development risk warning algorithm is: ;in: As an early warning index, For sequence matching accuracy, This represents the lag coefficient for the sequence stage. , The cognitive-motor correlation coefficient, To determine the adaptation constant, risk levels are judged according to the set numerical range criteria. An early warning index ≥4 indicates a high risk level, 2-3.9 indicates a medium risk level, and ≤1.9 indicates a low risk level. For children with disordered microphysiological signals and low sequence matching accuracy, early risk warnings related to motor development are issued to promptly identify potential developmental problems. Finally, a comprehensive developmental assessment report and early risk warning information that fully reflect the child's developmental status are generated.

[0042] S5. Personalized Intervention and Closed-Loop Optimization: Based on the generated comprehensive developmental assessment report and early risk warning information, the system differentiates children's motor development deficiencies, physical fitness, and motor-cognitive linkage performance, generating corresponding personalized intervention plans to better suit individual needs. For motor development deficiencies, interactive demonstrations are used to design corrective motor training, specifying training duration and repetition counts. For physical fitness, targeted physical training content is developed, specifying training intensity, frequency, and phased goals. For motor-cognitive linkage performance, lightweight training tasks combining cognitive and motor skills are designed, specifying task content and completion requirements. Throughout the intervention process, children's training completion, motor performance data, and phased feedback data are continuously recorded. Based on this data, the assessment model parameters are continuously optimized, making subsequent assessments more accurate and intervention plans more targeted, forming a closed-loop system of assessment-intervention-optimization to continuously improve the effectiveness of children's motor development guidance.

[0043] In summary, this quantitative testing method for basic motor skills in children's developmental assessment acquires macroscopic motor parameters, microscopic physiological signals, and multimodal data through comprehensive data collection and integration. A standardized multi-source data processing algorithm adapts to different stages to generate a unified score. Based on relevant theories, basic motor skills are categorized into three types: movement skills, stability skills, and object manipulation skills, and further divided into eight sequence stages, constructing a standard movement feature library. A movement-physiology co-matching algorithm accurately identifies the child's stage. A multi-task deep learning model calculates motor skill scores, and a multi-dimensional motor development risk warning algorithm generates assessment reports and warning information. Finally, personalized intervention plans are developed based on the assessment results, forming a closed-loop optimization mechanism that balances data standardization, matching accuracy, and intervention targeting throughout the process, providing a scientific and comprehensive solution for motor development assessment of children aged 3-12.

[0044] Example 2:

[0045] Specific implementation scenarios for intelligent assessment devices that quantify basic motor skills for children's developmental assessment.

[0046] In community children's health service centers, intelligent assessment devices for quantitative testing of basic motor skills for children's developmental assessment are installed to provide motor development assessment services for children aged 3-12, such as... Figure 2 As shown:

[0047] Data Acquisition Unit: This unit includes a wearable visual capture system and a cognitive task embedding module. After a child enters the assessment area, the visual capture system activates and aligns with the child's activity range, capturing in real-time the child's limb movement trajectories as they perform basic motor skills such as walking, jumping, catching a ball, and standing on one leg. It records macroscopic motion parameters such as joint rotation angles, movement duration, movement amplitude, and transition time. Staff then help the child put on the wearable sensor, which automatically begins collecting microscopic physiological signals such as muscle electrical signals, heart rate, heart rate variability fluctuations, electromyographic signals, and muscle coordination patterns. Simultaneously, the cognitive task embedding module presents simple, lightweight cognitive tasks such as graphic matching and number sorting to the child through an interactive screen, simultaneously collecting the child's accuracy and reaction time, forming multimodal data. The data acquisition unit timestamps the three types of data to ensure consistency in time sequence, and synchronously transmits them to the device's storage module for preservation, preventing data loss and providing a complete and comprehensive data source for subsequent processing by other units.

[0048] Data Processing Unit: This unit has a built-in stage-adaptive multi-source data standardization algorithm processing module. After receiving the raw multi-source data transmitted by the data acquisition unit, it automatically calls this algorithm to fuse macroscopic motion parameters, microscopic physiological signals, and multimodal data. The mathematical expression of the stage-adaptive multi-source data standardization algorithm is as follows: ;in To standardize and integrate scores across all dimensions, For the i-th type of raw collected data, , For the localized norm extreme values ​​of the same age and gender for the i-th type of data, For the dynamic weights of the i-th class of data, As a baseline adaptation coefficient based on children's age, the algorithm eliminates the dimensional differences between various types of data, enabling different types of data to have a unified analytical basis. Combined with dynamic weight allocation to adapt the credibility of different data, the algorithm embeds the adaptation coefficient of the sequence stage to calibrate the matching degree between the data and the corresponding sequence stage, and finally outputs a unified scale, full-dimensional standardized integrated score, which effectively improves the accuracy of subsequent sequence processing units and matching calculation units.

[0049] Sequence Processing Unit: After receiving the standardized integrated score from the data processing unit, based on the overall developmental sequence theory of children's motor skills and the decomposed movement sequence theory, and combining the score with statistical analysis of the movement characteristics of children aged 3-12 years of different ages and genders, basic motor skills are divided into three categories: movement skills, stability skills, and object manipulation skills, and further divided into 8 sequence stages. The 8 sequence stages are as follows: Stage 1 is the basic movement perception stage, suitable for children aged 3-4 years, focusing on basic limb movement and body perception abilities; Stage 2 is the basic stability establishment stage, suitable for children aged 4-5 years, focusing on static and simple dynamic body stability abilities; Stage 3 is the basic object contact stage, suitable for children aged 5-6 years, focusing on basic hand-object contact and grasping abilities; Stage 4 is the comprehensive movement coordination stage, suitable for children aged 6-7 years, focusing on the ability to complete multi-limb coordinated movement actions; Stage 5 is the dynamic stability control stage, suitable for children aged 7-8 years, focusing on the ability to adjust dynamic body stability in complex scenarios. Stage 6 is the simple object manipulation stage, suitable for children aged 8-9, focusing on simple pushing, throwing, and catching of objects; Stage 7 is the advanced mobile application stage, suitable for children aged 9-10, focusing on the flexible application of advanced mobile skills in different scenarios; Stage 8 is the comprehensive object manipulation stage, suitable for children aged 10-12, focusing on precise and complex comprehensive manipulation of objects. The system clearly defines the age range, gender compatibility differences, and exclusive action patterns for each stage, ensuring that the stage divisions fully align with children's physiological development. Simultaneously, a standard action feature library corresponding to each sequence stage is constructed. This feature library contains feature parameters of action amplitude, joint coordination state, and action completion rhythm for each stage, and is stored internally. Key features of children's actions are extracted and compared with the standard action feature library. Combined with the sequence stage adaptation coefficient calibration comparison results, abnormal action data is eliminated and positioning deviations are corrected, quickly and initially determining the sequence stage range to which the child belongs, significantly reducing the computational workload of the matching calculation unit.

[0050] Matching Calculation Unit: This unit incorporates a built-in action-physiological coordinating sequence matching algorithm processing module. This algorithm decomposes the standardized integrated score across all dimensions into the child's current action feature vector. This vector is then matched against the corresponding feature vector in the standard action feature library stored in the sequence processing unit. An age and gender correction coefficient is embedded to calibrate the matching results, ensuring the matching process fully adapts to individual child differences and improves matching accuracy. The mathematical expression for the action-physiological coordinating sequence matching algorithm is: ;in: For sequence matching accuracy, This is the feature vector of the child's current action. This represents the standard action feature vector corresponding to the sequence stage. The standard deviation of HRV fluctuation is in the millisecond range. The standard deviation of a specific muscle synergy pattern. This refers to the action-physiology weighting coefficient. The system uses age and gender correction coefficients. When the matching accuracy is ≥85%, the specific core sequence stage to which the child belongs is determined. When the matching accuracy is <85%, the child's action features are automatically re-extracted and compared again to ensure accurate sequence stage positioning. After accurate positioning is completed, the system collects specific quantitative parameters corresponding to the sequence stage to obtain data that reflects the developmental details of the child at that stage. At the same time, it marks the lag in the child's action sequence to improve the dimensions of the evaluation data.

[0051] The integrated assessment and early warning unit incorporates a multi-task deep learning model and a multi-dimensional motor development risk early warning algorithm processing module. First, the input layer of the multi-task deep learning model receives multi-dimensional data from the data acquisition unit and the sequence matching results from the matching calculation unit, completing data alignment. The multi-modal feature extraction layer extracts deep features from macroscopic motor parameters, microscopic physiological signals, and cognitive data through three parallel branches. After processing by convolutional, pooling, and fully connected layers, a fixed-dimensional feature vector is obtained, fully mining the deep value of the data. The feature fusion layer uses an attention mechanism to weight and fuse the feature vectors, highlighting key information related to children's motor development. The multi-task output layer calculates a motor skill score that accurately reflects children's motor abilities. Subsequently, the multi-dimensional motor development risk early warning algorithm inputs sequence matching accuracy, sequence stage lag coefficient, physiological-motor correlation factor, and cognitive-motor correlation coefficient. It quantifies and calculates the early warning index and classifies risk levels. The mathematical expression of the multi-dimensional motor development risk early warning algorithm is: ;in: As an early warning index, For sequence matching accuracy, This represents the lag coefficient for the sequence stage. , The cognitive-motor correlation coefficient, As a cognitive adaptation constant, an early warning index ≥4 indicates a high-risk level, an early warning index between 2 and 3.9 indicates a medium-risk level, and an early warning index ≤1.9 indicates a low-risk level. Early risk warnings are generated for children with disordered microphysiological signals and low sequence matching accuracy. Finally, the risk level of motor skill scores and risk warnings are integrated to generate a comprehensive developmental assessment report and early risk warning information, which are then fed back to staff to provide clear and specific reference for subsequent interventions.

[0052] Intervention Optimization Unit: After receiving the comprehensive developmental assessment report and early risk warning information generated by the fusion assessment and early warning unit, this unit automatically distinguishes the child's motor development shortcomings and physical fitness, as well as the linkage between motor and cognitive functions, and generates a personalized intervention plan tailored to the child's individual situation. During the child's subsequent intervention training, this unit continuously records the motor performance data and periodic feedback data to ensure the traceability of the intervention effect. Based on this data, it automatically optimizes the parameters of the device's built-in assessment model, ensuring that the device's assessment capabilities continuously align with the child's actual development, forming a closed-loop system of assessment-intervention-optimization, and continuously improving the quality of the assessment and intervention services provided by the device.

[0053] In summary, this intelligent assessment device synchronously acquires and stores multi-dimensional data through a data acquisition unit. The data processing unit then uses a stage-adaptive multi-source data standardization algorithm to output a standardized integrated score. The sequence processing unit categorizes basic motor skills into three types—mobility skills, stability skills, and object manipulation skills—and divides them into eight sequence stages, constructing a feature library. The matching calculation unit uses a motion-physiology co-matching sequence matching algorithm to accurately locate the child's sequence stage. The fusion assessment and early warning unit generates assessment reports and early warning information using a multi-task deep learning model and a multi-dimensional motor development risk early warning algorithm. The intervention and optimization unit develops personalized plans and continuously optimizes model parameters. All units work collaboratively to ensure efficient and accurate data processing, stage matching, assessment and early warning, and intervention optimization throughout the entire process, providing stable and reliable hardware support for the motor development assessment of children aged 3-12.

[0054] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any simple modifications, equivalent changes and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. A quantitative testing method for basic motor skills in children's developmental assessment, characterized in that, The specific steps of this method are as follows: S1. Full-dimensional data collection and integration: Through a visual capture system and wearable sensing devices, macroscopic motion parameters, multimodal data and microscopic physiological signals are collected simultaneously when children perform basic motor skills; the original multi-source data is fused and processed using a stage-adaptive multi-source data standardization algorithm to obtain a full-dimensional standardized integration score; S2. Sequence Stage Division and Feature Library Construction: Based on the comprehensive standardized integrated score, the basic motor skills of children aged 3-12 years are divided into sequence stages; a standard movement feature library corresponding to each sequence stage is constructed to initially determine the sequence stage range to which the child belongs; S3. Dynamic Sequence Matching and Parameter Acquisition: Using the action-physiology co-sequence matching algorithm, the full-dimensional standardized integrated score obtained in step S1 is decomposed into the child's current action feature vector; it is matched and calculated with the standard action feature library constructed in step S2 to determine the specific sequence stage to which the child belongs; for the determined sequence stage, the corresponding exclusive quantitative parameters are collected, and the action sequence lag is marked. S4. Multimodal fusion assessment and risk warning: Deeply fuse the multi-dimensional data collected in step S1 and the sequence matching results in step S3; Children's motor skill scores are calculated based on a multi-task deep learning model; a comprehensive developmental assessment report and early risk warning information are generated using a multi-dimensional motor development risk warning algorithm. S5. Personalized Intervention and Closed-Loop Optimization: Based on the assessment report and early warning information generated in step S4, generate a personalized intervention plan; Based on the effect data during the intervention process, the parameters of the evaluation model are optimized to form a closed-loop system of evaluation-intervention-optimization.

2. The method for quantitative testing of basic motor skills for children's developmental assessment according to claim 1, characterized in that, In step S1, the specific implementation steps for synchronously collecting data through the visual capture system and wearable sensing device are as follows: deploying the visual capture system to capture the limb movement trajectory of the child when performing basic motor skills, recording the joint rotation angle, the duration of the action, the amplitude of the action, and the time consumed by the action transition, forming macroscopic motion parameters; equipping the child with wearable sensing device to collect the child's muscle group electrical signals, heart rate, heart rate variability fluctuations, and muscle coordination patterns, forming microscopic physiological signals; embedding a lightweight cognitive task during the child's performance of basic motor skills, collecting the child's accuracy rate and reaction time in completing the cognitive task, forming multimodal data; aligning the collected macroscopic motion parameters, microscopic physiological signals, and multimodal data with timestamps, and synchronously transmitting them to the data processing terminal for storage, completing the synchronous collection of multi-source data.

3. The method for quantitative testing of basic motor skills for children's developmental assessment according to claim 1, characterized in that, In step S1, the mathematical expression for the stage-adapted multi-source data normalization algorithm is: ;in To standardize and integrate scores across all dimensions, For the i-th type of raw collected data, , For the localized norm extreme values ​​of the same age and gender for the i-th type of data, For the dynamic weights of the i-th class of data, The baseline fit coefficient is based on the child's age.

4. The method for quantitative testing of basic motor skills for children's developmental assessment according to claim 1, characterized in that, In step S2, the specific content of the sequence stage division is as follows: based on the theory of children's overall motor development sequence, the complete development trajectory of basic motor skills is decomposed; combined with the theory of decomposed motor sequence, the motor development patterns of the upper limbs, trunk, and lower limbs are decomposed; the full-dimensional standardized integration score obtained in step S1 is correlated; the motor characteristics of children aged 3 to 12 years of different ages and genders are statistically analyzed; and basic motor skills are divided into 8 sequence stages after being divided into three categories: movement skills, stability skills, and object manipulation skills.

5. The method for quantitative testing of basic motor skills for children's developmental assessment according to claim 1, characterized in that, In step S2, the specific content of constructing the standard action feature library corresponding to each sequence stage is as follows: the feature library contains feature parameters of action amplitude, joint coordination state, and action completion rhythm for each stage. The key features of the child's actions are extracted and compared with the standard action feature library. The comparison results are calibrated by combining the sequence stage adaptation coefficient. Abnormal action data generated during the collection process are removed, the positioning deviation is corrected, and the range of the sequence stage to which the child belongs is initially determined.

6. The method for quantitative testing of basic motor skills for children's developmental assessment according to claim 1, characterized in that, In step S3, the mathematical expression of the action-physiological coordinating sequence matching algorithm is: ;in: For sequence matching accuracy, This is the feature vector of the child's current action. This represents the standard action feature vector corresponding to the sequence stage. The standard deviation of HRV fluctuation is in the millisecond range. The standard deviation of a specific muscle synergy pattern. This refers to the action-physiology weighting coefficient. This is an age and gender correction factor.

7. The method for quantitative testing of basic motor skills for children's developmental assessment according to claim 1, characterized in that, In step S4, the multi-task deep learning model includes an input layer, a multimodal feature extraction layer, a feature fusion layer, and a multi-task output layer. The input layer receives the multi-dimensional data collected in step S1 and the sequence matching results in step S3 and completes the data alignment. The multimodal feature extraction layer has three parallel branches that extract features from macroscopic motion parameters, microscopic physiological signals, and cognitive data, respectively. Each branch extracts deep features from the data through convolutional and pooling layers, and maps the features into fixed-dimensional feature vectors through fully connected layers. The feature fusion layer uses an attention mechanism to weight and fuse the feature vectors output from the three branches, strengthening the weights of features related to children's motor development and weakening the influence of redundant features. The multi-task output layer includes a motor skill score calculation branch, which processes the fused feature vectors through fully connected layers and normalization layers to calculate the children's motor skill scores.

8. The method for quantitative testing of basic motor skills for children's developmental assessment according to claim 1, characterized in that, In step S4, the mathematical expression of the multi-dimensional motor development risk warning algorithm is: ;in: As an early warning index, For sequence matching accuracy, This represents the lag coefficient for the sequence stage. , The cognitive-motor correlation coefficient, This is the cognitive adaptation constant.

9. A method for quantitative testing of basic motor skills for children's developmental assessment according to claim 1, characterized in that, In step S5, the specific implementation steps of the personalized intervention plan are as follows: based on the assessment report and early warning information generated in step S4, distinguish the child's motor development shortcomings, physical fitness, and motor-cognitive linkage performance, and generate corresponding types of personalized intervention content. To address developmental deficiencies in motor skills, interactive demonstrations were used to design corrective training programs, clearly defining the training movements, duration, and repetitions. Targeted physical fitness training content was developed based on individual fitness levels, specifying training intensity, frequency, and phased goals. Lightweight cognitive-physical integration training tasks were designed to address the interaction between motor skills and cognitive abilities, clearly defining task content and completion requirements. Training completion status, motor performance data, and phased feedback data were recorded throughout the intervention process.

10. An intelligent assessment device for child development, applicable to the basic motor skills quantitative testing method for child development assessment as described in any one of claims 1-9, characterized in that, The device includes: Data acquisition unit: including visual capture system, wearable sensing device and cognitive task embedding module, used to simultaneously collect children's macroscopic motion parameters, microscopic physiological signals and multimodal data, and complete data timestamp alignment and transmission storage; Data processing unit: Built-in stage-adaptive multi-source data standardization algorithm processing module, used to fuse and process the raw multi-source data collected by the data acquisition unit, and output a full-dimensional standardized integrated score; Sequence Processing Unit: Used to combine the full-dimensional standardized integrated score to divide the basic motor skills of children aged 3-12 into 8 sequence stages, build and store the standard movement feature library corresponding to each sequence stage, and initially determine the range of the sequence stage to which the child belongs; Matching Calculation Unit: Built-in action-physiology coordinated sequence matching algorithm processing module, used to decompose the full-dimensional standardized integrated score into the current action feature vector, perform matching calculation with the standard action feature library, determine the specific sequence stage of the child and collect exclusive quantitative parameters; Fusion Assessment and Early Warning Unit: Built-in multi-task deep learning model and multi-dimensional motor development risk early warning algorithm processing module, used to deeply fuse multi-dimensional data and sequence matching results, calculate motor skill scores, and generate comprehensive development assessment reports and early risk warning information; Intervention Optimization Unit: Used to generate personalized intervention plans based on comprehensive developmental assessment reports and early risk warning information, record intervention process data, and optimize assessment model parameters to form a closed-loop optimization.

Images