Apparatus and method for evaluating children's writing disorder based on writing stylus timing information

Abstract

This invention relates to the field of automated assessment technology for children's writing disorders, and discloses a device and method for assessing children's writing disorders based on writing stylus timing information. The device includes a writing board with a power switch mounted through its surface and an electromagnetic pen slot through one side of the power switch. This invention achieves high-rate synchronous capture of pen tip position, X and Y coordinates, real-time pressure Z, and microsecond-level timestamps during the writing process using electromagnetic induction technology, constructing a complete spatiotemporal sequence of writing motion. This technological foundation overcomes the fundamental deficiency of traditional methods, which can only analyze static writing results, such as character shape and neatness, but cannot capture the details of the dynamic process. Based on this raw data stream, this invention systematically extracts at least 30 quantitative features covering six dimensions: time, speed, spatial geometry, pressure dynamics, behavioral norms, and character matching. The time dimension not only calculates the total duration but also analyzes the distribution and frequency of pauses.

Description

Technical Field

[0001] This invention relates to the field of automated assessment technology for children's writing disorders, specifically to a device and method for assessing children's writing disorders based on writing stroke timing information. Background Technology

[0002] Developmental Coordination Disorder (DCD) is a developmental disorder characterized by severe impairment of motor coordination, with onset in childhood. Clinically, DCD manifests as impaired motor control and executive function, significantly impacting daily activities and academic performance. The prevalence of DCD in school-aged children is 5%–6%, with 50%–70% of affected children experiencing motor impairments that persist into adolescence and even adulthood. According to the Diagnostic and Statistical Manual of Mental Disorders (DSM-5), writing difficulties are a key criterion for assessing DCD. Compared to alphabetic writing systems, Chinese characters have a higher incidence of writing difficulties due to their more complex structure, requiring greater precision in stroke quality, stroke order, radical structure, and pen pressure. Inadequate writing skills not only affect children's academic performance and school participation, but also negatively impact their self-awareness and self-efficacy. Currently, the mainstream assessment method for writing disorders is still scale-based assessment, such as the Detailed Assessment of Speed ​​of Handwriting (DASH). DASH is a norm-referenced test that can standardize the assessment of writing speed in children aged 9-16 and provide criteria for judging handwriting legibility. DASH is the only tool that can assess the speed of different writing tasks. In 2010, Barnett et al. further developed DASH 17+ to assess the writing ability of adolescents aged 17-25. Another example is the Concise Evaluation Scale for Children's Handwriting (BHK). BHK is used to assess writing quality and speed, serving as both a handwriting quality screening tool and the gold standard for diagnosing Latin alphabet writing disorders. BHK currently offers versions in Dutch, English, German, French, and Italian. Its operating guidelines and norms are derived from the original Dutch version. The handwriting quality norms are applicable to second-grade children (7-8 years old) and third-grade children (8-9 years old), while the writing speed norms are applicable to children in grades one through six (6.5-12.5 years old). Another example is the Children's Handwriting Assessment Tool - Manuscript / Cursive (ETCH-M / C). ETCH is a standardized assessment tool based on the Latin alphabet, designed to assess children's handwriting legibility, writing speed, hand function, and ability to operate writing tools. ETC is suitable for children aged 6-12 and includes a handwriting version (ETCH-M) and a cursive version (ETCH-C).The ETCH-M test includes six writing tasks similar to classroom activities: (1) writing uppercase and lowercase letters; (2) writing numbers; (3) close-range copying; (4) long-range copying; (5) dictation; and (6) sentence construction. Based on this, ETCH-C adds a "handwriting to cursive" task. The entire process lasts approximately 30 minutes. New technologies are also attempting to automate the assessment of writing difficulties using digital technology. For example, the EyeandPen (EP) device can synchronously record the writing process and eye movements (through an optical eye-tracking system) with a digital tablet. Children write on the tablet, which collects the pen point state (surface pressure level) and spatial and temporal data of the pen's movement on the plane. Built-in software analyzes this data to determine the execution speed (cm / s, representing the writing speed) and writing pause time. The combination of a writing tablet and an eye-tracking system can reveal changes in visual information during writing and pauses. Another example is the Computerized Chinese Handwriting Assessment Tool (CHAT). CHA was developed based on parameters (such as air time, speed, and pen pressure) measured by the early computer-based handwriting assessment tool Penmanship Objective Evaluation Tool (POET). In addition to the parameters measured by POET, CHAT can also measure Chinese character errors, such as stroke errors, stroke order, and accuracy. It consists of a graphics tablet used with a ballpoint pen, capturing handwriting data such as stroke order and pressure applied to the tablet when the user writes on grid paper. CHAT templates contain 90 commonly used Chinese characters. The writing process and performance are evaluated by CHAT. For example, an AI-based method and system for detecting writing disorders in children collects raw handwriting data and corresponding environmental parameters and emotional characteristics to obtain segmented correlated datasets. The data is analyzed based on behavioral characteristics, and segmented handwriting information is combined with emotional measurements for adaptive modeling. Abnormal data is labeled to ultimately determine whether a writing disorder exists. Therefore, a device and method for assessing children's writing disorders based on writing touch timing information are proposed.

[0003] However, in the process of realizing this invention, the inventors discovered that at least the following problems remain unresolved in the existing technology: 1. High assessment cost: The aforementioned scale-based scheme requires assessment personnel to evaluate each writing sample sequentially according to the scale items, resulting in high time and manpower costs. Assessment personnel also require professional training, further increasing costs and making large-scale application difficult. 2. Feature loss: As a paper-and-pencil test, the scale assessment cannot capture dynamic features during the writing process, such as changes in pen pressure, stroke trajectory, and pause time. Therefore, it cannot deeply analyze children's performance and potential obstacles during the writing process. Scoring and interpretation are subjective. Although the scale provides assessment standards, the scoring of handwriting legibility is easily influenced by individual expert insight, reaction bias, and comprehension differences, resulting in strong subjectivity. 3. Inconvenient support for Chinese character writing assessment: Existing writing assessment scales are mainly applicable to Latin letters, and assessment scales for Chinese character writing are still relatively lacking, limiting their application in Chinese writing assessment and writing disorder diagnosis. Children are suitable for Chinese character adaptation. Therefore, this invention designs a device and method for assessing children's writing disorders based on writing stroke timing information. Summary of the Invention

[0004] The purpose of this invention is to address the shortcomings of existing technologies, such as: high assessment costs (the aforementioned scale-based schemes require assessment personnel to evaluate each writing sample sequentially according to the scale items, resulting in high time and labor costs; assessment personnel require professional training, further increasing costs and hindering large-scale application); feature loss (as a paper-and-pencil test, scale assessments cannot capture dynamic features during the writing process, such as changes in pen pressure, stroke trajectory, and pause time, thus failing to deeply analyze children's performance and potential obstacles during writing); and subjectivity in scoring and interpretation (although the scales provide assessment standards, scores on handwriting legibility are easily influenced by individual expert insight, reaction bias, and comprehension differences, resulting in strong subjectivity); and limited support for Chinese character writing assessment (existing writing assessment scales are mainly applicable to Latin letters, and scales for Chinese character writing assessment are still relatively lacking, limiting their application in Chinese writing assessment and writing disorder diagnosis). Therefore, this invention designs a device and method for assessing children's writing disorders based on writing stroke timing information.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: a device for assessing children's writing difficulties based on writing stylus timing information, including a writing board, a power switch is installed through the surface of the writing board, an electromagnetic pen placement slot is provided through one side of the power switch, and a display screen is installed through one side of the electromagnetic pen placement slot.

[0006] The electromagnetic pen body is movably installed inside the electromagnetic pen placement slot. An electromagnetic pen charging port is installed through the electromagnetic pen body. A pen refill is fixedly installed inside the electromagnetic pen body.

[0007] The handwriting tablet has a charging interface installed through its interior, and a paper media placement area is provided through its interior surface. A paper media clip is movably installed on the surface of the handwriting tablet.

[0008] Preferably, a grid-patterned A4 sheet of paper is movably installed inside the paper medium buckle, and the inside of the A4 sheet of paper is permeated with grid lines.

[0009] Preferably, a cross-shaped A4 sheet of paper is movably installed at the bottom of the grid-shaped A4 sheet of paper, and the cross-shaped A4 sheet of paper has cross-shaped grids running through its interior.

[0010] Preferably, the handwriting tablet is based on active electromagnetic induction technology, uses dot matrix codes, and is equipped with an ultra-low power SOC to achieve wireless transmission, a standardized module for children's writing content, a data service terminal, and evaluation software.

[0011] Preferably, the electromagnetic pen body is based on active electromagnetic induction technology and has a dedicated pen tip for writing on a writing medium.

[0012] Preferably, the A4 paper used as the rice-shaped writing medium is placed on the surface of the writing board to provide a writing environment consistent with children's actual pen and paper writing, and the A4 paper used as the rice-shaped writing medium is made of paper material.

[0013] Preferably, the children's writing content standardization module configures corresponding writing task content according to the child's age or grade, including words, sentences or short passages, to form standardized writing requirements for different learning stages.

[0014] Preferably, the data service terminal connects to the handwriting tablet via wireless technology and runs the evaluation software.

[0015] Preferably, the assessment software receives writing stroke timing data during the child's writing process, calculates at least 30 writing behavior features through high-precision timestamp annotation, multi-dimensional feature extraction and intelligent modeling, and automatically generates a comprehensive writing disorder assessment report covering at least six dimensions of features, including time, speed, spatial geometry, spatial dynamics, behavioral features and character matching, and a weighted comprehensive score.

[0016] A method for assessing writing difficulties in children based on writing stroke timing information includes the following steps:

[0017] S1. Data Acquisition: Based on the digital handwriting tablet, the dynamic acquisition of children's continuous writing content during copying tasks is realized. The data specifically includes: continuous coordinate positioning of the pen tip on the two-dimensional plane and pressure intensity recording of coordinate synchronization; millisecond-level time stamps generated by the system clock, ultimately forming a structured time-series dataset.

[0018] S2. Data Preprocessing and Structured Modeling: Based on morphological analysis technology, the continuous writing pen stroke sequence data is segmented and processed, and the original writing content is broken down into individual Chinese characters and their corresponding stroke units, thereby forming a structured data representation with characters and strokes as the basic analysis objects. This decomposition method enables writing behavior to correspond to the shape structure and stroke organization of Chinese characters, providing a clear structural basis for the extraction of writing disorder-related features.

[0019] S3. Feature Engineering Processing: Combining the original pen stroke timing information with morphological segmentation results, writing indicators are extracted from six dimensions: time features, spatial features, speed features, geometric features, dynamic features, and character structure features. This forms a feature set containing more than thirty writing features. These features are used to characterize the abnormal writing rhythm, insufficient stroke control, character structure deviation, and writing disorder-related behavioral characteristics that children may exhibit during the writing process.

[0020] Compared with the prior art, the beneficial effects of the present invention are:

[0021] 1. This invention employs electromagnetic induction technology to synchronously capture the pen tip position, X and Y coordinates, real-time pressure Z, and microsecond-level timestamps during the writing process at a high sampling rate, constructing a complete spatiotemporal sequence of writing motion. This technological foundation overcomes the fundamental deficiency of traditional methods, which can only analyze static writing results, such as character shape and neatness, but cannot capture the details of the dynamic process. Based on this raw data stream, the system systematically extracts at least 30 quantitative features covering six dimensions: time, speed, spatial geometry, pressure dynamics, behavioral norms, and character matching. The time dimension not only calculates the total duration but also analyzes the distribution and frequency of pauses; the speed dimension not only calculates the mean but also analyzes the acceleration curve and peak distribution. This multi-dimensional dynamic feature system overcomes the limitations of existing technologies, such as single feature dimensions and insufficient utilization of temporal information, enabling the assessment to more comprehensively and realistically reflect the complex motion control and coordination mechanisms behind writing difficulties. By inputting the extracted multi-dimensional features into an optimized... The machine learning pipeline first selects the most discriminative subset of core features, such as character width variation coefficient and dynamic pressure standard deviation. Then, it uses machine learning models for modeling and prediction. This process is entirely data- and algorithm-driven, directly generating structured assessment reports and completely eliminating reliance on subjective expert ratings. The model can automatically identify "poor pressure stability" and "low spatial consistency" as the main risk factors for a given case. Compared to existing solutions that rely on manual observation or simple statistical analysis, it fundamentally solves the problems of "subjectivity bias" and "human dependence" in assessments, significantly improving the objectivity, consistency, and reproducibility of assessment results. By using fine-grained data and algorithmic modeling, it solves the problems of "subjectivity bias" and "human dependence," significantly improving assessment efficiency. The assessment results show good consistency and reliability with expert ratings, with a Cronbach's alpha coefficient of 0.7, demonstrating its advantage as an alternative to manual assessment methods.

[0022] 2. This invention employs a streaming processing architecture that enables simultaneous feature calculation and model inference during data acquisition. Experimental verification shows that within seconds of a child completing a writing task, the system can automatically generate a preliminary assessment report containing risk levels and multi-dimensional analysis. Compared to the lengthy process of traditional methods involving offline questionnaire completion, manual data entry, and subsequent analysis, which typically takes hours or even days, this represents a significant efficiency improvement. This "instant assessment - instant feedback" capability allows the invention to seamlessly integrate into classroom teaching or clinical screening scenarios, providing educators or therapists with real-time decision-making support to seize the optimal intervention opportunity. Through machine learning, the system can clearly demonstrate the specific contribution and direction of each feature to the final assessment conclusion, with character structure similarity negatively predicting the matching degree with a weight of -0.12. The report, reflecting that structural deviations may lead to rigid handwriting, clearly states: "The coefficient of variation of speed, with a contribution of +0.15, is the most significant factor contributing to the increased risk in this assessment; the higher the value, the stronger the coordination disorder." Furthermore, by comparing with large-sample norm data, the system can provide intuitive quantitative interpretations such as "Your stress fluctuation index is higher than 85% of children of the same age." This transparent analysis method is completely different from traditional "black box" models or methods that only provide a vague total score. It provides solid mathematical and logical support for the assessment conclusions, allowing both the assessor and the assessed to clearly understand the problem. This provides a precise "roadmap" for developing highly targeted personalized intervention plans. The ICC consistency index has significantly improved from 0.65 to over 0.85, and it supports the assessor in proposing improvement suggestions for specific indicators.

[0023] 3. This invention achieves full automation by automatically capturing "writing trajectory data," seamlessly extracting and intelligently analyzing the features of "process data" in the background, and finally generating a "professional assessment report" with one click. This closed-loop technology ensures high-fidelity transformation from raw behavior to assessment insights, minimizing errors and delays that may be introduced by human intervention. It not only completes a one-time assessment, but its structured data output also makes it possible to establish individualized long-term writing development files, track intervention effects, and conduct longitudinal comparative studies. This truly realizes continuous and intelligent monitoring and management of children's writing abilities. The content is standardized and selected according to the primary school Chinese textbook system to ensure that the writing content meets the writing learning requirements of the corresponding age group. The standardization of the writing content allows for comparison with norms, accurately characterizing children's writing problems. The technology relies on machine learning rather than deep learning and artificial intelligence, and the results are highly interpretable, providing educational significance for subsequent improvement of children's writing. This method is not only applicable to the detection of writing problems in normal children aged 3-12, but also to children with autism spectrum disorder, attention deficit disorder, and other special needs children. Attached Figure Description

[0024] Figure 1 This is a schematic diagram of the overall structure of the device for assessing children's writing difficulties based on writing stylus timing information proposed in this invention;

[0025] Figure 2 This is a partial perspective view of the electromagnetic pen used in the device for assessing children's writing difficulties based on writing stylus timing information proposed in this invention.

[0026] Figure 3 This is a schematic diagram of the paper medium—A4 paper with grid lines—of the device for assessing children's writing difficulties based on writing stylus timing information proposed in this invention;

[0027] Figure 4 This is a schematic diagram of the paper medium—a grid A4 paper—used in the device for assessing children's writing difficulties based on writing stylus timing information proposed in this invention.

[0028] In the diagram: 1. Handwriting tablet; 101. Power switch; 2. Electromagnetic pen slot; 3. Display screen; 4. Charging port; 5. Paper media storage area; 6. Paper media clip; 7. Electromagnetic pen body; 701. Electromagnetic pen charging port; 8. Pen refill; 9. A4 paper for grid paper; 10. Grid paper; 11. A4 paper for cross-shaped paper; 12. Cross-shaped paper. Detailed Implementation

[0029] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0030] like Figures 1-4 As shown, the present invention provides an apparatus and method for assessing children's writing difficulties based on writing stylus timing information, mainly as follows:

[0031] Step 1: High-precision stylus timing information acquisition and structuring

[0032] When children write on the digital writing tablet using a special electromagnetic pen, the system continuously and in real time collects complete spatiotemporal information of the writing motion at a sampling frequency of 100Hz: 100 times per second; each collection generates a structured data point containing precise information in five dimensions:

[0033] Coordinate values, x, y: precisely record the instantaneous two-dimensional position of the pen tip on the writing plane.

[0034] Pressure value: The contact pressure between the pen tip and the board surface is quantified in the range of 0-1024. Among them, the value 0 indicates that the pen tip is not in contact with the board surface and is moving in the air, and the value 1-1024 represents the contact pressure. The higher the pressure, the higher the value.

[0035] Timestamp: Marks the precise system time at which each data point was captured, with millisecond-level precision.

[0036] Stroke ID: Based on the logic of "pressure value is continuously greater than 0", the system automatically divides the continuous data stream into independent stroke units and assigns a stroke number to each data point.

[0037] Character ID: Through morphological analysis and spatial clustering algorithms, the system automatically identifies and segments each character in the written content, and assigns a unique character number to all data points belonging to the same character space range.

[0038] The data point information is shown in Table 1:

[0039] Table 1

[0040] X ^ y pressure timestamp Stroke_id <![CDATA[Char _ id]]> 0 3981.9436 22906.695 0 1760455341081 0 1 8687.05 -2716.9055 871 1760455350561 1 2.0 2 8686.02 -2713.9158 866 1760455350563 1 2.0 3 8686.988 -2710.9058 868 1760455350564 1 2.0 4 8686.979 -2709.906 875 1760455350564 1 2.0 5 8688.958 -2707.8857 874 1760455350564 1 2.0 6 8690.938 -2705.8652 881 1760455350564 1 2.0 7 8694.888 -2700.825 886 1760455350564 1 2.0 8 8699.836 -2695.775 889 1760455350564 1 2.0 9 8706.785 -2690.704 893 1760455350564 1 2.0 10 8714.734 -2685.623 890 1760455350565 1 2.0 11 8721.683 -2680.5525 893 1760455350565 1 2.0 12 8731.622 -2674.4512 892 1760455350565 1 2.0 13 8740.56 -2668.36 890 1760455350565 1 2.0 14 8749.509 -2663.2693 895 1760455350565 1 2.0 15 8757.447 -2657.1885 895 1760455350565 1 2.0 16 8766.396 -2652.0974 896 1760455350565 1 2.0 17 8774.345 -2647.0166 898 1760455350565 1 2.0 18 8781.305 -2642.9458 899 1760455350565 1 2.0 19 8787.273 -2639.885 901 1760455350565 1 2.0 20 8791.243 -2636.8447 895 1760455350565 1 2.0 21 8795.223 -2634.8042 896 1760455350565 1 2.0 22 8798.202 -2632.774 892 1760455350565 1 2.0 23 8800.191 -2631.7537 887 1760455350565 1 2.0 24 8801.182 -2630.7434 868 1760455350565 1 2.0 25 8802.172 -2629.7334 832 1760455350565 1 2.0 26 8802.172 -2629.7334 773 1760455350565 1 2.0

[0041] Step 2: Multi-dimensional Handwriting Feature Analysis and Quantization Calculation. Based on the raw time-series data obtained in Step 1, the system automatically analyzes and calculates quantized features covering six core dimensions—time, speed, spatial geometry, pressure dynamics, behavioral norms, and character matching—using its built-in feature engineering engine. The total number of features is no less than 30. Specifically, these include:

[0042] Time dimension: Analyze total writing time, percentage of effective writing time, and distribution and frequency of pauses.

[0043] Velocity dimension: Instantaneous velocity and acceleration are calculated by differentiating coordinates and timestamps, and then the average velocity, velocity variation coefficient, and peak acceleration kinematics are analyzed.

[0044] Spatial geometric dimensions: quantifying stroke curvature, character aspect ratio, structural proportion deviation, and spatial accuracy of stroke connections.

[0045] Pressure dynamics dimension: Calculate mean pressure, standard deviation of pressure fluctuation, and pressure-velocity coupling relationship.

[0046] Behavioral norms dimension: Detecting stroke order compliance, frequency and pattern of abnormal pen lifting / tracing actions.

[0047] Character matching dimension: The dynamic time warping algorithm is used to calculate the similarity and deviation between the writing trajectory and the standard template in terms of overall shape and key point position.

[0048] Step 3: Intelligent Analysis and Structured Report Generation. The multidimensional features calculated in Step 2 are input into a pre-trained machine learning evaluation model for intelligent analysis.

[0049] Feature selection and fusion: The model first selects the most discriminative feature subset through algorithms, and may perform feature fusion to construct a more representative higher-order index.

[0050] Prediction and Modeling: Using machine learning models, based on optimized features, comprehensive calculations are performed to achieve automated prediction of writing disorder risk and graded assessment of severity.

[0051] Automated report output: The system integrates the model prediction results with detailed quantitative data for each dimension, automatically generating a report within seconds. The report displays the child's specific performance in each dimension (time, speed, and space) in the form of charts and scores, and compares it with the norm, providing an overall assessment conclusion, including "suspected writing coordination disorder," risk level, analysis of major abnormal dimensions, and data-driven personalized intervention suggestions.

[0052] Process advantages: The entire assessment process achieves a fully automated closed loop from data collection, feature extraction, intelligent analysis to report output, without the need for manual intervention. It can be completed within seconds after the writing task is finished, ensuring the efficiency, objectivity, standardization and timeliness of the assessment, making it perfectly suitable for educational screening and clinical assessment scenarios with high timeliness requirements.

[0053] To verify the effectiveness of the proposed writing disorder assessment method based on writing stroke timing information, a systematic empirical study was conducted in a real primary school teaching setting. This study, using children's daily writing learning activities as a background, validated the identification effect of writing disorders by combining standardized copying tasks, manual scale assessment, and machine learning modeling.

[0054] Research Subjects and Experimental Setting: The empirical study was conducted in a normal primary school teaching environment. The research subjects were primary school students of different grades, with a sample of 160 students covering grades one through six, to ensure that the research results could reflect the writing performance characteristics of children at different stages of writing development. All experimental procedures were completed within the time allowed by classroom or teaching activities, and the writing tasks were consistent with daily teaching requirements, thereby reducing the interference of the experimental environment on children's writing behavior.

[0055] Writing Task Design and Data Collection: Addressing the differences in writing proficiency among students of different grades, this study, based on the primary school Chinese language textbook system, selected 30 Chinese characters from the People's Education Press textbook for each grade level as copying materials according to predetermined selection rules. The selected characters were representative in terms of structural type, number of strokes, and frequency of use, covering the main character shape characteristics encountered in children's writing learning.

[0056] During the experiment, the participants used a dedicated writing tablet to complete the copying task. This device simultaneously records the timing information of the children's pen strokes (coordinates, writing pressure, and timestamps) during the writing process, thus obtaining complete temporal data of the writing process. All children completed the copying task according to standardized writing requirements to ensure the standardization and comparability of the data collection process.

[0057] The assessment process for dyslexia: After completing the copying task, the study invited five graduate students majoring in education with a background in Chinese language education to independently assess the children's writing results. The assessment was based on dyslexia-related scales, and the assessment content covered multiple aspects such as character structure, stroke standardization, writing coherence, and overall legibility.

[0058] Five assessors scored the same writing sample, and the scores were used to characterize the degree of writing difficulties in children. During model development, the manual scale scores were used as a reference standard and supervised learning label to guide the machine learning model in learning the correspondence between different writing characteristics and the degree of writing difficulties.

[0059] Data modeling and validation methods: Based on the collected writing stylus timing data, morphological analysis is first used to segment the writing trajectory to obtain data representations at the single character and stroke levels. On this basis, more than thirty writing features are extracted from six dimensions—time, space, velocity, geometry, dynamics, and character structure—to form a feature dataset for modeling.

[0060] Using human dyslexia scale scores as supervised labels, a machine learning model was constructed to predict the severity of dyslexia in children. The model's ability to identify dyslexia and its stability were evaluated by comparing the prediction results with those of human assessments.

[0061] Experimental Results and Effects: Based on sample data from 160 students aged 7-13 years in grades 1-6, the experimental results show that the machine learning model constructed based on the above features can predict the scoring results of the human writing disorder scale well, and the model output shows a high degree of consistency with human assessment. Compared with analysis methods based on only a single writing indicator, the multi-dimensional writing feature modeling approach has a significant advantage in the accuracy of writing disorder identification.

[0062] The evaluation indicators are shown in Table 2:

[0063] Table 2

[0064]

[0065] Based on the completed prediction of scores for each sub-dimension of the Artificial Writing Disorder Scale, the key prediction results of writing disorders were further analyzed. First, multiple scoring dimensions of the artificial writing disorder scale were used as a comprehensive representation of the degree of children's writing disorders to construct a comprehensive writing disorder score prediction model. Experimental results show that, based on joint modeling of six dimensions and more than thirty writing features, the model has high consistency in predicting the comprehensive writing disorder score, with a determination coefficient of 0.84, a mean absolute error of 0.42, and a mean squared error of 0.21, indicating that the model can accurately reflect the overall level of children's writing disorders.

[0066] The predicted target information is shown in Table 3:

[0067] Table 3

[0068]

[0069] Furthermore, to verify the model's application capability in the context of dyslexia screening, children were divided into two categories based on the scoring thresholds of a manual dyslexia scale: those with dyslexia risk and those without significant dyslexia risk. Risk discrimination analysis was then performed on the model's output. The results showed that the model achieved good classification performance in the dyslexia risk discrimination task, with an overall accuracy of 85%, sensitivity of 83%, specificity of 88%, and an AUC of 90%, indicating that the model possesses good dyslexia risk screening capability while maintaining high recognition accuracy.

[0070] The evaluation numerical information is shown in Table 4:

[0071] Table 4

[0072]

[0073] Empirical Research Conclusions: Through empirical research in real teaching scenarios, this invention verifies that the proposed dyslexia assessment method can objectively predict children's dyslexia without relying on environmental factor attribution or behavioral abnormality detection mechanisms. This method uses a manual dyslexia scale as a reference standard, combined with children's actual writing data, providing an operable technical solution for the screening and teaching-aided assessment of dyslexia.

[0074] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A device for assessing children's writing difficulties based on writing stylus timing information, comprising a writing tablet (1), characterized in that, A power switch (101) is installed through the surface of the handwriting tablet (1), an electromagnetic pen placement slot (2) is provided through one side of the power switch (101), and a display screen (3) is installed through one side of the electromagnetic pen placement slot (2). The electromagnetic pen body (7) is movably installed inside the electromagnetic pen placement slot (2), the electromagnetic pen charging port (701) is installed through the electromagnetic pen body (7), and the pen refill (702) is fixedly installed inside the electromagnetic pen body (7). The handwriting tablet (1) has a charging interface (4) installed through its interior, a paper medium placement area (5) is provided through its interior surface, and a paper medium clip (6) is movably installed on the surface of the handwriting tablet (1).

2. The device for assessing children's writing difficulties based on writing stylus timing information according to claim 1, characterized in that, The paper media buckle (6) is movably installed with a grid-patterned A4 paper (9), and the inside of the grid-patterned A4 paper (9) is provided with grid-patterned grids (10).

3. The device for assessing children's writing difficulties based on writing stylus timing information according to claim 2, characterized in that, The bottom of the grid-patterned A4 paper (9) is movably fitted with a rice-patterned A4 paper (11), and the inside of the rice-patterned A4 paper (11) is permeated with rice-patterned grids (12).

4. The device for assessing children's writing difficulties based on writing stylus timing information according to claim 1, characterized in that, The handwriting tablet (1) is based on active electromagnetic induction technology, uses dot matrix code, and is equipped with an ultra-low power SOC to realize wireless transmission, a standardized module for children's writing content, a data service terminal, and evaluation software.

5. The device for assessing children's writing difficulties based on writing stylus timing information according to claim 1, characterized in that, The electromagnetic pen body (7) is based on active electromagnetic induction technology and has a dedicated pen core for writing on writing media.

6. The device for assessing children's writing difficulties based on writing stylus timing information according to claim 5, characterized in that, The rice-shaped paper medium A4 paper (11) is placed on the surface of the writing board to provide a writing environment consistent with children's real paper and pen writing. The rice-shaped paper medium A4 paper (11) is a paper material.

7. The device for assessing children's writing difficulties based on writing stylus timing information according to claim 4, characterized in that, The standardized module for children's writing content configures corresponding writing tasks based on children's age or grade, including words, sentences, or short passages, to form standardized writing requirements for different learning stages.

8. The device for assessing children's writing difficulties based on writing stylus timing information according to claim 4, characterized in that, The data service terminal connects to the handwriting tablet via wireless technology and runs the evaluation software.

9. The device for assessing children's writing difficulties based on writing stylus timing information according to claim 8, characterized in that, The assessment software receives writing stroke timing data of children during the writing process, calculates at least 30 writing behavior features through high-precision timestamp annotation, multi-dimensional feature extraction and intelligent modeling, and automatically generates a comprehensive writing disorder assessment report covering at least six dimensions of features, including time, speed, spatial geometry, spatial dynamics, behavioral features and character matching, and a weighted comprehensive score.

10. A method for assessing children's writing difficulties based on writing stylus timing information, based on the apparatus for assessing children's writing difficulties based on writing stylus timing information as described in any one of claims 1-9, characterized in that, Includes the following steps: S1. Data Acquisition: Based on the digital handwriting tablet, the dynamic acquisition of children's continuous writing content during copying tasks is realized. The data specifically includes: continuous coordinate positioning of the pen tip on the two-dimensional plane and pressure intensity recording of coordinate synchronization; millisecond-level time stamps generated by the system clock, ultimately forming a structured time-series dataset. S2. Data Preprocessing and Structured Modeling: Based on morphological analysis technology, the continuous writing pen stroke sequence data is segmented and processed, and the original writing content is broken down into individual Chinese characters and their corresponding stroke units, thereby forming a structured data representation with characters and strokes as the basic analysis objects. This decomposition method enables writing behavior to correspond to the shape structure and stroke organization of Chinese characters, providing a clear structural basis for the extraction of writing disorder-related features. S3. Feature Engineering Processing: Combining the original pen stroke timing information with morphological segmentation results, writing indicators are extracted from six dimensions: time features, spatial features, speed features, geometric features, dynamic features, and character structure features. This forms a feature set containing more than thirty writing features. These features are used to characterize the abnormal writing rhythm, insufficient stroke control, character structure deviation, and writing disorders that children may exhibit during the writing process.

Images