Summary: Researchers have developed an AI-powered handwriting analysis tool that could transform how educators and clinicians screen K–5 children for dyslexia and dysgraphia. By evaluating handwriting samples, the system detects behavioral indicators, spelling errors, motor patterns, and cognitive markers with strong accuracy, offering a faster, more scalable alternative to traditional screening methods.
Traditional screening methods for reading and writing disorders are often time-consuming, costly, and typically focused on a single condition. This AI approach is designed to be efficient, adaptable, and practical for use in classrooms and clinical settings, potentially helping to relieve pressure on an overburdened speech-language pathology and occupational therapy workforce and improving access to early intervention, particularly in underserved communities.
Key Facts:
- Multimodal analysis: The tool evaluates visual, motor, and cognitive elements of handwriting to create a rounded assessment.
- Early detection: It can flag signs of dyslexia and dysgraphia before they lead to significant academic or social-emotional challenges.
- Accessible screening: The method is designed to scale in school settings, supporting teachers, SLPs, OTs, and special educators where specialist resources are limited.
Source: University at Buffalo
A University at Buffalo-led study details how AI-enhanced handwriting analysis could serve as an early screening tool for dyslexia and dysgraphia in young children.
Published in the journal SN Computer Science, the study describes a co-designed framework that builds on established handwriting recognition methods while addressing practical classroom needs. The authors emphasize making screening faster and more comprehensive by combining multiple indicators in a single assessment tool.
The research team notes the potential for this technology to address the nationwide shortage of speech-language pathologists and occupational therapists, professionals who are critical to diagnosing and treating dyslexia and dysgraphia. By providing an efficient first-line screening tool, AI may enable earlier referral and targeted support.
“Catching these neurodevelopmental disorders early is critically important to ensuring that children receive the help they need before it negatively impacts their learning and socio-emotional development,” says Venu Govindaraju, PhD, SUNY Distinguished Professor in the Department of Computer Science and Engineering at UB, and the study’s corresponding author. “Our goal is to streamline and expand early screening tools so they are available in classrooms and communities that currently lack access.”
This work is part of the National AI Institute for Exceptional Education, a UB-led research initiative focused on developing AI systems that identify and support young children with speech and language processing challenges.
Building on established handwriting recognition research
Govindaraju and collaborators have a long history of applying machine learning and natural language processing to handwriting analysis—efforts that previously contributed to large-scale automation such as postal mail sorting. The new research adapts similar AI techniques to identify spelling errors, poor letter formation, disorganized writing, and other signs associated with dyslexia and dysgraphia.
Previous AI research has concentrated more on dysgraphia, which often produces overt motor differences visible in handwriting. Dyslexia can be harder to detect through handwriting alone because it primarily affects reading and language processing; however, patterns like frequent misspellings and specific writing behaviors can provide important clues.
The researchers also highlight a limiting factor for model development: a relative scarcity of child handwriting datasets suitable for training robust AI systems, which motivates their data collection efforts.
Collecting real-world samples from K–5 students
To ensure the models are practical for classroom use, the UB team consulted teachers, speech-language pathologists (SLPs), and occupational therapists (OTs) during design. “It is critically important to examine these issues and build AI-enhanced tools from the end users’ standpoint,” says co-author Sahana Rangasrinivasan, a PhD student in UB’s Department of Computer Science and Engineering.
The study team worked with co-author Abbie Olszewski, PhD, an associate professor in literacy studies at the University of Nevada, Reno, who helped develop the Dysgraphia and Dyslexia Behavioral Indicator Checklist (DDBIC). The DDBIC catalogs overlapping behavioral signs that can appear before, during, and after writing tasks.
Paper and tablet handwriting samples were gathered from kindergarten through fifth-grade students at an elementary school in Reno under an approved ethics protocol. The dataset was anonymized to protect student privacy. Researchers will use these samples to validate and refine the DDBIC, train AI models to perform DDBIC-style screening, and compare automated assessments to those administered by human experts.
AI features designed for practical screening
The study outlines how AI models can be combined to create a comprehensive screening suite that analyzes multiple handwriting dimensions, including:
- Motor characteristics such as writing speed, pressure, and pen trajectory to detect physical or coordination difficulties.
- Visual handwriting aspects like letter size, slant, and spacing to identify formation problems.
- Optical character recognition (OCR) and language analysis to spot misspellings, letter reversals, and transcription errors.
- Linguistic and cognitive markers, including grammar usage and vocabulary, to reveal deeper language processing or cognitive concerns.
The researchers describe a unified tool that synthesizes outputs from these models, summarizes findings, and provides a concise, actionable assessment for educators and clinicians. They stress that the framework is intended to support—not replace—professional judgment, making screening more efficient and informing decisions about further evaluation or intervention.
“This ongoing work demonstrates how AI can serve the public good by delivering tools and services to communities and practitioners who need them most,” says study co-author Sumi Suresh, PhD, a visiting scholar at UB.
Additional contributors to the study include Bharat Jayarman, PhD, director of the Amrita Institute of Advanced Research and professor emeritus in UB’s Department of Computer Science and Engineering, and Srirangaraj Setlur, principal research scientist at the UB Center for Unified Biometrics and Sensors.
About this AI and dyslexia research news
Author: Cory Nealon
Source: University at Buffalo
Contact: Cory Nealon, University at Buffalo
Image credit: Neuroscience News
Original Research (open access): “AI-Enhanced Child Handwriting Analysis: A Framework for the Early Screening of Dyslexia and Dysgraphia” by Venu Govindaraju et al., published in SN Computer Science.
Abstract
AI-Enhanced Child Handwriting Analysis: A Framework for the Early Screening of Dyslexia and Dysgraphia
Dyslexia and dysgraphia are two common specific learning disabilities that affect children’s academic progress and socio-emotional well-being. Early identification is essential to provide timely interventions that reduce long-term negative outcomes. Screening is the first step in identifying students who may need further assessment or targeted instruction.
Many current screening tools are designed to detect only one condition, require additional administration time beyond classroom activities, and can be costly to deploy. Most dyslexia screeners emphasize oral and speech-based measures and do not incorporate writing tasks. Analyzing children’s writing can yield complementary behavioral indicators that improve early detection.
This paper proposes a co-designed framework for building AI tools that augment existing screening practices and help practitioners—SLPs, OTs, general educators, and special educators—by making screening more efficient and actionable. The authors review current screening methods, examine prior AI-based approaches to detect dyslexia and dysgraphia, and highlight available handwriting datasets.
The framework supports integration with existing instruments like the Dysgraphia and Dyslexia Behavioral Indicator Checklist (DDBIC) and outlines a methodology for collecting offline and online handwriting samples to create a robust dataset for AI development. Co-design with end users ensures the resulting tools provide explainable, practical information that educators and clinicians can use to inform decisions and prioritize further evaluation or intervention.