Detecting Dysgraphia in Children Through Handwriting Image Analysis Using Hybrid Machine Learning Methodology
M. Vishnu Vardhana Rao, N. Sreeja, B. Sangeetha, T. Ramya Sri, I. Suneetha Rani, Kiran Kumari, Jayaditya Peddisetti · 2024
Nowadays writing disability is one of the major disorders in children, even parents also not identified early. Dysgraphia, a neurological disorder impacting writing abilities, poses significant challenges for children, especially as they begin to write. This disorder affects spelling, word spacing, and legibility, leading to slow and often unreadable writing. Early diagnosis is critical for effective intervention, yet traditional identification methods typically involve educational psychologists. Recent advances in machine learning offer promising automated solutions. This paper presents a Hybrid Machine Learning Methodology (HMLM) designed to detect dysgraphia in children aged 7-12 years through handwriting image analysis. The ability of this model to detect dysgraphia early supports timely interventions, potentially improving educational outcomes for affected children. HMLM architecture involves several stages namely Data Pre-Processing, Feature Extraction, a Predicting Model, and the final classification of dysgraphia or not based on the analysis of handwriting image dataset. The methodology combines Convolutional Neural Networks (CNN), Support Vector Machines (SVM), and Random Forest algorithms to analyze handwritten data collected from school children. %). The performance metrics such as accuracy, precision, recall, and F1 score are calculated to measure how well the model can distinguish between dysgraphia and non-dysgraphia samples. The proposed model shows notable effeteness, with CNN achieving an accuracy of 90.83%, surpassing SVM (81.88%) and Random Forest (81.25). This paper discusses the development, implementation, and comparative performance of the proposed machine learning models in the context of dysgraphia detection.