APAT-A Smart Tool for Screening of Academic Learning Difficulties in Children

Learning difficulties(LD), particularly in reading, mathematics, or language skills, can significantly hinder children’s academic progress. Identifying these difficulties early is essential for implementing effective interventions and preventing long-term educational setbacks. This study proposes an ML-based Academic Performance Assessment Tool (APAT) to enhance screening accuracy and efficiency. A review of existing screening methods was conducted, and a comprehensive dataset approved by psychologists was developed. This involved collaboration with schools, where teachers with at least 8 months of classroom experience provided valuable inputs for children aged 6 to 11 years (Grades 1-5). Questionnaires were used to assess potential LD in this age group. Additionally, academic performance data from students with and without LD was collected. Different Machine Learning(ML) models, including Support Vector Machines (SVMs), 1D Convolutional Neural Networks (1D-CNNs), Random Forests (RFs), and Decision Trees (DTs), were evaluated for their effectiveness in predicting LD. Among these, the ID-CNN model achieved the highest accuracy, reaching 97%. Using the 1D-CNN model, APAT was developed. APAT allows teachers to input questionnaire data to identify learning difficulties in students, ensuring accurate assessments and timely interventions, providing a comprehensive and effective approach to supporting students with learning difficulties.