A Linear Regularized Normalized Model for Dyslexia and ADHD Prediction Using Learning Approaches

Abstract

A learning disability called dyslexia typically affects school-age kids. Children have trouble spelling, reading, and writing words. Children who experience this problem often struggle with negative emotions, rage, frustration, and low self-esteem. Consequently, a dyslexia predictor system is required to assist children in overcoming the risk. There are many current ways of predicting dyslexia. However, they need to provide higher prediction accuracy. Also, this work concentrates on another disorder known as Attention-Deficit Hyperactivity Disorder (ADHD). The prediction process is more challenging as the prediction process shows some negative consequences. The data is typically gathered from online resources for the prediction process. This study examines how the predictor model predicts dyslexia and ADHD using learning strategies. Here, the most important features for accurately classifying dyslexia, non-dyslexia and ADHD are extracted using a new Support Vector Machine (SVM) for feature selection based on the norm and norm. Based on the weighted values, the predicted model provides improved subset features from the internet-accessible dataset. The accuracy, precision, F1-score, specificity, sensitivity, and execution time are all examined here using 10-fold cross-validation. The maximum accuracy reached with this feature subset during the prediction process is carefully reviewed. The experiment results imply that the anticipated model is used to accurately predict the defect and as a tool for CDSS. Recently, dyslexia and ADHD prediction has been greatly aided by computer-based predictor systems. The expected model also effectively fits the experimental design, bridging the gap between feature selection and classification.