Artificial Intelligence Enabled Hybrid Machine Learning Application for Dyslexia Detection using Optimized Multiclass Support Vector Machine and Personalized Interactive and Assistive tools using Adaptive Reinforcement

Dyslexia, a neurological disorder affecting children's reading abilities, requires early diagnosis for improved outcomes. Traditional methods are time-consuming and costly. To automate diagnosis, a hybrid machine learning and AI architecture is proposed. Utilizing a Raspberry Pi setup, the system processes speech signals from a USB microphone. Preprocessing includes noise reduction and feature extraction using Fourier transform for Mel frequency cepstral coefficients. Support Vector Machine models classify dyslexic and non-dyslexic speech data from Kaggle. Personalized interactive support based on dyslexia probability is provided, with an LCD displaying results and a speaker aiding in recitation. Experimental analysis demonstrates superior accuracy and efficiency over conventional methods.