A MACHINE LEARNING FRAMEWORK FOR EARLY-STAGE DETECTION OF AUTISM SPECTRUM DISORDERS

Abstract

With a look at reducing effect of suffering through early treatments instead of eradicating completely, this research proposes a framework considering first detection of “Autism Spectrum Disorder (ASD) using Machine Learning (ML)” approaches. “Four Feature Scaling (FS)” algorithms—“Quantile Transformer, Power Transformer, Normalizer, & Max Abs” Scaler-the proposed structure is used on four specific ASD data sets covering age from toddlers towards adults. “ADABoost, Decision Tree, Support Vector Machine, Linear Discriminant Analysis, Logistic Regression, Gaussian Naïve Bayes, Decision Tree, & Random Forest” abide some machine learning algorithms, which abide used on functions-scalled dataset. Top classifies & FS approaches considering each age group abide displayed by comparing classification output using different statistical criteria. Experimental results suggest certain voting classifies most accuracy of predicting ASD considering toddlers & children, & considering adults & adolescents it gets maximum accuracy. By using four functional choice techniques, project involves an analysis of a broad functional view certain highlights role of fine-tuning ML methods towards predict ASD in different age groups. This suggests certain health professionals can use functional analysis towards guide their decisions during ASD screening. Compared towards other methods considering detecting Initial ASD, promise from proposed structure is displayed. In order towards improve accuracy & reliability of ASD detection, algorithms certain were suggested use a cloth method connecting a voice classify among “Random Forest (RF) & Adaboost”. result was a fantastic 100% accuracy degree.