Machine Learning-Based Autism Spectrum Disorder Prediction: A Comparative Approach

Autism Spectrum Disorder (ASD) presents a significant challenge for early identification and action, given its complex and varied nature. Recent developments in machine learning (ML) offer an exciting chance to enhance ASD prediction, transforming the diagnostic and management processes. This research paper implements an exhaustive examination of ML-based ASD prediction methods, emphasizing a comparative assessment of various ML algorithms and feature selection techniques. The primary objective is to determine the most effective models for early ASD diagnosis, acknowledging the vital role of early intervention in improving developmental outcomes for individuals with ASD. By employing a diverse dataset and a spectrum of machine learning (ML) algorithms, such as K-Nearest Neighbors (KNN), Decision Tree, Random Forest, Support Vector Machine (SVM), and others, this research study illustrates the efficacy of ML techniques in prediction of Autism Spectrum Disorder (ASD). Notably, the study finds that Random Forest and Extra Tree models exhibit high precision in test outcomes. This underscores the promising potential of ML methodologies in accurately predicting ASD. The research methodology includes dataset selection, data preprocessing, feature extraction by utilizing Principal Component Analysis (PCA), and finally model classification, followed by rigorous evaluation metrics like precision, recall, F1Score, and confusion matrices that measure model performance. The results show the strengths and limitations of various ML algorithms and suggest the significance of feature selection to improve predictive accuracy. This paper provides valuable insights into crafting effective models for early ASD diagnosis and intervention, promoting a path toward augmenting the lives of people on the autism spectrum and their families.