Detection of Autism Spectrum Disorder by A Case Study Model Using Machine Learning Techniques An Experimental Analysis on Child, Adolescent and Datasets

Abstract

Autism Spectrum Disorder (ASD) is a lifelong neurodevelopmental disorder that affects a person's social interaction and communication skills. It is typically diagnosed in childhood but can be identified at any age. Behavioral symptoms of autism usually appear in the first two years of a child's life and continue into adulthood. Recently, there has been increased interest in using machine learning algorithms for medical diagnosis, including the diagnosis of autism spectrum disorder.

This study aimed to investigate the feasibility of using various machine learning algorithms, such as Naïve Bayes, Support Vector Machine (SVM), Random Forest, Logistic Regression, K-Nearest Neighbors (KNN), Decision Tree, and Gradient Boosting Classifier, to predict and analyze autism in children. The researchers utilized publicly available non-clinical ASD datasets for evaluation.

Different evaluation metrics, including accuracy, specificity, sensitivity, macro-average, and weighted average, were used to assess the performance of the machine learning models. The KNN-based model achieved the highest accuracy of 87.14% and outperformed the other models in terms of specificity. The Naïve Bayes model achieved an accuracy of 70.48%, while the SVM model had the highest sensitivity of 98.2%. The Decision Tree and Random Forest models achieved perfect scores of 100% in terms of macro-average, weighted average, and Mean Accuracy for all models was 85.52%.

Based on these results, the researchers concluded that the KNN-based model is the most effective for predicting and analyzing autism in children, with an accuracy of 87.14%. However, it is important to note that these findings are specific to the dataset and evaluation metrics used in the study. Further research and validation using diverse datasets are necessary to confirm the generalizability of these findings.