Superior ML Model for Autism Spectrum Disease Prediction to Enhance the Scientific Trails

Autism Spectrum Disorder (ASD) detection is difficult due to its complex nature and heterogeneous presentation. Machine learning (ML) techniques detect autism spectrum disorder from behavioral and clinical data. The K-Nearest Neighbor (KNN) algorithm with Chi-Square feature selection is proposed in this study for ASD detection. The KNN algorithm is a famous classification and regression machine learning algorithm. In autism detection, KNN can be used as a classification algorithm to determine whether or not a person has autism. The first step in implementing KNN for autism detection is to collect data on individuals with and without autism. This study's dataset includes 704 children aged 2 to 17, including 352 typically developing children and 352 children with ASD. Twenty features in the dataset are related to social interaction, communication, and behavior. To select the most relevant features for ASD detection, Chi-Square feature selection is used. The KNN algorithm is trained on the features chosen and then tested using 10-fold cross-validation. The results show that the KNN algorithm with Chi-Square feature selection detects ASD with an accuracy of 91.47%. Eye contact, gesturing, pointing, and social smiling are chosen features. Using ML techniques, the proposed method provides a promising approach for ASD detection. Using Chi-Square feature selection improves the KNN algorithm's performance by selecting the most relevant features for ASD detection. Future research can improve the performance of the proposed approach by incorporating new features and experimenting with different ML algorithms.