Autism Spectrum Disorder Prediction Using Machine Learning and Design Science

Abstract

Machine learning, a subset of Artificial Intelligence, has gained much recognition in facilitating disease prediction and the decision-making process in healthcare. One of the most often diagnosed developmental disorders in the world is Autism Spectrum Disorder (ASD). Around the world, it is reported to afflict 75 million people and the number of cases has gradually increased since studies began in the 1960s. The symptoms generally include communication deficits, sensory processing differences, and repetitive actions or behaviors. This research develops a model to detect ASD using Principal Component Analysis and Machine Learning algorithms to classify and predict the risk of ASD among pregnant women. Data was collected from National Hospital in Abuja, Nigeria. According to the results, PCA and Random Forest produced the best accuracy of 98.7%. Bayesian probability was employed to evaluate and verify the reliability of the model. The created model can aid doctors in diagnosing ASD.