DOI: https://doi.org/10.59321/BAUETJ.V4I2.12
AUTHOR(S)
H M Mostafizur Rahman1, Masum Bakaul2*, Ahnaf Saif Choudhury3, Niaz Makhdum1, Md. Ejharul Haque4
ABSTRACT
Autism Spectrum Disorder (ASD) is a neurological condition that affects a person’s behavior and presents challenges in communication, cognition, and social skills. It can be particularly challenging in areas with inadequate education and limited diagnostic facilities. Despite years of research, scientists still face difficulties answering some questions about autism. However, certain common symptoms can help identify the disorder in children between 18 to 24 months old. This research aims to assist people in rural areas of Bangladesh by predicting ASD in children aged 1.50 to 2.00 years based on these symptoms. Following Cohen-Barren’s ASD detection criteria, a dataset was collected from three clinics in Madaripur, a district in the western part of Bangladesh. The dataset was cleaned, prepared for machine learning models, and split into 80% for training and 20% for testing. The Logistic Regression achieved the highest accuracy of the eight machine learning algorithms tested and achieved 99.30% accuracy. Although the dataset is small compared to the area’s population, future work will include testing more data and developing a web application for broader use.