A new machine learning model, named AutMedAI, has been developed to predict autism in young children with an accuracy of almost 80%. The model uses a combination of 28 different parameters, including age of first smile and presence of eating difficulties, to identify patterns in data and facilitate early detection of autism.