Autism Spectrum Disorder (ASD) and Attention Deficit /and Hyperactivity Disorder (ADHD) are two common type of Neurodevelopmental Disorder (NDD) with increasing prevalence rate. Although children with NDD can be benefited by early intervention, most children are engaged in service only after the age of the golden treatment period. This is usually caused by the delay in diagnosis and the long waiting time for treatment. In this paper, our aim is to find out a solution to enable early screening of children with NDD and better engage them into the early training and education service. We applied machine learning approach to test the effectiveness of predicting ASD and ADHD risk by behaviors during infancy. We trained and tested four machine learning models on the Revised Infant Temperament...[ Read more ]

Autism Spectrum Disorder (ASD) and Attention Deficit /and Hyperactivity Disorder (ADHD) are two common type of Neurodevelopmental Disorder (NDD) with increasing prevalence rate. Although children with NDD can be benefited by early intervention, most children are engaged in service only after the age of the golden treatment period. This is usually caused by the delay in diagnosis and the long waiting time for treatment. In this paper, our aim is to find out a solution to enable early screening of children with NDD and better engage them into the early training and education service. We applied machine learning approach to test the effectiveness of predicting ASD and ADHD risk by behaviors during infancy. We trained and tested four machine learning models on the Revised Infant Temperament Questionnaires (RITQ) data from the ASD group, ADHD group and Typically Developing (TD) group. The high accuracy, sensitivity and specificity achieved by the machine learning models in classification of ASD support the use the temperament items to predict ASD risk in infancy while they are found to be not good predictors for ADHD. Our research work also provide evidence for the implementation of machine learning algorithm to the parent-directed NDD screening tool which provide parents a better way to screen their children for NDD risk accurately and effectively. Supported by our research work and scientific review, we implemented machine learning algorithm in the early screening tools in out platform which also provides service matching and online appointment of therapeutic service providers for children in need to reach the suitable service promptly.