Combination of Machine Learning and Functional Networks Concept for Diagnosis of Autism Spectrum Disorder

The subjective assessment of experts and cost of evaluation make it much harder to diagnose an early stage autism. This could lead to other complications in later life. In this study, we approach the problem of autism identification based on functional connectivity via selecting and interpreting input features. We access the difference in alpha band connectivity in autistic and neuro-typical individuals and demonstrate how this difference can be used to identify individuals with autism. Our study highlights the importance of alpha band connectivity coupled with supervised machine learning in the diagnosis of autism spectrum disorder.