Autism being a spectrum or developmental disorder characterized by lack of social skills and repetitive behavior is being diagnosed using several methods in order to identify the onset of this disease. One such method is used by Autism Diagnostic Observation Schedule which inspects videotapes on the basis of an assessment between an examiner and a child for understanding the child’s behavior. According to a paper published in Science Translational Machine by researchers from the University of North Carolina at Chapel Hill and Washington University School of Medicine, doctors could precisely forecast which child might develop autism before hitting 24 months and with a 96 percent accuracy rate. A fully cross-validated machine learning was developed which used the scans of the 6-month-old infants. 59 high risk brain scans were taken over 230 regions and the whole brain was mapped creating matrices of functional connectivity from each child’s MRI data.  The algorithm further analyzed the brain scans of the 6-month-old infants and it properly predicted 9 out of 11 infants had the symptoms of autism at 24 months, with a sensitivity of 81.8%. Such data-driven approach is a good indicator of predictive measure that suggests that AI and machine learning could someday possibly recognize diseases with accurateness and extend treatments for the mass and maybe halt the headway of the disorders themselves.

Read more at: https://www.analyticsinsight.net/artificial-intelligence-machine-learning-can-be-used-to-predict-autism-in-children/