AI able to identify autism in children with 98.5% accuracy
A newly developed artificial intelligence (AI) system has demonstrated an accuracy rate of 98.5% at diagnosing autism in children age 24 to 48 months.
The three-stage AI system analyzes diffusion tensor MRI (DT-MRI) scans—an imaging technique that tracks water movement in the brain's white matter. By isolating brain tissue images and extracting markers indicating connectivity between brain regions, a machine learning algorithm compares patterns in children with autism to those with typical development.
“Autism is primarily a disease of improper connections within the brain," study co-author Gregory Barnes, MD, PhD, of the Norton Children's Autism Center said in a statement. "DT-MRI captures these abnormal connections that lead to the symptoms that children with autism often have, such as impaired social communication and repetitive behaviors."
The AI system, applied to a DT-MRI dataset from 226 children, exhibited 97% sensitivity, 98% specificity, and an overall accuracy of 98.5% in identifying autism. The full results are set to be presented at the Radiological Society of North America (RSNA) annual meeting in Chicago later this month.
Closing the gap
The CDC's 2023 Community Report on Autism highlights a concerning gap in early diagnoses for children with autism spectrum disorder. According to Barnes and the other researchers, the AI model used in their study has the potential to address these challenges by reducing assessment and treatment time and costs. They said the AI not only identifies affected neural pathways but also provides information on the expected impact on brain functionality and a severity grade, aiding in early therapeutic intervention.
But, Barnes added that AI can’t replace the need for a human to make an assessment.
"Imaging offers the promise of quickly detecting autism in an objective fashion," he said. "We envision an autism assessment that begins with DT-MRI followed by an abbreviated session with a psychologist to confirm the results and guide parents on next steps. This approach could reduce the psychologists' workload by up to 30%."
Looking ahead, the researchers said they are actively seeking clearance from the U.S Food and Drug Administration to use their AI in real-world clinical settings.
Chad is an award-winning writer and editor with over 15 years of experience working in media. He has a decade-long professional background in healthcare, working as a writer and in public relations.