Please view the main text area of the page by skipping the main menu.

Researchers develop AI-based technique for accurate autism spectrum diagnoses

A research team has developed an artificial intelligence (AI)-based system to more accurately identify people with autism spectrum disorder (ASD), a group of complex neurological disorders including Asperger's syndrome characterized by difficulties in communication and social interaction.

    The new technique -- which has prompted high hopes for swifter diagnosis and the development of new medications -- was announced in an April 14 article in the British journal Nature Communications.

    In Japan, around one person in 100 is said to have ASD -- but there had hitherto been no objective indexes to distinguish the range of conditions from other mental disorders.

    The research team -- which includes members from the Advanced Telecommunications Research Institute International (ATR) in Kyoto Prefecture as well as from the University of Tokyo -- used magnetic resonance imaging (MRI) scans to measure the brain activity of 181 volunteers in Japan and 88 in the United States, with and without ASD. An AI algorithm then analyzed and categorized how 140 specific sections of the brain worked together to carry out their respective functions.

    The AI found that strong activity in 16 specific connections between different regions of the brain predicted with 85 percent accuracy which people in the Japanese subject group had been previously diagnosed with ASD. The accuracy rate was 75 percent for the U.S. subject group.

    Also in The Mainichi

    The Mainichi on social media