AI-Based Tool Achieves 98.5% Accuracy in Diagnosing Autism in Toddlers

AI-Based Diagnosis for Autism in Young Children

Researchers have developed a method using Artificial Intelligence (AI) technology that can diagnose Autism Spectrum Disorder (ASD) in children aged 2 to 4.

According to a report by UPI News on the 25th (local time), a research team led by Dr. Ayman El-Baz from the Department of Bioengineering at the University of Louisville in the United States announced the development of an AI system that can diagnose autism through the analysis of Diffusion Tensor Magnetic Resonance Imaging (DT-MRI).

DT-MRI is a special imaging technique that captures how water molecules move along the white matter pathways in the brain.

Autism Spectrum Disorder primarily manifests as a condition where the neural networks in the brain are not correctly connected.

According to the research team, DT-MRI captures abnormal neural connections that could lead to autism symptoms such as social communication impairments and repetitive behaviors.

This AI technology analyzes the distinctive patterns of an autistic brain compared to a normal brain through machine learning algorithms.

The research team analyzed the DT-MRI brain images of 226 children aged 24 to 48 months using this AI technology from the “Autism Brain Imaging Data Exchange” (ABID). Of these children, 126 were autistic, and 100 were normal.

High Success Rate in Autism Detection

This AI system demonstrated a success rate of 98.5% in distinguishing autism.

Currently, the tools used for diagnosing autism are subjective. In cases close to the boundary line between autism and expected growth, subjective analysis typically occurs.

Therefore, the research team emphasized the urgent need for the development of new and objective technology for early diagnosis of autism.

According to the 2023 Autism Community Report by the U.S. Centers for Disease Control and Prevention (CDC), less than half of children with autism receive developmental evaluations before the age of 3.

Even children who meet the criteria for Autism Spectrum Disorder, 30%, do not receive an official diagnosis until they are 8 years old.

Seeking FDA Approval for Commercialization

The research team stated that using the newly developed AI technology first to evaluate autism and then having psychologists confirm the results could reduce their workload by up to 30%.

They hope to get the Food and Drug Administration (FDA) approval to commercialize this AI technology.

However, Dr. Susan Hyman, a specialist in developmental and behavioral pediatrics at the University of Rochester Children’s Hospital, highlighted the challenges of taking a brain MRI of a child with autism, as they cannot stay still during the scan. She suggested it should be done while they are sleeping to avoid the need for a sedative, which could pose a medical risk and to which parents might not agree.

She presented the results of this research at the Radiological Society of North America (RSNA) held in Chicago.

By. Ahn Yoo Jin