National Center for Mental Health (NCMH). (Image courtesy of NCMH)
SEOUL, Sept. 11 (Korea Bizwire) — A team of South Korean researchers has developed an artificial intelligence model that allows parents to screen toddlers for autism spectrum disorder at home using nothing more than a smartphone.
The study, led by Professors Chun Geun-ah and Kim Hwi-young of Severance Hospital and Professor Kim Boong-nyun of Seoul National University Hospital, analyzed data from 1,242 children aged 18 to 48 months who visited nine hospitals across Korea. The findings were published this week in npj Digital Medicine, a journal partnered with Nature.
Autism spectrum disorder (ASD) is a developmental condition marked by difficulties in communication and social interaction, as well as restricted, repetitive behaviors. Early diagnosis and treatment can significantly improve outcomes, yet many cases are detected late. In the United States, the Centers for Disease Control and Prevention has reported that about one in three children with ASD is not diagnosed until after age eight.
Voice-Based AI Model
The newly developed AI analyzes voice recordings of children performing simple interaction tasks, such as responding to their name, mimicking parental gestures, engaging in pretend play with toys, or asking for help. Parents are instructed to record the tasks on a smartphone, after which the AI integrates the audio with results from a standard questionnaire-based screening test.
While traditional surveys reach about 70 percent accuracy, the AI’s ability to detect vocal tone, rhythm and interaction patterns sharply improved performance. Researchers said the model distinguished typically developing children from those at risk with more than 94 percent accuracy, and further separated high-risk groups from confirmed autism cases with 85 percent accuracy.
Toward Earlier Intervention
The team hopes the tool will allow parents to flag concerns at home and seek professional evaluation sooner, leading to earlier intervention. “Too often children come to us only after their symptoms have become severe,” Professor Chun said. “This AI can be used in the home to make earlier diagnosis possible, improving the chances for better treatment outcomes.”
The project was supported by Korea’s National Center for Mental Health as part of its initiative to develop digital therapeutics for developmental disorders.
Kevin Lee (kevinlee@koreabizwire.com)
