Elijah Hill plays with his letter while sitting in the floor of his room, Saturday, Dec. 11, 2023, in Westfield, Ind. Families reliant on Medicaid worry changes coming in January will limit access to the therapy as Indiana and other states cut the size of the low-income healthcare program.(AP Photo/Darron Cummings)
(NewsNation) — A groundbreaking study shows deep learning models were successful in diagnosing children with autism using retinal photographs.
The study involving 958 participants aimed to explore whether artificial intelligence could serve as an objective tool for autism screening, given the challenges posed by limited resources and the increasing prevalence of the disorder.
The findings published in the JAMA Network Open shows new possibilities for utilizing retinal photographs as a noninvasive and accessible method for autism screening.
The AI models demonstrated remarkable accuracy in distinguishing between individuals with autism and those with typical development, showcasing potential applications for early detection.
Autism spectrum disorder is characterized by social communication impairment and restricted/repetitive behaviors, affecting a growing number of individuals globally.
Further research and validation are needed to establish the generalizability of the AI models and their application in diverse populations, according to the researchers from Yonsei University College of Medicine in Seoul, Korea.
English (United States) ·