Researchers Develop AI App to Improve Autism Screening
-
Researchers at Duke University are exploring using AI for autism screening in real-world settings like pediatric offices. The goal is to develop an objective, scalable early autism detection tool.
-
The current standard autism screening method is a 20-question parent survey, which has limitations like language barriers and inconsistent doctor follow-up.
-
The Duke researchers developed a digital app that displays movies to elicit behaviors that AI techniques can precisely measure to determine autism likelihood.
-
In a study, the app showed high sensitivity and specificity for autism detection, including in girls and children of color.
-
AI has potential benefits for healthcare like increasing access, efficiency, and equity - though it won't fully replace human providers.