Researchers at the Electronics and Telecommunications Research Institute (ETRI) have developed a groundbreaking artificial intelligence (AI) technology designed to detect early signs of autism spectrum disorder (ASD) in young children. This advancement could significantly increase access to early screening and intervention, improving long-term developmental outcomes.
Detecting Early Signs of Autism Through Social Interaction
The newly developed “Social Interaction Recognition AI” works by analyzing children’s reactions to specially designed “social interaction-inducing content.” The system captures video footage of infants and toddlers as they engage with the content and evaluates their responses within just six minutes.
This AI technology identifies key indicators of ASD, such as a lack of social communication, reduced eye contact, repetitive behaviors, and limited responsiveness. These behaviors typically appear between 12 and 24 months of age, sometimes even earlier, underscoring the need for timely detection and treatment.
Bridging the Diagnosis Gap
Currently, many children face delays of two to six years between the first signs of ASD and an official diagnosis, primarily due to a shortage of specialists, limited awareness, and resource constraints. ETRI’s innovation aims to close this gap by offering a faster and more accessible screening method.
As per the press release, ETRI collaborated with Professor Yoo Hee-Jung from the Department of Psychiatry at Seoul National University Bundang Hospital to evaluate the accuracy of the screening index. Together, they analyzedbehavioral data from 3,531 infants and toddlers under 42 months of age to develop a realistic observational scenario for the AI tool.
Creating a New Standard in ASD Detection
Building on this scenario, the team designed interactive content that prompts various social behaviors, including responding to names, making eye contact, imitating actions, and pointing. The AI system tracks these behaviors using advanced capabilities such as gaze tracking, emotional state prediction, and gesture recognition.
This makes the technology the first of its kind to use a multidisciplinary, convergence-based approach in the field of autism screening. It offers a more objective and quantitative alternative to traditional methods, which often rely heavily on subjective interpretation.
Expanding Accessibility Through Innovation
Since 2020, ETRI has operated a living lab at the Korea Institute of Robotics and Technology Convergence (KIRO) Seoul Center to test and refine the system. Over the past five years, researchers have gathered valuable observational data and continuously improved the technology.
The AI screening tool is designed for ease of use in various settings, including preschools, developmental centers, and even homes. By simplifying the screening process, the system lowers the barrier to accessing mental health services and encourages broader societal engagement with early autism detection.
Looking Toward a More Inclusive Future
Dr. Yoo Jang-Hee, Principal Researcher of the Social Robotics Research Section, emphasized the broader implications of the work. “We hope this will help shorten the time between symptom detection and diagnosis, while also shifting public perceptions of autism,” he said. “Our goal is not just to tackle technically complex problems, but also to contribute to solving deeply important issues like autism.”