ETRI develops AI technology for early screening of autism spectrum disorder

ETRI researchers have developed an artificial intelligence (AI) technique for the early screening of autism spectrum disorder (ASD). The technology is designed to detect early signs of ASD in children more quickly, which could expand opportunities for early screening and intervention and greatly improve accessibility.

Electronics and Telecommunications Research Institute (ETRI) announced that it has developed a “social interaction recognition AI” technology that can screen for autism spectrum disorder by analyzing “social interaction-inducing content” for infants and toddlers and video footage collected within six minutes of viewing it.

Autism spectrum disorder can be detected by observing certain behaviors and development, such as lack of social communication skills and restricted, repetitive behaviors. Early screening and intervention is critical for ASD, as early detection and appropriate medical intervention can have a positive impact on developmental outcomes.

However, due to a lack of experts, limited social awareness, and constraints on time and resources, it is reported that it takes two to six years from symptom detection to actual diagnosis. The key symptoms of ASD can appear as early as 12 to 24 months of age, and in some cases even earlier, emphasizing the importance of early screening and intervention.

ETRI researchers collaborated with Prof. Yoo Hee-Jung’s team from the Department of Psychiatry at Seoul National University Bundang Hospital to analyze the sensitivity of the screening index for ASD, based on data from 3,531 cases of infants and toddlers aged 42 months or younger. Based on this analysis, the researchers developed a scenario for observing infants and toddlers in which AI technology can be applied.

Based on this scenario, they developed the world’s first “social interaction-inducing content” that can induce and observe various social responses such as ▲ showing an interesting object, ▲ responding to a name, ▲ imitating behavior, ▲ pointing gesture, ▲ eye contact, etc.

In addition, they developed a “Social Interaction Recognition AI” technology that captures the interaction processes of infants and toddlers watching the content with cameras and performs △analysis of personal attributes, △prediction and monitoring of emotion state, △gaze tracking and response to name detection, △detection of pointing gestures, △detection of imitative and stereotypical behaviors, and etc.

The researchers established a living lab for screening of autism spectrum disorder at the Korea Institute of Robotics and Technology Convergence (KIRO) Seoul Center in 2020 and have been conducting observational tests and data collection on infants and toddlers for the past five years, thus advancing the technology of the field.

This technology is the world’s first multidisciplinary, convergence-based artificial intelligence (AI) technology for the screening of autism spectrum disorder, overcoming the limitations of existing screening tools and presenting a new solution that enables more objective and quantitative evaluations.

It lowers the barrier to mental health services, making it easier for pre-schools, childcare centers, developmental centers, and even homes to screen children as they interact with content.

It is expected to be used as a practical way to address the issue of early screening of infants and children by improving social awareness of autism spectrum disorders and activating preventive testing and early intervention.

Dr. Yoo Jang-Hee, Principal Researcher of the Social Robotics Research Section, said “We hope that this will help shorten the time between symptom detection and diagnosis, along with changing societal perceptions of autism. In addition, it is important for our research to solve hard problems, but we also hope that it will also contribute more to solving important problems like autism.”

Meanwhile, this technology was selected as one of the “100 National R&D Excellence Achievements in 2024” and has been recognized for its excellence through more than 50 domestic and international patent applications and 18 publications in international journals (SCIE).

1) Autism Spectrum Disorder (ASD): A type of neurodevelopmental disorder characterized by difficulties with social communication and interaction, and restricted and repetitive behaviors and interests. The term “spectrum” refers to the fact that the severity and types of symptoms vary from person to person.

2) Early screening/early intervention: Detecting conditions such as developmental disabilities as early as possible and, if necessary, providing medical and psychological interventions to prevent worsening or improve symptoms

3) Gaze tracking and response to name: The position of the child’s gaze toward a specific object (gaze point) and the degree to which the child responds when its name is called (response to name)

4) Imitative/stereotyped behavior: Imitative behavior is copying the movements or actions of others, and stereotypical behavior is repeating certain actions or words

5) Living Lab: A research model that engages users to experiment and advance technology in spaces that resemble real-life environments. In this case, we collected data and validated the technology while infants and their caregivers actually interacted with the content.

###

This research was conducted as part of the ‘SW Computing Industry Core Technology Development Project’ of the Institute of Information and Communications Technology Planning and Evaluation (IITP) with the support of the Ministry of Science and ICT, and Prof. Yoo Hee-Jung’s team of Seoul National University Bundang Hospital, Prof. Kim Hong-Guk’s team of Gwangju Institute of Science and Technology (GIST), and Dr. Kim Min-Kyu’s team of the Korea Institute of Robotics and Technology Convergence (KIRO) participated together.