AI, Digital Mental Health, and Computational Behavioral Science
Generative AI-based Detection of Challenging Behaviors in Children with Autism
Sep 2022 - Present
In collaboration with Prof. Junghyuk Park at George Washington University, this project develops AI algorithms for the automatic detection of challenging behaviors in children with autism spectrum disorder (ASD).
The first collaborative paper was published in 2025, and advanced follow-up studies are currently underway. A Korea–U.S. joint research grant proposal is also being prepared.
Development of Dynamic Persona-Based Conversational AI Agents
Nov 2025 - Present
This project develops a conversational AI agent designed to support university students experiencing academic burden, career anxiety, and interpersonal stress. Using focus group interviews and survey data, the project investigates user needs and AI preferences, while developing a prototype capable of dynamically adapting personas according to users’ emotional states and situational contexts.
Development and Validation of On-Device AI for Complex Emotion Recognition in Drivers
Oct 2025 - Present
Funded by the Ministry of Trade, Industry and Energy, this project develops AI technologies and dedicated SoC systems capable of real-time recognition of drivers’ complex emotions using multimodal data including facial expressions, physiological signals, gaze, and speech. The project integrates Korean emotion models to improve precision and builds a platform compatible with automotive electronic systems.
Youth Mental Health Care Platform Development
Sep 2023 - Apr 2024
Supported by the Ministry of Science and ICT and the National Information Society Agency, this collaborative project with healthcare company HUNO Inc. developed and operated a customized mental health care platform for young populations including adolescents, soldiers, university students, and employees.
University Student Mental Health Database and Predictive Modeling Project
Apr 2021 - Feb 2026
This project established a large-scale longitudinal mental health database integrating clinical, cognitive, social, and biological indicators to identify risk and recovery trajectories among university students. The project further aimed to build a campus-wide mental health prevention and intervention system.
Professor Kyungmi-Chung and collaborators develop AI-based behavioral assessment tool
AV-FOSReducing the burden of analyzing problem behaviors in children with Autism Spectrum Disorder (ASD) using AI
Professor Kyungmi Chung of the Department of Psychology, together with Professor Jung Hyuk Park from the Department of Biomedical Engineering at George Washington University, has introduced an AI-based video analysis model for assessing problem behaviors in children with Autism Spectrum Disorder (ASD). Children with ASD are at a higher risk of exhibiting severe problem behaviors such as self-injury, aggression, and destructive behaviors, which can limit family adaptation, educational opportunities at school, and participation in the community. Traditionally, assessment of these behaviors relies on lengthy direct observation by trained professionals, a process that is both time-consuming and costly, while also being limited in accessibility due to a shortage of specialists.
To address these limitations, the research team collected 216 home videos of 83 children with ASD displaying problem behaviors in naturalistic settings. Using the Revised Family Observation Schedule, Third Edition (FOS-R-III), a clinical observational measure designed to record parent–child interactions, the team coded problem behaviors observed in the videos at 10-second intervals. These coded data served as the ground truth for developing an AI algorithm and an automated behavioral analysis system called “AV-FOS.” With approximately 10 minutes of video input, AV-FOS can provide information on both the type and severity of problem behaviors. This study is meaningful in demonstrating the feasibility of automated assessment of problem behaviors in children with ASD within home environments. With further research to expand the range of measurable behaviors and improve accuracy, AV-FOS could become a practical tool for parents and teachers to easily monitor the severity of problem behaviors, while also serving as a valuable support tool for clinicians in making more accurate assessments and treatment decisions.
Article Title:AV-FOS: Transformer-Based Audio-Visual Multimodal Interaction Style Recognition for Children With Autism Using the Revised Family Observation Schedule 3rd Edition (FOS-R-III) [View Article]
Journal:IEEE Journal of Biomedical and Health Informatics