From Cohorts to Clinical Insights: Multimodal AI research at Korea’s National Institute of Health

All are welcome to attend -  no need to register, just turn up!

Speaker: Dr. Hye-Yeong Jo, Korea National Institute of Health (KNIH)

Talk Overview

• Current status of the national bio-big-data infrastructure established by KNIH and key findings from the COVID-19 multi-omics project
• Major research outcomes generated from the program
• Plans to develop a COVID-19–focused foundation model using 900 single-cell datasets from the multi-omics project

Abstract

The Korea National Institute of Health (KNIH), under the Korea Disease Control and Prevention Agency (KDCA), conducts national-level biomedical research with a focus on disease prevention and population health. This seminar introduces KNIH’s recent AI research initiatives and highlights key outcomes across multiple domains.

KNIH has applied AI and machine learning to a range of healthcare data types, including neuroimaging (PET and MRI), wearable-derived lifelog signals, and unstructured clinical text. These efforts span early diagnosis of neurodegenerative diseases, digital phenotyping for chronic condition monitoring, and the development of LLM-based tools to support genomic interpretation and clinical data structuring.

The seminar concludes with a discussion of KNIH’s COVID-19 multi-omics program. Since 2020, KNIH has built a longitudinal multi-omics dataset—including WGS, scRNA-seq, TCR/BCR-seq, and cytokine profiling—from 720 COVID-19 patients, the vaccinated and healthy controls. These data have been used to develop AI-based severity prediction models and are publicly available through the National Biobank of Korea. Building on this infrastructure, KNIH is currently developing AI-based foundation models to support preparedness and rapid response to future emerging infectious diseases.

These initiatives reflect KNIH’s broader efforts to integrate AI into public health research and offer perspectives relevant to national-scale health data science.