Multidimensional Trends and AI-Based Forecasts of HIV Incidence in Korea, 1985–2024
Journal of Epidemiology and Global Health · 2026
DOI: 10.1007/s44197-026-00549-4 ↗Kurz gefasst
Fachartikel in der Zeitschrift Journal of Epidemiology and Global Health, 2026. Verfasst von Seungho Baek und 3 weiteren Autor:innen. Beteiligte Einrichtung: AI·Information·Reasoning (AI/R) Laboratory, Computational Science Center, Korea Institute of Science and Technology, Seoul, 02792, Republic of Korea. Hauptthemen laut MEDLINE: HIV Infections. Die Arbeit wurde bisher 1-mal zitiert (Crossref) und nennt 36 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY-NC-ND 4.0).
Abstract (Originalsprache)
BACKGROUND: Since the first reported HIV case in Korea in 1985, the incidence of HIV/AIDS has continued to rise, emphasizing the need for effective prevention and management strategies. While global initiatives aim to end the epidemic by 2030, Korea has observed demographic shifts, particularly among young men and foreign nationals, that warrant further investigation. METHODS: We analyzed national HIV incidence data from 1985 to 2024, stratified by nationality, age, transmission route, and testing motivation. To forecast future trends, we employed a stacking ensemble model combining ARIMA, Random Forest, and LSTM, each trained to capture different aspects of temporal and demographic patterns. Model performance was evaluated using RMSE and MAE. RESULTS: Annual new HIV infections rose steadily until the mid-2010s, peaking at 1,223 cases in 2019, before stabilizing around 975–1,000 cases by 2024. Korean men showed a recent decline, while foreign nationals, especially foreign women, displayed continuous growth. Young men aged 25–34 emerged as the most affected group, comprising nearly half of male cases in 2023. The ensemble model outperformed individual approaches, projecting that new HIV cases will stabilize at approximately 1,024 annually through 2030. CONCLUSION: Analysis of HIV incidence in Korea using 40-year trend data reveals a historically rising trend that has recently begun to stabilize, with projections suggesting this plateau will continue through 2030. Ensemble-based forecasting models leveraging four decades of surveillance data may enhance predictive accuracy and serve as a framework for other settings facing shifts in HIV epidemiology.
Schlagwörter
MeSH-Schlagwörter (MEDLINE)
Stichwörter der Autor:innen
Autor:innen und Institutionen
- AI·Information·Reasoning (AI/R) Laboratory, Computational Science Center, Korea Institute of Science and Technology, Seoul, 02792, Republic of Korea.
- AI-Robot Department, University of Science and Technology, Daejeon, 34113, Republic of Korea.
- AI·Information·Reasoning (AI/R) Laboratory, Computational Science Center, Korea Institute of Science and Technology, Seoul, 02792, Republic of Korea. eau@ust.ac.kr.
- AI-Robot Department, University of Science and Technology, Daejeon, 34113, Republic of Korea. eau@ust.ac.kr.
- Department of Prevention and Management, Inha University Hospital, Incheon, Republic of Korea, 22332. yeonju.kim5@gmail.com.
- Department of Medicine, Inha University College of Medicine, Incheon, 05368, Republic of Korea. yeonju.kim5@gmail.com.
Zitieren
APA
Baek, S., Cho, H., Kim, C., & Kim, Y. (2026). Multidimensional Trends and AI-Based Forecasts of HIV Incidence in Korea, 1985–2024. Journal of Epidemiology and Global Health. https://doi.org/10.1007/s44197-026-00549-4
Vancouver
Baek S, Cho H, Kim C, Kim Y. Multidimensional Trends and AI-Based Forecasts of HIV Incidence in Korea, 1985–2024. Journal of Epidemiology and Global Health. 2026. doi:10.1007/s44197-026-00549-4
BibTeX
@article{Baek2026,
author = {Baek, Seungho and Cho, Haneol and Kim, Chansoo and Kim, Yeonju},
title = {Multidimensional Trends and AI-Based Forecasts of HIV Incidence in Korea, 1985–2024},
journal = {Journal of Epidemiology and Global Health},
year = {2026},
doi = {10.1007/s44197-026-00549-4}
}
EUVIMED