Clustering of OECD countries by chronic disease burden: identifying associated health system and socioeconomic factors via machine learning
Journal of Medicine and Palliative Care · 25.10.2026 · 7 (5) · 1185-1195
DOI: 10.47582/jompac.1961328 ↗Kurz gefasst
Fachartikel in der Zeitschrift Journal of Medicine and Palliative Care, online erschienen am 25.10.2026. Verfasst von Seda Sönmez. Die Arbeit nennt 31 Literaturangaben.
Abstract (Originalsprache)
Aims
The aim of this study is to cluster OECD (Organisation for Economic Co-operation and Development) countries based on their chronic disease burden and to rank the health system and socioeconomic variables that discriminate these clusters using machine learning.
Methods
Secondary data obtained from international databases were analyzed in two stages for 14 variables across 38 OECD countries. First, a K-means clustering analysis was conducted based on age-standardized DALY (disability adjusted life years) rates for five major chronic disease groups and two associated risk factors. The statistical significance of cluster differences was tested using the Kruskal-Wallis test and the Bonferroni-corrected Dunn test. Second, health system and socioeconomic variables discriminating between clusters were ranked using the standard and balanced Random Forest models.
Results
OECD countries were grouped into three clusters: cluster 1 (n=4; Turkiye, United States, Mexico, Costa Rica), which exhibits a high burden of diabetes and kidney diseases and chronic respiratory diseases along with significant income inequality; cluster 2 (n=8; Central and Eastern Europe, Greece), which has the highest out-of-pocket payment rates along with a high burden of cardiovascular disease and tobacco-related risks; and cluster 3 (n=26; including Western European countries, Japan, Korea), which combines a relatively low burden of chronic diseases with high levels of health expenditure. Six out of seven health system and socioeconomic variables showed significant differences across the clusters. The standard Random Forest model identified current health expenditure and per capita income as the most discriminating variables, whereas in the balanced model, which accounted for class imbalance, income inequality ranked first.
Conclusion
The profiles of the burden of chronic diseases in OECD countries are associated with their socioeconomic and health system characteristics, particularly income inequality.
Zitieren
APA
Sönmez, S. (2026). Clustering of OECD countries by chronic disease burden: identifying associated health system and socioeconomic factors via machine learning. Journal of Medicine and Palliative Care, 7(5), 1185-1195. https://doi.org/10.47582/jompac.1961328
Vancouver
Sönmez S. Clustering of OECD countries by chronic disease burden: identifying associated health system and socioeconomic factors via machine learning. Journal of Medicine and Palliative Care. 2026;7(5):1185-1195. doi:10.47582/jompac.1961328
BibTeX
@article{Snmez2026,
author = {Sönmez, Seda},
title = {Clustering of OECD countries by chronic disease burden: identifying associated health system and socioeconomic factors via machine learning},
journal = {Journal of Medicine and Palliative Care},
year = {2026},
volume = {7},
number = {5},
pages = {1185--1195},
doi = {10.47582/jompac.1961328}
}
EUVIMED