EKSAKTA: Berkala Ilmiah Bidang MIPA
Open Access

Heart Attack History Classification Using XGBoost-Based Feature Selection and ADASYN

1 Department of Mathematics, Faculty of Mathematics and Natural Sciences, Universitas Negeri Surabaya, Surabaya, Indonesia
2 Department of Actuarial Science, Faculty of Mathematics and Natural Sciences, Universitas Negeri Surabaya, Surabaya, Indonesia
Download PDF

Abstract

Cardiovascular disease remains a major public health concern, highlighting the importance of data-driven approaches to understanding heart attack history. In the context of Sustainable Development Goal 3 (SDG 3), this study aims to classify individuals based on their reported history of physician-diagnosed heart attack using data from the fifth wave of the Indonesia Family Life Survey (IFLS5). The dataset comprises 32,173 individuals, including 315 positive cases (0.98%). The proposed approach combines XGBoost-based feature selection, Adaptive Synthetic Sampling (ADASYN), and seven classification algorithms. From ten initial predictors, four features were selected: Diabetes, HighCholesterol, BMI, and Weight_kg. Model evaluation used leakage-safe stratified 10-fold cross-validation and an independent holdout set, with a probability threshold of 0.50. Logistic Regression achieved the highest cross-validation ROC-AUC (0.603 ± 0.045), while SVM obtained the highest holdout ROC-AUC (0.6257). FDA approximation achieved the highest holdout recall (0.6190), but its precision was only 0.0161, resulting in 2,378 false-positive predictions. These findings indicate limited classification performance and a substantial false-positive burden. The study provides an exploratory analysis of reported heart attack history and should not be interpreted as a clinical diagnostic or future-risk prediction tool.

References

1
[1] Muharram, F. R., Multazam, C. E. C. Z., Mustofa, A., Socha, W., Andrianto, Martini, S., Aminde, L., and Yi-Li, C. (2024). The 30 Years of Shifting in The Indonesian Cardiovascular Burden: Analysis of The Global Burden of Disease Study. Journal of Epidemiology and Global Health, 14, 193–212.
2
[2] Chong, B., et al. (2025). Global burden of cardiovascular diseases: projections from 2025 to 2050. European Journal of Preventive Cardiology, 32(11), 1001–1015.
3
[3] Katsimardou, A., Imprialos, K., Stavropoulos, K., Sachinidis, A., Doumas, M., and Athyros, V. (2020). Hypertension in Metabolic Syndrome: Novel Insights. Current Hypertension Reviews, 16(1), 12–18.
4
[4] Lecube, A. (2024). Impact of obesity and diabetes on health and cardiovascular disease. Atencion Primaria, 56(12), 103045.
5
[5] Afifah, F., and Dewi Irawati Soeria Santoso (2024) Preventive Effect of Exercise on Kidney Fibrotic Pathway in Hypertension, EKSAKTA: Berkala Ilmiah Bidang MIPA, 25(3), 323–333.
Downloads
Download data is not yet available.

How to Cite

1.
Heart Attack History Classification Using XGBoost-Based Feature Selection and ADASYN. EKSAKTA [Internet]. 2026 Nov. 21 [cited 2026 Nov. 22];27(06):1130-45. Available from: https://eksakta.ppj.unp.ac.id/index.php/eksakta/article/view/775

Most read articles by the same author(s)