A COMPARATIVE STUDY ON INTERPRETABLE MACHINE LEARNING BASED SURVIVAL MODELS FOR CVD WITH CLINICAL DATA
DOI:
https://doi.org/10.69980/aa2j8s13Keywords:
Cardiovascular disease, survival analysis, interpretable machine learning, Cox proportional hazards, ensemble survival models, clinical dataAbstract
Cardiovascular disease stands as a major cause of death around the globe, thus stressing the need for proper prediction of prognosis and survival in the clinical setting. In classic diagnostic methods the disease presence is treated as a main factor but survival analysis provides the possibility of predicting the time-to-event by including the model information about the censored observation and disease progression. This study presents an interpretable survival modeling framework to understand cardiovascular disease outcomes using clinical data. Classical survival models, ensemble machine learning–based survival models, and deep learning–based survival methods are developed and tested systematically. Performance on the part of the model is examined using concordance index (C-index); interpretability through hazard ratios, coefficient analysis, and feature relevance is stressed. The experimental results show better predictive performance in ensemble survival models, and the requirement for clinical interpretability in classical survival models. These findings provide support for incorporating transparent survival models into clinical decision-support systems for cardiovascular disease prognosis.
References
1.Alizadehsani, R., et al. heart disease diagnosis by machine learning algorithms: A comprehensive survey. Expert Systems with Applications, 137, 1–16.
2.Breiman, L. (2001). Random forests. Machine Learning, 45(1), 5–32.
3.Chen, T., & Guestrin, C. (2016). XGBoost: A scalable tree boosting system. Proceedings of the 22nd ACM SIGKDD, 785–794.
4.Cox, D. R. (1972). Regression models and life-tables. Journal of the Royal Statistical Society: Series B, 34(2), 187–220.
5.Deo, R. C. (2015). Machine learning in medicine. Journal of the American College of Cardiology, 66(21), 2668–2679.
6.Freund, Y., & Schapire, R. (1997). A decision-theoretic generalization of on-line learning. Journal of Computer and System Sciences, 55(1), 119–139.
7.Friedman, J. H. (2001). Greedy function approximation: A gradient boosting machine. Annals of Statistics, 29(5), 1189–1232.
8.Harrell, F. E. (2015). Regression Modeling Strategies. Springer.
9.Ishwaran, H., et al. (2008). Random survival forests. Annals of Applied Statistics, 2(3), 841–860.
10.Katzman, J. L., et al. (2018). DeepSurv: Personalized treatment recommender system using a Cox model. Journal of Machine Learning Research, 19, 1–28.
11.Khera, R., et al. (2018). Machine learning methods for cardiovascular risk prediction. PLoS ONE, 13(10), e0204869.
12.Kleinbaum, D. G., & Klein, M. (2012). Survival Analysis: A Self-Learning Text. Springer.
13.Lee, C., et al. (2018). DeepHit: A deep learning approach to survival analysis. AAAI Conference on Artificial Intelligence.
14.Liu, Y., et al. (2018). Ensemble learning for cardiovascular risk prediction. BMC Medical Informatics and Decision Making, 18(1), 1–12.
15.Obermeyer, Z., et al. (2016). Predicting the future—big data, machine learning, and clinical medicine. Science, 352(6293), 1520–1522.
16.Polat, K., & Güneş, S. (2007). Heart disease diagnosis using SVM. Expert Systems with Applications, 32(2), 244–252.
17.Rahman, M., et al. (2019). Clinical risk prediction using machine learning. BMC Medical Informatics and Decision Making, 19(1), 1–13.
18.Shouman, M., et al. (2012). Applying ensemble classifiers for heart disease diagnosis. International Journal of Computer Applications, 38(4), 1–6. Sun, J., et al. (2020).
19.Survival analysis with machine learning methods. IEEE Access, 8, 125007–125017.
20.Therneau, T. M. (2020). A Package for Survival Analysis in R.
21.Uno, H., et al. (2011). Evaluating prediction rules for survival data. Statistics in Medicine, 30(10), 1105–1117.
22.Zeng, X., et al. (2020). Interpretability of machine learning models in healthcare. Scientific Reports, 10, 1–12.
