EUVIMEDEUVIMED entdecken
Crossref

Machine learning insights for cardiovascular risk prediction in diabetic patients: Emphasis on renal and cardiac markers using random forests

Julian Borges

Artificial Intelligence in Health · 19.05.2026 · 3 (3) · 025490111

DOI: 10.36922/aih025490111 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Artificial Intelligence in Health, online erschienen am 19.05.2026. Verfasst von Julian Borges.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Cardiovascular disease remains a leading cause of mortality among individuals with diabetes, yet many machine learning studies report inflated performance due to inadequate validation. This study evaluates whether standard, interpretable models can predict heart failure mortality using a rigorously validated analytic pipeline. Two publicly available datasets from the University of California, Irvine, Machine Learning Repository were analyzed independently: the Early Stage Diabetes Risk Prediction Dataset (n = 520) and the Heart Failure Clinical Records Dataset (n = 299). These datasets are not patient-linked; findings are framed as methodological feasibility rather than direct clinical prediction. The heart failure dataset was used for model development. Logistic regression and random forest classifiers were trained and evaluated using stratified five-fold cross-validation, with all metrics computed from pooled out-of-fold predictions. Preprocessing and class imbalance handling were confined to training folds to prevent information leakage. All analyses were conducted using R with the caret, randomForest, pROC, and ggplot2 packages. In a pooled out-of-fold evaluation, random forest achieved higher discriminative performance than logistic regression (area under the curve 0.91 versus 0.86). Random forest exhibited higher specificity, whereas logistic regression showed higher sensitivity, reflecting distinct error profiles. Feature importance analyses and Shapley additive explanations consistently identified serum creatinine, ejection fraction, age, and follow-up time as dominant predictors. Limitations include modest sample size, reliance on a single public dataset, and absence of external validation. These findings underscore the importance of conservative validation strategies and transparent baseline modeling for clinically responsible artificial intelligence research.

Zitieren

APA

Borges, J. (2026). Machine learning insights for cardiovascular risk prediction in diabetic patients: Emphasis on renal and cardiac markers using random forests. Artificial Intelligence in Health, 3(3), 025490111. https://doi.org/10.36922/aih025490111

Vancouver

Borges J. Machine learning insights for cardiovascular risk prediction in diabetic patients: Emphasis on renal and cardiac markers using random forests. Artificial Intelligence in Health. 2026;3(3):025490111. doi:10.36922/aih025490111

BibTeX
@article{Borges2026,
  author = {Borges, Julian},
  title = {Machine learning insights for cardiovascular risk prediction in diabetic patients: Emphasis on renal and cardiac markers using random forests},
  journal = {Artificial Intelligence in Health},
  year = {2026},
  volume = {3},
  number = {3},
  pages = {025490111},
  doi = {10.36922/aih025490111}
}
Kostenlos starten →