EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Beyond Black-Box: SHAP-Driven System for Personalized Cardiovascular Interventions

Kailla Salsabila, Zaki Muhamad

Journal of Technology & Applied Sciences · 30.09.2026 · 1 (3) · 119-127

DOI: 10.64878/jtas.v1i3.262 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Journal of Technology & Applied Sciences, online erschienen am 30.09.2026. Verfasst von Kailla Salsabila und 1 weiteren Autor:innen. Die Arbeit nennt 32 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Cardiovascular disease is a leading cause of mortality worldwide, and its prevention needs accurate and preventive measures. Machine learning (ML) has demonstrated strong predictive power in the medical domain, but its black-box nature hinders adoption in the clinic due to limited decision transparency. The main novelty and objective of this study is the development of a hybrid system that transforms a passive diagnostic algorithm into a proactive Clinical Decision Support System (CDSS) by integrating prescriptive modeling, Explainable AI (XAI), and a rule-based inference engine. The CRISP-DM methodology was applied to evaluate a dataset of 70,000 medical records. In the preprocessing phase, physiological irregularities were removed, and a body mass index (BMI) feature was created. We optimized the Random Forest ensemble, deployed it using a web interface, and interpreted it with the SHAP framework. The empirical results demonstrated that the model achieved an accuracy of 73.66% and an F1 Score of 71.80%. In practice, this level of performance provides a reliable baseline for large-scale early population screening, effectively reducing false alarms in healthy subjects. The SHAP analysis indicated systolic blood pressure and obesity as the major risk factors. The rule-based engine automatically translated the risk factors into tailored actions (e.g., specific caloric deficits). Finally, this architecture links computational prediction with clinical implementation. However, the limitation of self-reported observational data is the potential recall bias, which requires future prospective clinical validation.

Zitieren

APA

Salsabila, K., & Muhamad, Z. (2026). Beyond Black-Box: SHAP-Driven System for Personalized Cardiovascular Interventions. Journal of Technology & Applied Sciences, 1(3), 119-127. https://doi.org/10.64878/jtas.v1i3.262

Vancouver

Salsabila K, Muhamad Z. Beyond Black-Box: SHAP-Driven System for Personalized Cardiovascular Interventions. Journal of Technology & Applied Sciences. 2026;1(3):119-127. doi:10.64878/jtas.v1i3.262

BibTeX
@article{Salsabila2026,
  author = {Salsabila, Kailla and Muhamad, Zaki},
  title = {Beyond Black-Box: SHAP-Driven System for Personalized Cardiovascular Interventions},
  journal = {Journal of Technology & Applied Sciences},
  year = {2026},
  volume = {1},
  number = {3},
  pages = {119--127},
  doi = {10.64878/jtas.v1i3.262}
}
Kostenlos starten →