EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

From Clinical Profiles to Actionable Decisions: Explainable Artificial Intelligence for Early Heart Disease Detection

Dr. Uttara Patnaik, N Suresh Rao, Dr Ajab Singh Choudhary, DINGARI ACHARYA KIRAN, Dr Sheena P Kochumon, Yong Ching Yin

Computer and Decision Making: An International Journal · 05.11.2026 · 3 (6S) · 3067-3076

DOI: 10.1922/comdem.v3i6s.18893 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Computer and Decision Making: An International Journal, online erschienen am 05.11.2026. Verfasst von Dr. Uttara Patnaik und 5 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

The early identification of risks and reliable predictive methods is important, since the CHD is a major health problem worldwide. The purpose of this study was to assess explainable artificial intelligence to predict CHD risk over a 10-year period with demographic, behavioral, clinical, and biochemical features. This was a quantitative, observational and computational study, utilizing clinical records from the Framingham Heart Study, which included 4,240 people and 15 predictor variables. The four machine learning algorithms, namely Logistic Regression, Random Forest, Support Vector Machine, and XGBoost were tested using stratified training and testing procedures. Classification metrics and receiver operating characteristic analysis and probability calibration were used to assess predictive performance. SHapley Additive exPlanations (SHAP) were used to find important clinical predictors. Results showed that there were substantial differences among groups by CHD outcome in age, blood pressure, cholesterol, glucose and smoking intensity. Logistic Regression had the best discriminatory ability with an ROC-AUC of 0.700 and a sensitivity of 0.597. SHAP analysis revealed that age, cigarette use, systolic blood pressure, and male sex were significant factors in the prediction of cardiovascular risk. Although the method was developed with good interpretation in mind, the discrimination and calibration range were moderate which limited the clinical usefulness of the method. In summary, explainable machine learning holds promise for personalized cardiovascular risk prediction and prevention. Reliable and actionable cardiovascular prediction systems need to be validated and clinically evaluated in various health care environments.

Zitieren

APA

Patnaik, D. U., Rao, N. S., Choudhary, D. A. S., KIRAN, D. A., Kochumon, D. S. P., & Yin, Y. C. (2026). From Clinical Profiles to Actionable Decisions: Explainable Artificial Intelligence for Early Heart Disease Detection. Computer and Decision Making: An International Journal, 3(6S), 3067-3076. https://doi.org/10.1922/comdem.v3i6s.18893

Vancouver

Patnaik DU, Rao NS, Choudhary DAS, KIRAN DA, Kochumon DSP, Yin YC. From Clinical Profiles to Actionable Decisions: Explainable Artificial Intelligence for Early Heart Disease Detection. Computer and Decision Making: An International Journal. 2026;3(6S):3067-3076. doi:10.1922/comdem.v3i6s.18893

BibTeX
@article{Patnaik2026,
  author = {Patnaik, Dr. Uttara and Rao, N Suresh and Choudhary, Dr Ajab Singh and KIRAN, DINGARI ACHARYA and Kochumon, Dr Sheena P and Yin, Yong Ching},
  title = {From Clinical Profiles to Actionable Decisions: Explainable Artificial Intelligence for Early Heart Disease Detection},
  journal = {Computer and Decision Making: An International Journal},
  year = {2026},
  volume = {3},
  number = {6S},
  pages = {3067--3076},
  doi = {10.1922/comdem.v3i6s.18893}
}
Kostenlos starten →