Evaluating the Clinical Efficacy and Algorithmic Equity of Explainable Artificial Intelligence (XAI) Models in Predicting Cardiovascular Mortality Across Economically Disadvantaged and Demographically Diverse Populations

• Abilly Elly

  Texas University

  Author

Cardiovascular disease remains the leading cause of mortality globally, with disproportionately higher death rates observed among economically disadvantaged and racially diverse populations. Artificial intelligence (AI) models have demonstrated significant promise in predicting cardiovascular outcomes; however, concerns persist regarding algorithmic bias and inequitable performance across demographic subgroups. This study evaluates the clinical efficacy and algorithmic equity of explainable AI (XAI) models for predicting cardiovascular mortality, with particular focus on socioeconomic and demographic disparities. Using a retrospective cohort design analyzing 62,482 patient records across diverse U.S. populations, we developed and validated Random Survival Forest and DeepSurv models, incorporating comprehensive social determinants of health variables alongside traditional clinical risk factors. The DeepSurv model achieved superior predictive performance with an area under the curve of 0.89 (95% CI: 0.87-0.91), sensitivity of 83.5%, and specificity of 81.7%, consistent with recent meta-analytic findings . Importantly, SHAP-based explainability analysis revealed that socioeconomic factors—particularly median household income and educational attainment—ranked among the top three predictors of cardiovascular mortality, comparable to traditional risk factors such as age and smoking status. Model performance was assessed across demographic subgroups, revealing that XAI-enhanced models with equitable variable selection reduced algorithmic bias compared to standard approaches. These findings demonstrate that XAI models can achieve high predictive accuracy while promoting health equity when deliberately designed with representative training data and comprehensive social determinants of health.

• PDF

Copyright (c) 2026 Abilly Elly (Author)

This work is licensed under a Creative Commons Attribution 4.0 International License.