Work

A Fairness-Aware Explainable Artificial Intelligence Clinical Decision Support Framework Using Electronic Health Records

Year: 2026
Type: article
Abstract: Artificial Intelligence (AI) based Clinical Decision Support Systems (CDSS) are increasingly being adopted in Electronic Health Record (EHR) based healthcare settings to enhance disease prediction, clinical diagnosis and operational efficiency. Current healthcare AI systems, however, are facing serious challenges such as demographic bias, transparency, unreliable predictions, poor explainability and low clinician trust. We propose a fairness-aware explainable AI framework for trustworthy and transparent EHR-based CDSS. The proposed framework integrates machine learning prediction models with SHAP-based explainability, fairness evaluation metrics, subgroup bias attribution, reliability testing, and clinician-centered evaluation into one trustworthy healthcare AI architecture. Experiments are done using standard EHR datasets like UCI Heart Disease Dataset and MIMIC-IV. We compared different models like Logistic Regression, Random Forest, Gradient Boosting and XGBoost using metrics such as accuracy, ROC-AUC, Equalized Odds, Demographic Parity, robustness score and explainability consistency. Experimental results demonstrate that the proposed framework improves the reliability of clinical prediction, reduces subgroup bias, enhances interpretability, and increases clinician trust compared to conventional black-box healthcare AI systems. This work advances towards trustworthy, transparent, and fairness-aware healthcare AI systems fit for real-world clinical deployment.
Institution: Alliance University
Language: English
Open Access status: diamond
OpenAlex ID: W7236361161
FWCI:
Cites: 11
Cited by:
Related to: 10