Evidence brief: Interpretable ensemble machine learning framework for cardiovascular disease prediction using EMR data and large language models in Ethiopia

Abstract

Cardiovascular diseases (CVDs) are leading causes of morbidity and mortality globally, with a growing burden in low- and middle-income countries such as Ethiopia. Early detection is limited by resource constraints, low screening uptake, and a lack of predictive tools tailored to local healthcare systems. This study presents an interpretable ensemble machine learning framework for predicting CVD risk via structured electronic medical record (EMR) data from public hospitals in Addis Ababa. We trained an XGBoost classifier on 20,960 anonymized records containing demographic, clinical, and physiological attributes. Preprocessing involves handling missing values, outlier capping, one-hot encoding, rare-category grouping, and dimensionality reduction. SHapley additive explanations (SHAPs) were used for feature attribution, and a large language model (Gemini) was used to translate SHAP outputs into plain-language narratives to enhance interpretability. The model achieved an accuracy of 0.99, with strong precision (0.99), recall (0.98), and F1-scores across both classes. SHAP analysis identified general_plan, history of present illness (HPI), musculoskeletal system (MSS) and diagnosis as key predictors. The integration of SHAP and LLMs provided transparent, clinician-friendly insights into model outputs, supporting adoption in resource-limited settings. This study demonstrates that combining ensemble learning with explainability techniques can yield highly accurate and interpretable CVD prediction models, offering potential for integration into clinical decision-support systems in Ethiopia.

1. Introduction

Cardiovascular diseases (CVDs) account for approximately 17.9 million deaths annually, representing 32% of all deaths worldwide [1]. In Ethiopia, urbanization, demographic shifts, and lifestyle transitions have contributed to a growing CVD burden, particularly in cities such as Addis Ababa [2]. However, limited diagnostic infrastructure, delayed care-seeking behavior, and insufficient screening programs hinder early detection and timely intervention [3,4]. These challenges are further compounded by resource constraints and the lack of localized clinical decision-support tools, leaving clinicians without effective means to identify at-risk individuals early in the disease trajectory.

Machine learning (ML) offers significant promise in improving early diagnosis and risk prediction of CVDs by leveraging complex, high-dimensional data patterns [5]. Nevertheless, the adoption of ML models in Ethiopian clinical practice has been limited, largely due to a lack of interpretability—most models function as “black boxes,” making it difficult for healthcare professionals to understand and trust their outputs [6]. This barrier is particularly critical in resource-limited settings, where transparency and simplicity are essential for frontline adoption.

This study addresses these gaps by developing a highly interpretable ensemble machine learning model for CVD prediction using structured electronic medical record (EMR) data from public hospitals in Addis Ababa. A unique aspect of this work is the use of SHapley Additive exPlanations (SHAPs) to identify and quantify the contributions of clinical, demographic, and physiological features such as diagnosis, general care plans, and musculoskeletal system (MSS) indicators. To further enhance transparency, a large language model (LLM) was employed to translate SHAP values into natural-language narratives, making the model’s outputs more accessible and actionable for clinicians without technical expertise.