High-Accuracy and Interpretable Anaemia Diagnosis Using Hybrid Ensemble Learning with Explainable AI

Abstract

Anaemia is a widespread public health issue in India, particularly among women and children, leading to serious health complications and increased healthcare costs. Despite various government initiatives to reduce its prevalence, challenges such as delayed diagnosis, limited access to screening facilities, and nutritional deficiencies continue to hinder effective management. Traditional diagnostic methods rely on laboratory testing and clinical evaluation, which can be time-consuming and difficult to implement for large-scale population screening. This study proposes an explainable ensemble learning framework for anaemia diagnosis and clinical decision support. The system utilizes multiple machine learning algorithms, including Support Vector Machine (SVM), Decision Tree, K-Nearest Neighbors (KNN), Gradient Boosting, Random Forest, and XGBoost, to predict anaemia using patient health data. Ensemble voting techniques are employed to combine the strengths of individual models and improve overall prediction accuracy and robustness. To enhance transparency and support clinical acceptance, Explainable Artificial Intelligence (XAI) methods such as SHAP and LIME are integrated into the framework. These techniques provide interpretable explanations by identifying the contribution of important features, including haemoglobin level, age, and gender, to each prediction. The proposed approach enables accurate, transparent, and scalable anaemia detection, supporting early intervention and assisting healthcare professionals in making informed and reliable clinical decisions.