Interpretable AI Models for Accurate Malaria Detection in Healthcare Applications
Jasvant Mandloi, Harshita Mandloi, Rakesh Kumar Bhujade, Stuti Bhujade · International Journal of Emerging Technology and Advanced Engineering · 2025
Malaria continues to pose a significant global health challenge, especially in areas with limited resources, requiring diagnostic tools that are both precise and comprehensible for healthcare professionals. This study offers a comprehensive framework for malaria detection through Machine Learning (ML) models, enhanced by explainable artificial intelligence (XAI) tools, to examine both predictive efficacy and the interpretability of model predictions. The framework is utilized on a new dataset comprising P. vivaxinfected human blood smear images. To deal with class imbalance and feature variability, a lot of data preprocessing was done, such as synthetic minority oversampling and feature scaling. An ensemble model got a test accuracy of 96.8% and a ROC-AUC of 0.992. XAI techniques found that cell texture, color intensity, and parasite density were the most important morphological predictors. The findings suggest that combining machine learning with explainable AI offers a viable approach to developing precise, transparent, and clinically applicable diagnostic tools, aligning with the principles of sustainable engineering by facilitating early, accurate, and accessible healthcare interventions.