Developing a Transparent Anemia Prediction Model Empowered with Explainable AI

Anaemia affects millions of people around the world and happens when the blood does not contain enough haemoglobin to carry oxygen properly. Caused by nutritional deficiencies, genetic conditions, or chronic diseases, making accurate prediction models essential for early diagnosis and effective treatment. However, traditional diagnostic approaches and statistical models often lack clarity and fail to provide transparent insights for clinical interpretation. Although AI-based prediction systems show improved accuracy, many operate as black- box models, making it difficult for doctors to trust and understand their predictions. To solve this issue, this research proposes a transparent and interpretable anaemia prediction model using machine learning algorithms such as SVM, Decision Trees, K- Nearest Neighbour’s, and Gradient Boosting, along with Explainable AI (XAI) techniques like SHAP and LIME. These tools clearly explain how specific features influence predictions, enhancing trust and usability in medical environments. The proposed model has an accuracy of 98.13% with a 1.87% miss-rate, outperforming previously published approaches.