Interpreting Cervical Cancer Risk Predictions Using Optimized Random Forest and Explainable AI Techniques

Cervical cancer remains a major public health concern, requiring accurate and interpretable risk prediction tools for early diagnosis. Machine learning models often lack transparency, limiting their adoption in clinical settings. This study proposes a robust framework using an optimized Random Forest classifier for cervical cancer risk prediction, combined with Explainable AI (XAI) techniques such as LIME. The Random Forest model is optimized using key hyperparameters to enhance predictive accuracy. LIME provides local explanations, visualizing feature contributions for individual predictions, while global feature importance identifies the most influential risk factors. To ensure clinical relevance, feature scaling reversal is applied, enabling interpretable outputs. The proposed framework achieves high accuracy, offers transparent decision-making, and enhances trust among medical professionals. This study demonstrates the potential of explainable machine learning for improving cervical cancer diagnosis.