• Mrinalinee Singh
• Dr. Rahul
• Dr.Upendra Kumar Srivastava

Abstract: Breast cancer remains one of the leading causes of cancer-related mortality among women worldwide, where early diagnosis plays a pivotal role in improving survival rates and treatment outcomes. Although machine learning (ML) techniques have demonstrated remarkable potential in automated breast cancer detection, the majority of existing studies prioritize predictive accuracy while overlooking model transparency, interpretability, and clinical trustworthiness. This limitation hinders the practical integration of artificial intelligence systems into real-world healthcare environments. To address this challenge, this research proposes a novel explainability-driven evaluation framework that systematically integrates predictive performance assessment with explainable artificial intelligence (XAI) analysis for early breast cancer diagnosis.The study conducts a comprehensive comparative investigation of state-of-the-art machine learning models, including Logistic Regression, Decision Tree, Random Forest, Support Vector Machine, K-Nearest Neighbors, Gradient Boosting, and Extreme Gradient Boosting (XGBoost). Unlike conventional comparative studies that rely solely on classification metrics, the proposed framework introduces a multi-dimensional evaluation strategy that simultaneously measures diagnostic accuracy, model robustness, feature stability, computational efficiency, and interpretability. Furthermore, SHAP (SHapley Additive Explanations) and LIME (Local Interpretable Model-Agnostic Explanations) are employed to provide both global and local explanations, enabling transparent identification of clinically significant features influencing diagnostic decisions. Experimental findings demonstrate that high-performing ensemble models achieve superior classification accuracy; however, their interpretability varies significantly compared with inherently transparent algorithms. The proposed explainability-oriented framework reveals critical trade-offs between predictive performance and clinical comprehensibility, offering a more realistic assessment of model suitability for healthcare deployment. Moreover, the study establishes an explainability-based ranking mechanism that assists clinicians and researchers in selecting trustworthy AI models beyond conventional accuracy-centric evaluation criteria. The novelty of this work lies in the integration of predictive analytics and explainability into a unified evaluation paradigm, providing actionable clinical insights while preserving diagnostic reliability. The proposed framework contributes toward the development of transparent, trustworthy, and clinically deployable AI systems for breast cancer screening and decision support, thereby advancing the broader adoption of explainable machine learning in precision healthcare.

Index Terms - Breast Cancer Diagnosis, Explainable Artificial Intelligence (XAI), SHAP, LIME, Machine Learning, XGBoost, Clinical Decision Support Systems, Interpretable AI, Early Cancer Detection, Precision Healthcare.

"Explainability - Driven Comparative Evaluation of Machine Learning Models for Early Breast Cancer Diagnosis", International Journal of Emerging Technologies and Innovative Research (www.jetir.org), ISSN:2349-5162, Vol.13, Issue 10, page no.c301-c317, October-2026, Available :http://www.jetir.org/papers/JETIR2610235.pdf

"Explainability - Driven Comparative Evaluation of Machine Learning Models for Early Breast Cancer Diagnosis", International Journal of Emerging Technologies and Innovative Research (www.jetir.org | UGC and issn Approved), ISSN:2349-5162, Vol.13, Issue 10, page no. ppc301-c317, October-2026, Available at : http://www.jetir.org/papers/JETIR2610235.pdf