Scholarlink

Now a days the Identification of breast cancer remains a significant worldwide healthcare challenge, requiring the development of specific, transparent, and interpretable diagnostic tools. This work introduces an Explainable AI (XAI) approach like SHAP-enhanced Random Forest (RF) model which strategies to reduce the inherent transparency of traditional Machine Learning (ML) models. Utilizing SHAP (SHapley Additive exPlanations), the suggested model analyses prediction results by breaking them down into feature-level contributions, hence enhancing the comprehension of the diagnostic process. When applied to a clinical dataset, the model achieved a classification accuracy of 97.20%, surpassing conventional methods. Critical attributes, including tumor dimensions and texture, are recognized as substantial factors in diagnostic forecasting, equipping healthcare practitioners with comprehensible and practical insights. The proposed methodology improves the reliability of AI-based breast cancer diagnostics, facilitating informed clinical decision-making and providing the groundwork for future developments in XAI applications in healthcare.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Empowering Breast Cancer Diagnostics: SHAP-Enhanced Explainable AI

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Zulfikar Ali Ansari,
  • Rafeeq Ahmed,
  • Manish Madhava Tripathi,
  • Nafees Akhter Farooqui,
  • Md Shamsul Haque Ansari,
  • Shadab Siddiqui,
  • Shahin Fatima

摘要

Now a days the Identification of breast cancer remains a significant worldwide healthcare challenge, requiring the development of specific, transparent, and interpretable diagnostic tools. This work introduces an Explainable AI (XAI) approach like SHAP-enhanced Random Forest (RF) model which strategies to reduce the inherent transparency of traditional Machine Learning (ML) models. Utilizing SHAP (SHapley Additive exPlanations), the suggested model analyses prediction results by breaking them down into feature-level contributions, hence enhancing the comprehension of the diagnostic process. When applied to a clinical dataset, the model achieved a classification accuracy of 97.20%, surpassing conventional methods. Critical attributes, including tumor dimensions and texture, are recognized as substantial factors in diagnostic forecasting, equipping healthcare practitioners with comprehensible and practical insights. The proposed methodology improves the reliability of AI-based breast cancer diagnostics, facilitating informed clinical decision-making and providing the groundwork for future developments in XAI applications in healthcare.