Scholarlink

The promise of integrating Artificial Intelligence (AI) into the medical sphere has been outstanding with the models already scoring over 90% accuracy in various specialities such as radiology, oncology, and cardiology. An unexplainability of the model is one of the largest threats to clinical adoption. This proposes a platform of explainable artificial intelligence (XAI) which incorporates the usage of SHAP, LIME, and Grad-Cam as methods in explainable model application in healthcare predictive models. Our XAI-augmented models were highly generalized (AUC > 0.92) and much more explainable (ranked on a 50,000-patient record data set) in comparison to three clinical tasks. More than 80% of the clinician respondents gave greater trust and comprehension on the basis of visual and textual clarification. Our paradigm fills the black-box AI model-explainability gap in actual clinical decision-making and sparks trust, accountability, and patient-informed care.

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

Enhancing Clinical Decision-Making through Explainable AI: Integrating Interpretability Techniques for Transparent Healthcare Models

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • M. Chandra Sekhar,
  • Balajee Maram

摘要

The promise of integrating Artificial Intelligence (AI) into the medical sphere has been outstanding with the models already scoring over 90% accuracy in various specialities such as radiology, oncology, and cardiology. An unexplainability of the model is one of the largest threats to clinical adoption. This proposes a platform of explainable artificial intelligence (XAI) which incorporates the usage of SHAP, LIME, and Grad-Cam as methods in explainable model application in healthcare predictive models. Our XAI-augmented models were highly generalized (AUC > 0.92) and much more explainable (ranked on a 50,000-patient record data set) in comparison to three clinical tasks. More than 80% of the clinician respondents gave greater trust and comprehension on the basis of visual and textual clarification. Our paradigm fills the black-box AI model-explainability gap in actual clinical decision-making and sparks trust, accountability, and patient-informed care.