Scholarlink

Schwannomatosis is a rare tumor disease characterized by multiple benign nerve-sheath tumors. The complexity of diagnostic pathways and management guidelines highlights the need for robust decision-support tools for health professionals. This paper presents an Explainable AI (XAI) system designed to assist in the diagnosis and management of Schwannomatosis disease. The XAI system integrates ERN GENTURIS clinical guidelines into Controlled Natural Language (CNL) within the Cognica framework transforms them into structured decision logic. By translating natural language policies into transparent, interpretable recommendations the system enhances diagnostic accuracy and supports informed treatment decisions. Through simulated case studies, we evaluate its effectiveness in improving trust and usability among clinicians. We also discuss current system limitations and propose future enhancements for expanding clinical applicability.

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

An Explainable AI System for Clinical Decision Support in Schwannomatosis

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Melpo Pittara,
  • Anastasia Kyriacou,
  • Adamos Koumi,
  • Maria Matsangidou,
  • Eirini Schiza,
  • Constantinos S. Pattichis,
  • Antonis Kakas

摘要

Schwannomatosis is a rare tumor disease characterized by multiple benign nerve-sheath tumors. The complexity of diagnostic pathways and management guidelines highlights the need for robust decision-support tools for health professionals. This paper presents an Explainable AI (XAI) system designed to assist in the diagnosis and management of Schwannomatosis disease. The XAI system integrates ERN GENTURIS clinical guidelines into Controlled Natural Language (CNL) within the Cognica framework transforms them into structured decision logic. By translating natural language policies into transparent, interpretable recommendations the system enhances diagnostic accuracy and supports informed treatment decisions. Through simulated case studies, we evaluate its effectiveness in improving trust and usability among clinicians. We also discuss current system limitations and propose future enhancements for expanding clinical applicability.