人工智慧 (Artificial Intelligence, AI) 技術正迅速改變全球醫療生態，醫用 AI 系統在疾病診斷、風險預測、個人化治療等領域已展現高度潛力。然而，這類技術同時引發關於倫理原則、風險管理與資訊透明度等議題之廣泛關注。歐盟於 2024 年通過之《人工智慧法案》 (Artificial Intelligence Act, AI Act) 採取風險為主之分類架構，特別對高風險 AI 系統，訂定數據品質、偏差控制、透明度與問責機制等合規要求，並結合《醫療器材法規》 (Medical Device Regulation, MDR) 與《通用資料保護條例》 (General Data Protection Regulation, GDPR)，構成整合性之法規體系，對保障病患權益與促進科技創新形成制度支撐。
本研究透過比較法學方法與文本分析，剖析歐盟 AI 監管框架中針對醫用 AI 的規範機制，並與臺灣現行相關法規 (如《人工智慧基本法》草案、《個人資料保護法》及《醫療器材管理法》) 進行對照，藉此分析兩地於倫理規範、風險評估、法規協調與國際接軌方面之差異與挑戰。研究發現，臺灣目前於高風險AI分類標準、跨部門責任歸屬及透明揭露方面尚未具備系統化機制，導致合規準則不足、問責機制不明確，亦不利於推動本土創新技術之國際化發展。因此，本文建議臺灣應參酌歐盟經驗，建立以風險評估為核心、強化問責與透明度的合規架構，並推動國內外之法規調和，以回應醫用 AI 所帶來的倫理挑戰與制度需求。

Artificial Intelligence (AI) is rapidly transforming global healthcare, with medical AI systems demonstrating remarkable potential in areas such as disease diagnosis, risk prediction, and personalized treatment. However, this technological advancement raises complex ethical, legal, and regulatory concerns, particularly regarding risk management, transparency, and accountability. In response, the European Union (EU) enacted the Artificial Intelligence Act (AI Act, Regulation (EU) 2024/1689), introducing a risk-based classification framework with stringent compliance requirements for high-risk medical AI systems. These include mandates for high data quality, bias mitigation, operational transparency, and accountability mechanisms. The AI Act operates in conjunction with the Medical Device Regulation (MDR, Regulation (EU) 2017/745) and the General Data Protection Regulation (GDPR, Regulation (EU) 2016/679), forming a comprehensive regulatory ecosystem that protects patient rights while supporting innovation.
This study adopts a comparative legal approach to examine the EU's regulatory strategies for Medical AI and contrasts them with Taiwan's current legislative developments, including the draft Artificial Intelligence Basic Law, the Personal Data Protection Act, and the Medical Devices Act. Through legal text analysis, this thesis identifies gaps in Taiwan's legal system, such as the absence of concrete risk classification criteria for AI, insufficient coordination between sectoral laws, and limited mechanisms for transparency and ethical oversight. These shortcomings could hinder the safe deployment of Medical AI and impede Taiwan's ability to align with international standards. Accordingly, this research recommends that Taiwan adopt a forward-looking regulatory framework inspired by the EU model—centered on risk assessment, transparency, and accountability—while promoting regulatory harmonization across legal domains to ensure both ethical integrity and technological progress in the Medical AI field.