Elsevier Unveils Rigorous Evaluation Framework to Mitigate Risk in Generative AI Clinical Decision Support Tools 

Elsevier, a global leader in medical information and data analytics, unveiled a groundbreaking evaluation framework for assessing the performance and safety of generative AI-powered clinical reference tools. This innovative approach has been developed for all Elsevier Health generative AI solutions, including ClinicalKey AI, Elsevier's advanced clinical decision support platform, and sets a new standard for responsible AI integration in healthcare. It will be featured in a future issue of the Open Access Journal of the American Medical Informatics Association (JAMIA).  

The framework, designed with input from clinical subject matter experts across multiple specialties, evaluates AI-generated responses along five critical dimensions: query comprehension, response helpfulness, correctness, completeness, and potential for clinical harm. It serves as a comprehensive assessment to ensure that AI-powered tools not only provide accurate and relevant information but also align with the practical and current needs of healthcare professionals at the point of care. 

Omry Bigger, President of Clinical Solutions at Elsevier: “This evaluation framework not only supports innovation and advancements to improve patient care but adds an extra layer of review and assessment to ensure physicians are armed with the most accurate information possible. It’s a critical step in the implementation of responsible AI for healthcare providers and patients.”  

In a recent evaluation study of ClinicalKey AI, Elsevier worked with a panel of 41-board certified physicians and clinical pharmacists to rigorously test responses generated by the tool for a diverse set of clinical queries. That panel evaluated 426 query-response pairs, and results demonstrated impressive performance, with 94.4% of responses rated as helpful, 95.5% assessed as completely correct, with just 0.47% flagged for potential improvements. 

Leah Livingston, Director of Generative AI Evaluation for Health Markets at Elsevier, said: “These results reflect not just strong performance, but the real value of bringing clinicians into the evaluation process. By designing an evaluation framework around what matters most to physicians—accuracy, relevance, and clinical safety —we’re helping ensure that AI tools truly add value to care delivery. This approach supports clinicians in quickly accessing the right information, ultimately reducing cognitive burden.” 

Elsevier is continuing to implement AI responsibly in its portfolio of AI solutions and is also involved in industry-wide initiatives. As a proud partner of the Coalition for Health AI, the company is actively contributing to industry-wide standards for responsible AI deployment in healthcare settings. 

The release of the evaluation framework represents a significant step forward in the responsible integration of AI technologies in healthcare, paving the way for more efficient, accurate, and patient-centered clinical decision-making.  

To learn more details about the evaluation framework, a white paper is available for download: Evaluation framework for generative AI tools used for clinical decision support. 

關於 Elsevier

身為科學資訊與分析的全球領導者，Elsevier 協助研究人員與醫療照護專業人員推動科學發展，改善醫療成果，造福社會。我們以可信賴、以實證為基礎的內容和先進的 AI 數位技術為基礎，透過創新的解決方案促進洞察力和關鍵決策。

140 多年來，我們一直為研究和醫療保健界的工作提供支援。我們全球 9,500 名員工，包括 2,300 名技術人員，致力於支援研究人員、圖書館館長、學術領袖、資金提供者、政府、研發密集型公司、醫生、護士、未來醫療保健專業人員和教育工作者的重要工作。我們的 2,900 種科學期刊和經典參考工具書包括其領域中最重要的書籍，包括 Cell Press、The Lancet 和 Gray's Anatomy。 我們與愛思唯爾基金會 (Elsevier Foundation 打開新的分頁／視窗) 合作，與我們服務的社群攜手合作，在發展中國家和世界各地的科學、研究和醫療保健領域推動包容性和多樣性。 Elsevier 是 RELX 打開新的分頁／視窗 的一部分，RELX 打開新的分頁／視窗 是一家為專業和商業客戶提供以資訊為基礎的分析和決策工具的全球供應商。有關我們的工作、數位解決方案和內容的更多資訊，請造訪 www.elsevier.com 。