The first-ever guidelines for AI use in oncology are now available
Industry Buzz
AI in oncology is only as ethical as the rules clinicians help build around it.
—Stacey Lee, JD, associate professor of health policy and management at the Johns Hopkins Bloomberg School of Public Health
The European Society for Medical Oncology (ESMO) released new guidance last week outlining best practices for integrating AI language tools into oncology. []
According to the agency, the goal is to ensure that innovation leads to a “measurable benefit for patients and workable solutions for clinicians.” []
The recommendations, titled ESMO guidance on the use of Large Language Models in Clinical Practice (ELCAP), were published in the Annals of Oncology. []
Stacey Lee, JD, associate professor of health policy and management at the Johns Hopkins Bloomberg School of Public Health, says that clinicians are already using AI and navigating it daily, often without clear guardrails. “This kind of structure gives developers and clinicians a shared framework for responsible innovation. It's not about slowing progress; it's about giving it clinical credibility,” Lee tells MDLinx.
Opportunities and risks
The guidance provided a three-tier framework for how to safely use large language models (LLMs) in cancer care. It covered three categories:
Patient‑facing applications: Chatbots used for condition education and symptom support
Physician-facing tools: Decision support, along with decision documentation and translation assistance
Background institutional systems: Data extraction, automated data analysis, and clinical trial matching
The authors also highlighted opportunities and potential risks that clinicians should be aware of when using AI. For example, healthcare providers should continuously monitor their tools for biases and changes in performance. The patient-facing applications should complement clinical care, not replace it.
“AI in oncology is only as ethical as the rules clinicians help build around it,” says Lee.
Related: 5 specialties AI can’t replace (yet)Consensus on the need for regulation
Regardless of whether healthcare workers support or oppose AI, most agree that it needs to be regulated.
Jennifer Kang-Mieler, PhD, department chair and Director of the Center for Healthcare Innovation at Stevens Institute of Technology, hopes that the guidance will build confidence in AI systems and promote collaboration among researchers and clinicians.
“The standardization of guidelines [is] essential to ensure that the models are used correctly and effectively, prevent errors or bias, and protect privacy,” Dr. Kang-Mieler tells MDLinx.