TLDR: The European Society for Medical Oncology (ESMO) has released its pioneering “ESMO Guidance on the Use of Large Language Models in Clinical Practice” (ELCAP). These guidelines, developed by a 20-member international panel, aim to ensure the safe, effective, and ethical integration of AI language models into oncology. ELCAP categorizes LLM applications into three types: patient-facing, healthcare professional-facing, and background institutional systems, providing specific recommendations for each to balance innovation with patient safety and data protection.
LUGANO/BERLIN – The European Society for Medical Oncology (ESMO) has introduced its groundbreaking “ESMO Guidance on the Use of Large Language Models in Clinical Practice” (ELCAP), marking the first comprehensive framework for the safe and effective integration of artificial intelligence (AI) language models in oncology. The guidance was officially published in ESMO’s peer-reviewed journal, Annals of Oncology, on October 18, 2025, coinciding with a dedicated session on ChatGPT and cancer care at the ESMO Congress 2025 in Berlin, highlighting the escalating importance of AI in cancer treatment and research.
Developed between November 2024 and February 2025, ELCAP is the product of an extensive effort by a 20-member international panel. This diverse group comprised experts from various fields, including oncology, AI, biostatistics, digital health, ethics, and patient advocacy, all convened under the auspices of the ESMO Real World Data & Digital Health Task Force.
ESMO President Fabrice André emphasized the society’s commitment to patient benefit and clinical practicality. “ESMO’s priority is to ensure that innovation translates into measurable benefit for patients and workable solutions for clinicians,” stated André. He added, “With ELCAP, ESMO provides a pragmatic, oncology-specific framework that embraces AI while upholding clinical responsibility, transparency and robust data protection.”
The guidelines acknowledge that the opportunities and risks associated with Large Language Models (LLMs) vary significantly depending on the user – whether they are patients, clinicians, or institutions. To address this, ELCAP structures its recommendations into three distinct categories, translating high-level principles into 23 consensus statements for daily practice:
Type 1: Patient-facing applications – These include tools such as chatbots designed for patient education and symptom support. ESMO recommends that these applications should complement existing clinical care, operate within supervised pathways, include explicit escalation protocols, and ensure robust data protection.
Type 2: Healthcare professional-facing tools – This category encompasses applications like decision support systems, documentation aids, and translation tools for clinicians. Such tools necessitate formal validation, transparent disclosure of their limitations, and clear human accountability for all clinical decisions made with their assistance.
Type 3: Background institutional systems – These are LLMs integrated with electronic health records for tasks such as data extraction, automated summaries, and matching patients to clinical trials. These systems require rigorous pre-deployment testing and continuous monitoring for potential biases and performance fluctuations.
Professor Miriam Koopman from the University Medical Center Utrecht, Netherlands, a co-author of the paper, highlighted that while LLMs offer significant potential to support medical oncologists in clinical decision-making and reduce administrative burdens, they also carry risks such as introducing biases or providing less tailored information to patients.
André concluded, “Responsible use of AI in oncology requires shared standards as much as it requires algorithms; ELCAP sets out how to deploy language models in ways that improve the quality, equity and efficiency of cancer care, without compromising trust in clinical judgement.”
The European Health Data Space (EHDS) Regulation, which came into force in March 2025, is expected to further support the implementation and validation of LLMs by establishing a common framework for health data exchange across the EU and harmonizing data privacy and governance.
Also Read:
- American Medical Association Launches Center for Digital Health and AI to Reshape Healthcare
- BostonGene Secures Five 2025 Stevie® Awards for Pioneering AI in Oncology Technology
ESMO continues to champion digital innovation in cancer care, with the inaugural ESMO AI and Digital Oncology Congress 2025 scheduled for November 12-14, providing a dedicated platform for the latest advancements in this transformative field.
