The ELCAP guidance sets standards for responsible AI use in oncology, protecting patients and supporting clinicians
LUGANO, Switzerland and BERLIN, Oct. 20, 2025 /PRNewswire/ -- As artificial intelligence tools like ChatGPT become increasingly common in healthcare, ESMO has taken a decisive step to ensure their safe and effective use in cancer care. ESMO has unveiled its Guidance on the Use of Large Language Models in Clinical Practice (ELCAP), the first structured framework for safely integrating large language models (LLMs) into oncology practice in parallel with a session at the ESMO Congress in Berlin. The paper is published in the Society's peer-reviewed journal Annals of Oncology.
"Innovation must serve oncologists and, ultimately, patients, not confuse or mislead them," said Fabrice André, ESMO President. "With ELCAP, we offer a practical guide for using AI responsibly in oncology, one that protects data, ensures clinical oversight, and supports informed decision-making."
ELCAP divides AI applications into three categories, with specific safety and governance recommendations:
"The value of AI depends on who is using it," said Miriam Koopman, ESMO Real World Data and Digital Health Task Force Chair. "ELCAP sets expectations for each context, ensuring patients are protected, clinicians are supported, and institutions are accountable."
"Trust in AI doesn't come from technology alone, it comes from shared standards," concluded Andrè. "ELCAP is our contribution to making AI a safe and effective ally in oncology."
With 45,000 oncology professionals in its global network, ESMO can be looked up as the global reference for responsible AI adoption in cancer care.
Article here: https://www.annalsofoncology.org/article/%20S0923-7534(25)04698-8/fulltext
- Picture is available at AP -
Contact:
ESMO Press Office press@esmo.org +41 91 9731907
Comunicato stampa sponsorizzato - Responsabilità editoriale PrNewswire. I giornalisti Adnkronos non sono in nessun modo coinvolti né responsabili per i contenuti dei comunicati trasmessi.
