The European Society for Medical Oncology (ESMO) has released the Basic Requirements for AI-Based Biomarkers in Oncology (EBAI) framework, "the first comprehensive guidance for the safe, trustworthy and effective integration of AI-derived biomarkers in cancer care." The framework provides guidance for developers, clinicians, regulators and healthcare institutions on means of integrating digital technologies in cancer care.
To view the complete framework, click here.
Source mentioned:
ESMO Basic Requirements for AI-based Biomarkers In Oncology (EBAI)
DOI: 10.1016/j.annonc.2025.11.009
First-ever guidance on validating AI-based biomarkers is overdue, because biomarkers built by algorithms need the same rigour as any laboratory test before they influence treatment. Without clear validation requirements, promising models risk entering practice on performance claims that do not hold up in real clinics. Standards like this are regularly discussed on onc brothers, where the test for any new tool is whether it survives contact with everyday oncology care.
ReplyDeleteThe EBAI framework really is a watershed moment — holding algorithm-derived biomarkers to the same validation rigour as any laboratory test is exactly what the field needed. The most promising part is the multi-stakeholder framing: developers, clinicians, regulators, and healthcare institutions all speaking the same language about what "valid" actually means. Without standards like this, impressive retrospective AUCs keep crumbling on contact with real clinics, and patients are the ones who pay for that gap.
ReplyDelete