Updated
Updated · Oncodaily · Jul 27
Yan Leyfman Launches 3-Disease AI Benchmark for Hematologic Cancers
Updated
Updated · Oncodaily · Jul 27

Yan Leyfman Launches 3-Disease AI Benchmark for Hematologic Cancers

1 articles · Updated · Oncodaily · Jul 27

Summary

  • A preregistered benchmark covering AML, CLL and mantle cell lymphoma is recruiting hematologists and oncologists to supply the physician judgments that AI models will be measured against.
  • The project targets disputed clinical decisions where NCCN, ELN and iwCLL guidance stops short and expert clinicians can reasonably choose opposite approaches.
  • Participants are asked to spend 5-10 minutes on anonymous surveys with no patient data, and can complete one, two or all three disease modules before the August deadline.
  • Arturo Loaiza-Bonilla said the effort aims to move AI testing beyond straightforward guideline questions toward harder real-world cases that "keep hematologists up at night."

Insights

Will using human doctors as the ultimate benchmark prevent AI from discovering superior cancer treatments?
How will AI navigate life-or-death cancer decisions when even top medical experts completely disagree?