Yael Waknine, Founder and Principal of Oncology Medical Education at MedPoint Solutions, shared on LinkedIn:
“What happens if you hand an AI a stack of recent EGFR-mutant NSCLC abstracts and ask for a CME deck?
Probably something like this: The AI notices that ivonescimab and pembrolizumab both hit PD-1.
It concludes they’re basically the same drug. They are not the same drug.
Slide 7 cites HARMONi-A to argue that pembrolizumab plus chemo is a solid option after EGFR-TKI progression.
Slide 12 cites KEYNOTE-789, where pembrolizumab plus chemo after TKI failure missed its PFS and OS endpoints.
Slide 19 cites osimertinib resistance data to support immunotherapy after progression. The data are about MET amplification and small-cell transformation. Neither is an indication for pembrolizumab.
No slide mentions what happens when osimertinib follows a recent checkpoint inhibitor. (Pneumonitis. Pneumonitis is what happens.)
By slide 23, the AI has synthesized two years of the field into the medical equivalent of a man arguing with himself in a parking lot.
The peer reviewer asks whether the learning objectives are ‘intentionally Socratic.’
The lesson:
‘AI reads the literature the way a golden retriever reads Kant.’
Tremendous enthusiasm. Genuinely no idea what’s happening. Occasionally sits on something important.
AI can draft a deck in an afternoon. Catching slide 7 still takes someone who knows a PD-1/VEGF bispecific isn’t pembrolizumab with extra steps.
The parking lot guy is still out there, though.”
You can also read: HARMONi-2 Shows Overall Survival Benefit With Ivonescimab Over Pembrolizumab in First-Line PD-L1–Positive NSCLC
