VeriSim adds six literature-derived noise types to patient simulations and shows medical LLMs lose 15-25% diagnostic accuracy under realistic conditions.
The simulator cannot fabricate or alter these invariant facts regardless of noise profile
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.AI 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
VeriSim: A Configurable Framework for Evaluating Medical AI Under Realistic Patient Noise
VeriSim adds six literature-derived noise types to patient simulations and shows medical LLMs lose 15-25% diagnostic accuracy under realistic conditions.