Storia: Why benchmarking clinical LLMs from OpenEvidence, Doximity is complicated — Warptech Lab News