Pick a modality and a canonical finding. The engine returns the deterministic follow-up recommendation, guideline used, and risk category — no free-text inference.
Loading findings…