ad-kat/clinical-contradiction-detector ? reverse-engineered prompt

Reverse engineered prompt

Build me a clinical note checker for hospital records that looks at a patient’s past discharge summaries and flags when something doesn’t add up across visits.

I want it to take in two or more notes for the same patient, compare them, and catch cases like a medication being listed even though there’s a documented allergy, or a diagnosis that seems to disappear and then show up again without a clear reason. If a pair of notes looks suspicious, it should do a deeper extraction step, then return a clear finding with the contradiction type and a simple severity label like high, medium, or low.

Please include a small web dashboard where I can paste or load notes, see the flagged contradictions, and click into the reason it was flagged. It should feel like a working demo with synthetic patient data, and the backend should be easy to run locally in Docker. If you need to look up current docs for any libraries or model loading details, go ahead and do that.

Are you gonna build this?

make sure you review the code using coderabbit

Try freeSponsored — opens CodeRabbit in a new tab