Mini-project
Healthcare & Hospitals
Clinical notes extractor
Pull symptoms, diagnoses and medicines out of clinical notes and summarise each visit.
The brief
Doctors' notes hold the most useful information in a hospital, but as free text nobody can search. Build a pipeline that pulls out symptoms, diagnoses and medicines from de-identified notes, and writes a short summary of each visit. Good looks like: accuracy measured on hand-labelled notes, a comparison of a medical NER model with an LLM, and summaries a doctor finds correct.
What you will build
- Select and label 50 notes by hand
- Extract entities with a medical NER model
- Extract the same with an LLM and compare
- Summarise each visit in a fixed format
- Report accuracy and the errors that matter clinically
Free, open data to build it with
Check each dataset's licence before you publish your work. Your mentor gives your team its own version of the brief (data, scope or region), so every team's project is its own.