MedAlign

MedAlign tests instruction following grounded in longitudinal electronic health records and clinician responses.

unassessed

This page is a discovery lead. Nobody has yet assessed it against the catalogue contract, so it carries no disposition. Absence of evidence here is not evidence of staleness.

Category	instruction-following
Subcategory	electronic health records
Page status	active
Metric	COMET and BERTScore
Direction	higher_is_better
Dataset size	303
Publisher	Stanford Medicine and collaborators

What it measures

MedAlign gives a model an instruction or question grounded in an event-stream style patient record. It evaluates whether the model can read the record and produce the clinician-generated completion.

Task format

EHR-grounded natural-language instruction and generated clinical response.

Models reporting this benchmark

No model card in ModelSpec reports this benchmark yet.

Data

This page as JSON · Edit on GitHub