Validating event counts before plotting a clinical evidence table
A percentage alone does not reveal the denominator, so an extraction row should keep the event count, participant total, study arm and an exact source location. The article offers a small dependency-free Python validator that uses synthetic teaching data and rejects booleans, non-integers, impossible counts, zero denominators and missing source locators. It does not infer missing values or back-calculate counts from rounded percentages. The example's valid rows yield proportions of 0.1 and 0.3, and its negative cases confirm that invalid rows are refused. The author stresses that passing these structural checks does not confirm the copied numbers, the randomized population or outcome comparability, and that recurrent-event data needs a different schema.
This is an AI-generated summary. ShortSingh links to the original source for the complete article.

Discussion (0)
Log in to join the discussion and vote.
Log in