Evaluation science

How healthcare AI is actually measured: judges, error rates, severity grading and why the numbers rarely compare.