Table 3.

Average recall, precision, and F-measure of the deep learning approach and McMaster’s Balanced Treatment filter according to the Clinical Hedges gold standard (N=20).

Measure	Deep learning (%)	McMaster’s CQa balanced filter (%)	P value
Recall	96.9	97.0	.63
Precision	34.6	40.9	<.001
F-measure	51.0	57.5	<.001

^aCQ: PubMed Clinical Query Treatment filter