Table 3. Barzooka screening tool performance in an external validation set of publications from authors affiliated with Charité Universitätsmedizin—Berlin.

Class	Manually labeled cases	False positives	Precision	Recall	F1 score
Flow chart	123	20	0.84	0.87	0.86
Pie chart	38	4	0.89	0.87	0.88

The Charité external validation set included 1000 papers published between 2015 and 2019.