Fig. 3.

The ratio of false-positive SDRs confounded by indication or disease spill-over and unclassified SDRs relevant for further manual validation; the ratio should ideally be as close to zero as possible, with as few confounded SDRs as possible (numerator) delivered by the method in relation to the relevant SDRs (denominator). From left to right for each drug analysis: the ratios when analyzing by the conventional PRR, PRR-TA (model 2 and 6), and for bicalutamide and abiraterone, also the PRR-TA (model 3)