Figure 2
From: Reliable anti-cancer drug sensitivity prediction and prioritization
Drug prioritization pipeline. This figure shows a drug prioritization pipeline originating from combining the CMax viability with SAURON-RF and CP. The output of the CP pipeline (cf. Fig. 1) deployed with our SAURON-RF method are sets for the classification task and intervals for the regression task. Here, sets that contain only one element indicate that we can be confident about the initial point prediction (single class) of the trained model. Thus, we can identify effective drugs by filtering for sets solely comprising the class corresponding to drug sensitivity (1: sensitive). Due to the across-drug comparability of the CMax viability, we can rank these drugs by their predicted efficiency using, for example, the upper limit of the regression interval that represents a value not being surpassed with high probability.
