Fig. 1

Figure reports the pipeline workflow, consisting of (from left side): (i) Prostate serum and EPS-urine datasets; (ii) the preprocessing phase, allowing to remove the inconsistent values and to correct the missing values; (iii) the feature selection phase allows us to keep only the most important features to improve the application of the classification algorithms; (iv) the ML phase, consisting in choosing among five different classification algorithms; (v) finally, the voting phase, which consists of a soft vote and hard vote, to support disease prediction using two unknown datasets (one for serum and one for urine respectively)