Prediction of the Aggressive Status of Prostate Cancer on the Basis of Preoperative Data

Abstract

The problem concerning the prediction of the aggressive status of prostate cancer (PCa) is examined on the basis of preoperative data. This problem is solved using data on 360 patients with the established (aggressive or indolent) postoperative status of the disease. The collection of factors containing five informative indicators of prediction (from primarily accessible sixteen) is revealed and employed to create the diagnostic index used to predict the aggressive PCa status. In compliance with cross-validation data, the prognostic algorithm enables us to find the group involving 55% of patients with an aggressive status in the absence of patients with an indolent PCa status. The three-class prediction algorithm making it possible to ascertain whether any patient belongs to the group with the low, high, or uncertain risk of the aggressive disease stage is proposed.