Evaluation of the Discrimination Ability of the Decision Rule Allowing to Predict the Probability of Death in Oncosurgical Patients with Comorbid Pathology.

At the moment, there are many decision rules and mathematical models that reduce the risk of postoperative mortality and complications. A small part of such medical mathematical models (scales) is successfully used in practice, but there is also a part that eventually remains on the shelves and becomes morally obsolete. The purpose of this work is to evaluate the discrimination ability of the prognostic model underlying the decision rule that allows ranking patients into groups with favorable and unfavorable outcomes and into a group of patients subject to preoperative preparation to maintain the performance of the mathematical model Oncoprognosis 1.0. The discrimination ability carried out by constructing an area under the Receiver Operating Characteristic (ROC) curve. The investigation allowed conduct that any decision rule requires revision over time, its clarification and, if necessary, adjustments and updates.