Diagnostic Model that Takes Medical Preferences into Account. Prediction of the Clinical Status of Prostate Cancer

Abstract

Abstract—A mathematical model is proposed to describe and solve problems of medical diagnostics and forecasting based on a risk criterion. Within the framework of the model, problems with ranked diagnoses are considered, whose solution benefits from taking medical preferences into account. A diagnostic algorithm, which is the implementation of this model, is used to solve the problem of predicting the clinical status of prostate cancer. A comparative analysis of the quality of the prediction for four model options was carried out, informative prognostic indicators were revealed, and the results were interpreted. Taking medical preferences into account increases the accuracy of prediction for patients with more frequent and aggressive tumor process due to loss of accuracy for patients with less frequent and aggressive tumor process.