Choosing an optimal treatment strategy for patients with benign prostatic hyperplasia using a neural network

In this study, we analyzed the role of several etiopathogenetic and clinical factors important for the prediction of conservative treatment outcomes in patients with benign prostatic hyperplasia (BPH). The study included 849 patients who received conservative treatment (n = 441) or underwent surgery (n = 408) for BPH between 2012 and 2015. Some clinical features of BPH were found to increase the risk of conservative treatment failure. We have developed a neural network for predicting the outcome of conservative treatment for BPH. We recommend using a three-layer network with the number of neurons in the output layer equal to the possible number of outcomes of the disease. The neural network ensures the choice of an optimal treatment strategy by changing the value of the vector of the input layer of neurons and estimating the value of the output vector.