Hierarchical Bayesian neural networks: an application to a prostate cancer study

1 INTRODUCTION

The main objective of this article is to explore the neural network technique for predicting certain features indicative of non-organ-confined prostate cancer upon the basis of the be deriveds of certain diagnostic tests administered to patients suffering from prostate cancer. Our approach is full Bayesian and provides the posterior (or predictive) probability of vicinity of these features in the patients based upon certain inputs. Physicians can then make decisions upon the basis of these probabilities, particularly, in marginal cases (eg when these posterior probabilities are in the neighborhood of 50%) to order further diagnostic trials rather than making an