Project overview

Over the last three years, research clinicians at Aberdeen Royal Infirmary (ARI) and computer scientists at Robert Gordon University (RGU) have conducted a pilot study to investigate the feasibility of applying Bayesian Networks to prostate cancer patient data. In this collaborative study, a cohort of 320 prostate cancer patient cases was assembled. The collection consists of retrospective data from patients diagnosed and treated for prostate cancer at ARI over a period of 2.5 years during 2002-2004. Bayesian Networks were constructed on up to 37 different patient and clinical variables demonstrating the power of the approach in handling large quantities of complex interacting data. As part of this work, the technique was successfully applied to construct nomograms, producing credible probability distributions from this limited data set. There is now an opportunity to apply this approach to create and validate nomograms based on a comprehensive UK data collection. This has the potential to enhance clinical practice in the UK.