Modelling Patient Pathways for Prostate Cancer

Prostate carcinoma is the most commonly diagnosed malignancy affecting men beyond middle age. Although little is known about its direct causes, a set of risk factors and symptoms have been empirically observed. These include patient age, race, family history, health status, raised Prostate-Specific Antigen(PSA) levels in the blood. These are now used as markers to diagnose and assess the presence, grade and stages of prostate cancer in patients. Unfortunately however, neither the tests nor the risk factors are completely conclusive, leading to unnecessary medical procedures with significant financial and emotional costs for both the patient and his family. The medical professional also has to partition and traverse a vast space of parameters in order to decide on an appropriate disease management plan for a particular patient.

Learning the relationship between patient risk factors and treatment outcomes will assist in the selection of an appropriate treatment pathway for an individual patient. In this work we use evolutionary algorithms to evolve Bayesian Networks on patient data. In particular, we investigate the potential of our evolved networks in improving on Partin tables or nomograms, currently used to assist urologists in predicting the extent to which the disease has progressed.

Research Team:

• John McCall
• Ratiba Kabli

Collaborators:

• Eng Ong, Samuel McClinton, James N'Dow, Department of Urology, University of Aberdeen