Modeling the Survival of Hip Fracture Patients Using a Conditional Phase-Type Distribution

Previous research has introduced the conditional phase-type distribution as a model that can suitably represent a skewed survival distribution conditioned on a network of inter-related variables. The technique has successfully been applied to modeling the stay of elderly patients in hospital which typically includes extreme stays in hospital resulting in a highly skewed survival distribution. This paper uses the Conditional phase-type distribution in a different setting whereby the focus is on patients who are in hospital due to having suffered a hip fracture. The model represents the patient length of stay in hospital conditioned on a Bayesian network of inter-related patient variables. In particular, the network highlights the key factors influencing length of stay and identifies relationships between long stay patients, delays in surgery and the patients condition. The delays identified in the system, could therefore be addressed to hopefully reduce the length of stay in hospital and increase the flow of patients in the hospital system.