Multimorbidity is the presence of at least two diseases in an individual. MM is one of the major challenges faced by health systems across the world. In the UK, the care of multimorbid patients accounts for more than half of the primary and secondary care costs, with an expectation that the cost will rise. This project will utilise the Ninewells Hospital & Medical School dataset that follows the Scottish Tayside and Fife population. We will first identify MM patients, focusing on those with complex MM (more than 4 conditions). We will then utilise Bayesian Mixture Modelling to detect the prevalent clusters of MM (Papathomas et al. 2012, Jing and Papathomas 2021), with patients in the same cluster characterised by conditions of a homogenous range and nature. The use of Bayesian modelling will allow for the evaluation of uncertainty with regard to the number of clusters and also the typical subject profile within each one of the MM clusters. Profile Regression (Liverani et al. 2015), a popular extension of Bayesian Mixture Modelling, will be used to model the effectiveness of different treatments for these patients. As this concerns a large dataset, subsampling techniques will be used, which will lead to further developments in the field of Bayesian Mixture Modelling when handling large datasets.
For more information, please see the School's Postgraduate Research page, and in particular the information about Statistics PhD opportunities.