Statistical approaches to revealing structure in complex health data

This thesis develops approaches for the modelling and analysis of structured data, considering two case study datasets containing features such as missing data, irregular time periods, workplace grouping, and spatial observations from different spatial scales. The models developed helped to: create individual health risk profiles over time, obtain ideal locations of health facilities to maximise their coverage, evaluate the impact of health facility access. Overall this thesis makes substantive contributions that extend models to account for data structures, provide corresponding new software tools, improve health surveillance and health resource usage, in the hope of improving health of the public.