Research in the Combining Health Information, Computation and Statistics Group
Research in CHICAS is concerned with the development and application of novel statistical methods motivated by substantive problems in the biomedical and health sciences.
Current methodological research interests within the group include: spatial and spatio-temporal statistical models and methods, longitudinal data analysis and latent graphical modelling.
We engage in collaborative research links with many clinical and academic institutions, nationally and internationally. Current application areas include: human and veterinary epidemiology; real-time disease surveillance; environmental exposure measurement; tropical disease prevalence mapping; longitudinal studies in cardio-thoracic surgery, cognitive psychology, mental health and renal medicine.
Brief summaries of our main areas of current research follow.
Environmental Epidemiology deals with issues of environmental exposures and their health effects. We study the determinants of the distribution of disease, as well as the environmental conditions and hazards that may pose risks to human health. Identifying and quantifying the adverse health-effect of exposures to environmental contaminants assists the conduct of risk assessment and surveillance and may add to our understanding of disease aetiologies. Current applied projects in this area include studies of adverse birth outcomes (with University of Newcastle upon Tyne), campylobacter (with Preston Public Health Laboratory and Liverpool University Veterinary School) and childhood meningitis (with Alder Hey Hospital, Liverpool).
Spatial Epidemiology focuses on the description and analysis of geographic variations in disease, and relates these to demographic, environmental, behavioural or genetic risk-factors and infectious transmission processes. We are working on a range of applications in disease mapping, geographic correlation studies and disease clustering. The advances in geographic information systems, in statistical and geostatistical methodology, as well as the increasing availability of high-resolution spatially or spatio-temporally referenced health and environmental quality data enable a fuller understanding of observed patterns of spatial variation in disease incidence and prevalence. Spatial statistical modelling is especially useful in developing country settings, where the lack of census or registry data requires the formulation, fitting and validation of spatial stochastic models for spatial interpolation and prediction of disease prevalence.
In collaboration with the Johns Hopkins University School of Public Health, Baltimore, WHO Tropical Disease Research, Geneva, and the International Research Institute for Climate and Health, New York, we are modelling spatio-temporal incidence patterns for a variety of endemic tropical diseases including Loa loa, malaria and meningitis.
Longitudinal data analysis
Longitudinal data analysis is needed in any study where subjects are followed up over time. Our methodological work in this area includes joint modelling of multivariate repeated measurement sequences and time-to-event outcomes, and hierarchically specified models with latent Gaussian graphical structure and non-standard sampling distributions.
In collaboration with colleagues from Lancasterís Department of Psychology we are investigating causal models of executive behaviour in pre-school children and in patients diagnosed with Alzheimerís. We are also collaborating with colleagues from the Spectrum Centre for mental health research into the analysis of longitudinal studies of bipolar disorder, and with colleagues in the Biomedical Sciences Unit into the evaluation of novel biomarkers for Parkinsonís disease.
In collaboration with Hope Hospital, Salford, we are developing dynamic time series models for real-time monitoring of progression to renal failure in primary and secondary care patients. With the Brompton Hospital, London, we are modelling long-term outcomes following cardio-thoracic surgery.
Microarray data analysis
Microarray data analysis The analysis of microarray data remains an active area of research since the invention of the technology in the late nineties. A basic objective in many microarray experiments is to determine sets of differentially expressed genes, which may provide clues to genetic pathways. The focus of our work in this area is to adopt a model-based approach to the analysis of gene expression data, in which multi-dimensional questions of scientific interest are addressed by computing predictive probabilities, rather than through the more traditional approach of significance testing.
For further details of research in CHICAS please contact Cathy Thomson.