Research Fellows Directory
Professor S Richardson
Imperial College London
My research programme has progressed in several directions. Firstly, I have focussed on developing new methods to deal with complex data patterns in epidemiology. I have studied different situations where data is missing in such a way that it can create biases when estimating quantities of interest. Using the Bayesian framework and relevant prior information, my work has aimed at giving methodological guidance on issues such model choice for non ignorable missing data and on the sensitivity of inference to the presence of unmeasured confounders. I have also developed decision tools for deriving an improved classification of an ensemble of parameters in hierarchical models or for detecting unusual patterns in short time series, with application to small area data analysis and surveillance.
Secondly, I have worked in the area of statistical genomics and its applications. A wealth of detailed genetics and functional data measured at the level of the genome can now be gathered to capture multi-stage disease mechanisms. It is important to develop efficient tools and algorithms for “integrative statistical methods” that fully exploit the multiple aspects of the data, but remain feasible to use on such large-scale data sets, with thousands of individuals and millions of pieces of information. I have extended my previous work on regression in high dimensional space to consider the joint analysis of thousands of (correlated) responses (phenotypes), such as gene expression measures, which are associated to many predictors, such as genetic markers. Such analyses provide new information towards understanding the function of important genes that influence several phenotypes simultaneously. In parallel, I have been interested in methods for evaluating false discovery rates in the high dimensional context. Finally, I have participated in the application of statistical genomics tools to investigate a variety of health conditions (myelopathy and severe obesity).
Interests and expertise (Subject groups)