Dr Ruth Keogh
BSc MSc DPhil
I joined the Medical Statistics Department at LSHTM in 2012.
I studied Mathematics and Statistics at the University of Edinburgh, an MSc in Applied Statistics at the University of Oxford, and a DPhil in Medical Statistics/Epidemiology also in Oxford. Prior to coming to LSHTM I worked at the MRC Biostatistics Unit, Cambridge (2008-2012) and the Cancer Epidemiology Unit, University of Oxford (2006-2007).
I teach on the MSc in Medical Statistics. This includes organising a module on Survival Analysis, and also teaching modules on Generalized Linear Models and Advanced Research Methods
I enjoy supervising and advising a number of PhD students and act as the Research Degrees Coordinator for the Department of Medical Statistics.
I am currently funded by a Medical Research Council Methodology Fellowship (2015-2019). My interests lie in statistical methodology for the analysis of observational data, and in related applications, especially in epidemiology.
One of my main research interests is in methods for predicting survival using large patient databases using an approach called landmarking. This work is motived by the aim of making dynamic predictions of survival and times to other events for people with Cystic Fibrosis using data from the UK Cystic Fibrosis Patient Registry and the US Cystic Fibrosis Foundation Patient Registry. My prediction models make use of up-to-date measures on an individual’s health status to provide personalised predictions.
I am also interested in how we can estimate causal effects of treatments and exposures on outcomes, especially survival, using observational data.My research interests also include:
- Use of multiple imputation to handle missing data in case-control studies and to make use of full cohort information in nested case-control and case-cohort studies.
- Methods for handling missing data in Cox regression when there are time-varying exposure effects.
- Methods for correcting for the effects of exposure measurement error, especially in the field of nutritional epidemiology.