skip to content

MRC Biostatistics Unit

Summary

My research interests include methodology for drawing causal inferences from observation data and for handling selection biases and missing data.  A causal inference problem that particularly interests me is time-dependent confounding.  This arises when a time-varying exposure (e.g. treatment) is affected by time-varying confounders that are themselves affected by the exposure.  This confounding complicates the estimation of treatment effects and the identification of optimal treatment regimes for individual patients.  Selection biases of interest include those arising from having an unrepresentative sample or an observation processes that is not independent of the process of interest.  For example, electronic health records may contain more information on sicker patients than on healthier patients.  My missing data interests include foundations of missing data theory, multiple imputation methods, inverse probability and covariate-balancing weights, and double robust methods.  I also have long-standing interests in risk prediction, HIV/AIDS epidemiology and injecting drug use epidemiology.
 

Selected Papers

Samartsidis P, Seaman SR, Harrison A, Alexopoulos A, Hughes GJ, Rawlinson C, Anderson C, Charlett A, Oliver I and De Angelis D. (2024).  A Bayesian multivariate factor analysis model for causal inference using time-series observational data on mixed outcomes. Biostatistics 25: 867-884

Seaman SR, Samartsidis P, Kall M and De Angelis D. (2022) Nowcasting COVID-19 deaths in England by age and region. Journal of the Royal Statistical Society Series C (Applied Statistics) 71: 1266-1281

Farewell D, Daniel R and Seaman S. (2022) Missing at random: a stochastic process perspective.  Biometrika 109: 227-241

Seaman S, Dukes O, Keogh R and Vansteelandt S. (2020) Adjusting for time-varying confounders in survival analysis using structural nested cumulative survival time models. Biometrics 76: 472-483

Wen L. and Seaman S.R. (2018) Semi-parametric methods of handling missing data in mortal cohorts under non-ignorable missingness. Biometrics 74: 1427-1437

Seaman S.R. and Vansteelandt S. (2018) Introduction to double robust methods for incomplete data. Statistical Science 33: 184-197

S.R. Seaman and R.H. Hughes (2018) Relative efficiency of joint-model and full-conditional-specification multiple imputation when conditional models are compatible: The general location model. Statistical Methods in Medical Research 27: 1603-1614

S.R. Seaman, R.H. Keogh (2015) Handling missing data in matched case-control studies using multiple imputation. Biometrics 71: 1150-1159

S.R. Seaman, M. Pavlou, A.J. Copas. (2014) Methods for Observed-Cluster Inference when Cluster Size is Informative: a Review and Clarifications. Biometrics 70: 449-456

S.R. Seaman, J. Galati, D. Jackson, J. Carlin. (2013) What is Meant by "Missing at Random"? Statistical Science 28: 257-268