My statistical methods focus is on extensions of the theory of generalized linear models and failure time models to deal with multivariate responses and incomplete data. This inevitable leads to estimation procedures for complex models, including likelihood based EM strategies and Bayesian simulation based inference. The use of "causal modeling" to assess the effect of non-random non-compliance in randomized studies has been a topic of interest for a number of years. I would like in the near future to pursue the extension of this causal modeling to longitudinal observational studies. An earlier interest which has proven useful more recently concerns an extension of the estimation theory for generalized linear models to a situation in which each unit has a different non-linear link function. I also share the urge of many fellow biostatisticians to understand how modern statistical modeling tools are best used in the area of genetic and molecular epidemiology. Interest focuses on linkage disequilibrium mapping and association studies for case-control data and various family designs, including twins. How to deal with complex ascertainment schemes and with longitudinally measured multivariate phenotypes is included in the challenges. Most of my applied research is focused on cancer.
- 1973 MSc (Mathematics), University of Helsinki, Finland
- 1987 PhD (Statistics), University of Helsinki, Finland
- 1989 Docent (Statistics), University of Helsinki, Finland