What do we do?

We develop and apply statistical machine learning and computational statistics methodology to answer biologically and epidemiologically-driven questions.  We develop spatial and spatio-temporal models for disease mapping and for risk assessment and environmental exposure estimation. We use advanced statistical approaches to explore high-dimensional complex genetic, epigenetic, transcriptomic, proteomic and metabolomic  datasets to model associations between genes and disease, identify features driving disease dynamics and progression, and to uncover corresponding cellular networks, pathways, and biomarkers.

How do we do it?

We answer key statistical questions to address the above challenges by employing both Bayesian and Frequentist approaches. We develop computationally efficient methods  for profiling from high-throughput platforms and fitting complex spatial and spatio-temporal models. We use Markov chain Monte Carlo approaches, shrinkage and regularisation methods,  approximate Bayesian computation such as INLA and simulated likelihood methods, as well as non-parametric Bayesian models. In addition, we employ latent-variable models such as partial least squares and develop computational methods for high-dimensional model selection and signal identification.

Why is it important?

Advanced statistical methods are crucial for interpreting and making sense of modern complex biomolecular, epidemiological and environmental data. Modern biostatistical methods help to better understand the genetic structure of diseases, disease progression and pathways, and to study the impact of environmental exposures on chronic diseases.