Professor Mark S Gilthorpe

Professor Mark S Gilthorpe

Professor of Statistical Epidemiology

0113 343 1913

Summary: Statistical Epidemiology

Location: Room 11.21, Leeds Institute for Data Analytics (http://lida.leeds.ac.uk/)

Teaching Commitments: Module manager and teacher on the MSc in Statistical Epidemiology for "Introduction to Modelling" and "Advanced Modelling Strategies". Programme Lead and Demonstrator on the Summer School "Advanced Modelling Strategies: challenges and pitfalls in robust causal inference with observational data".

Overview

Qualifications

BSc, PhD, FHEA

Research Area

In observational research, evidence of underlying association between an exposure and subsequent disease is difficult to demonstrate unequivocally. There are many situations in which we seek causal insight that cannot be obtained through experimentation; evidence of causation may then only be inferred from observational data. Epidemiology, the methodological foundation of observational research, provides a structure for the conduct of biomedical observational research through appropriate study design and statistical analysis.

Existing study designs continue to experience analytical challenges and we also face new challenges with ‘unstructured’ health data: i.e. data that are not obtained through traditional ‘study design’, but acquired through other means. So-called ‘big data’ (small or large), derived through means of routine collection and possibly linked to other datasets, now yield complex observational data that require new methodological approaches to data analysis.

It is within this context of observational data research that my contributions revolve around: ending poor statistical practices in observational data analytics; developing new observational data analytic methods (particularly for seemingly intractable problems); and promoting robust causal inference for observational research. I employ philosophical reflection into data generation processes and combine this with graphical model theory to explain and resolve poor statistical practices in the analysis of observational data, developing new methods or adapting existing techniques to improve analytical rigour.