平特五不中

Michael Wallace, PhD

Postdoctoral Fellow, Department of Epidemiology, Biostatistics and Occupational Health, 平特五不中

Personalizing Medicine: New Ideas for Dynamic Treatment Regimes

ALL ARE WELCOME

Abstract:

Personalized medicine is a rapidly expanding area of health research wherein subject level information is used to inform treatment. Dynamic treatment regimens (DTRs) are one means by which personalized medicine can be studied theoretically and applied in practice. DTRs are sequences of decision rules which take subject information as input and provide treatment recommendations as output. Such regimens therefore tailor each treatment decision to a patient's unique circumstances, but can also identify management plans which optimize long-term outcomes by accommodating potentially obscure delayed treatment effects and other complex interactions. However, taking such factors into account can complicate the problem of causal inference in this context. One approach considers the blip: a structural nested mean model of the expected difference in the (potentially counterfactual) outcome when using a baseline treatment instead of the observed treatment. DTR estimation in this context therefore relies on estimating blip parameters and numerous methods have been proposed for this purpose. In this talk I present an approach which uses standard weighted ordinary least squares regression to control for the potentially confounding effects of covariate-dependent treatment. This builds on two established methods: Q-learning and G-estimation, offering the doubly-robust property of the latter but with ease of implementation akin to the former. I'll outline the underlying theory and demonstrate the double-robustness and efficiency properties of the approach through illustrative examples. Finally I'll discuss model assessment, demonstrating diagnostic plots for the method, and how the double robustness property itself may be leveraged to investigate model validity.

Bio:

Michael Wallace is a postdoctoral fellow in the Department of Epidemiology and Occupational Health, working with Professor Erica Moodie (also EBOH) and Professor David Stephens (Mathematics and Statistics). His research focuses on developing methodology for the identification of dynamic treatment regimes: disease management plans that vary depending on patient-level information. Michael received his undergraduate training in mathematics at Trinity College, Cambridge, before pursuing a Master's in statistics at University College London. His PhD thesis, which concerned covariate measurement error in regression modelling, was completed at the London School of Hygiene and Tropical Medicine. Outside research, Michael has a strong interest in promoting statistics (and statistical thinking) to those beyond the statistical community, and serves on the Editorial Board of the Royal Statistical Society/American Statistical Association magazine Significance.