About the seminar
Optimal prophylaxis and treatment regimens for a disease may be best evaluated using a sequential multiple assignment randomised trial (SMART). A SMART is a multi-stage study that randomises a participant to an initial treatment then, depending on their observed response, randomises the same participant to an alternative treatment. Response-adaptive randomisation may, in some settings, improve the trial participants’ outcomes and expedite trial conclusions, compared to fixed randomisation. But `myopic’ response-adaptive randomisation strategies, blind to multistage dynamics, may also result in suboptimal treatment assignments.
We propose a `dynamic’ response-adaptive randomisation strategy based on Q-learning, a well-known approximate dynamic programming algorithm that uses backward induction to model multistage treatment and outcome dynamics. Our motivation was a real-world COVID-19 prophylaxis and treatment SMART with qualitatively distinct binary endpoints at each stage. Sequences of binary endpoints cannot be modelled with standard Q-learning and sequential binary endpoints may also require qualitatively distinct weightings to ensure that the design evolves to assign participants to regimens with the highest utility. We describe how a simple decision-theoretic extension to Q-learning can be used to model sequential binary endpoints with distinct utilities. Using simulation, we show that, under a set of binary utilities, a `dynamic’ approach increases expected utilities compared to the fixed approach, sometimes markedly, for all model parameters, whereas a `myopic’ approach can decrease expected utilities.
About the speaker
Dr Robert Mahar received a PhD in Biostatistics at the University of Melbourne in 2019 following his completion of a Master of Biostatistics at the University of Queensland in 2014. He relocated from coastal Victoria to Brisbane at the start of 2022 and has since joined QUT as a Visiting Fellow. He continues to hold a full-time joint appointment as a Research Fellow in Biostatistics at both the University of Melbourne School of Population and Global Health Centre for Epidemiology and Biostatistics and the Murdoch Children’s Research Institute Clinical Epidemiology and Biostatistics Unit.
Embedded within multiple Australian clinical trial networks, he develops innovative and efficient randomised studies in areas including Staph bacteraemia treatment, bowel cancer screening, Covid prophylaxes and treatment, and improving outcomes of preterm kids. Informed by his applied work, his main methodological research focus is on the Bayesian decision-theoretic design and analysis clinical trials, primarily on adaptive platform trials and sequential multiple assignment randomised trials
Agenda:
This is a hybrid seminar. Please join us in GP-Z208 if you are on campus, or by zoom.
Details:
| Start Date: | 14/02/2023 [add to calendar] |