Research
Research
For a full list of my publications, please see my Google scholar.
I am interested in methodological developments for Network Meta-Analysis (NMA), which is a method used to estimate the relative efficacy or safety of multiple treatments by synthesizing clinical trial data. I have worked on a few different topics in NMA. During my PhD, I was also involved in the development of BUGSnet, an R package that facilitates the conduct and reporting of Bayesian NMA.
Component Network Meta-Analysis (CNMA) is an extension of NMA that is used to model multi-component treatments. I identified an important difference in the specification of the common frequentist and Bayesian implementations of CNMA, which causes the Bayesian approach to be less flexible, and I proposed a robust Bayesian alternative. I won the ASA Health Policy Statistics Student Paper Award for this work in 2022. Read the pre-print or the paper (not open access).
The results of an NMA are often summarised using a treatment hierarchy, that is, a list of treatments ordered from most to least preferred. Hierarchies can be misleading as 1) there is uncertainty in the estimated ranks, and 2) different ranks do not necessarily reflect statistically or clinically meaningful differences between treatments. I had the chance to work on solutions to both 1) and 2) on a research visit to the Knowledge Discovery and Synthesis group at the University of Freiburg in Winter 2024 thanks to a NSERC Michael Smith Foreign Study Supplement.
1) Precision of Treatment Hierarchy (POTH): A simple metric for quantifying uncertainty in treatment hierarchies [Paper, R package]
2) Incorporating minimally important differences in treatment hierarchies in Bayesian NMA [Paper]
Standard NMA models provide estimates of the average relative effects in the population represented by the studies used to conduct the NMA. Methodological advances enable personalized prediction of relative effects for specific covariate profiles. In my post-doc, I am developing doubly-robust Bayesian approaches to estimating Individualized Treatment Rules (ITRs) using NMA.
During my PhD, I was involved in interdisciplinary research motivated by challenges in the quantification of methane emissions from the oil and gas industry.
Bayesian approach to quantifying uncertainty in measurements of methane emissions from storage and production facilities [Paper]
Interpretable and computationally feasible estimation of provincial methane emissions using aerial survey data [Paper, R package]