I am a third year PhD student at Warwick Medical School. My research is in medical statistics for clinical trials that use adaptive seamless designs. My thesis focusses on developing methods for point and interval estimation specifically for survival outcomes that account for treatment selection bias in trials that use adaptive designs and comparing those to normally distributed or binary outcomes. This project is a collaboration between Warwick Medical School (WMS) and Novartis Pharma AG (Novartis).
Before starting my doctoral studies I worked as a Biostatistician at the Cancer Research UK Clinical Trials Unit, University of Birmingham. Here I worked on implementing Bayesian adaptive designs for many early phase clinical trials in leukemia and lymphoma research as well as performing statistical analyses for late phase clinical trials.
Prior to this, I obtained a 1st class BSc (hons) in Mathematics from the University of Birmingham, followed by an MSc in Medical Statistics from the University of Leicester.
Medical statistics, Survival Analysis, Estimation, Adaptive Designs for Clinical Trials, Bayesian Analysis, Meta-Analysis, Haemato-Oncology
PhD Research Detail
My research is focussed on point and interval estimation for survival outcomes in clinical trials that use adaptive designs.
In late phase clinical trials, adaptive designs are utilised due to their efficiency. For example, they are used to answer phases II and III objectives of drug development using a single seamless phase II/III clinical trial with two stages. Interim analysis of stage 1 data answers the phase-II objectives such as treatment selection and a final confirmatory analysis of both stages 1 and 2 data answers phase-III objectives. Although efficient, data dependent adaptation introduces complexity in hypothesis testing and estimation. My research focuses on estimation in a multi-arm two-stage clinical trial where in stage 1, the apparently most effective experimental treatment and control are selected to be investigated further in stage 2. Using data from both stages of such a trial in estimation introduces selection bias because stage 1 data are used both for estimation and selection. Additional bias is introduced if there is a possibility of early stopping for futility/efficacy. Several methods that account for treatment selection and the possibility for early stopping have been proposed for normally distributed outcomes. However, for time-to-event outcomes, methods of estimation that account for bias in this setting are limited. Survival outcomes are generally used in cancer studies with interest in overall or progression free survival as primary outcomes. Such outcomes have added complexities in estimation due to parametric vs non-parametric test choices, censored observations and correlation between stages 1 and 2 data. The main aim of my PhD is to extend these methods developed for normally distributed outcomes to survival outcomes.
My PhD project is supervised by:
Dr. Peter Kimani (WMS)
Professor Nigel Stallard (WMS)
Dr. Ekkehard Glimm (Novartis)
Professor Frank Bretz (Novartis)
This project is jointly funded by the Medical Research Council and Novartis Pharma AG as part of an industrial case award.