I am internationally leading in methods and applications in the fields of clinical prediction modelling and notably, its intersection with causal inference. I have extensive experience in teaching and learning, and am Programme Director of the MSc Health Data Science. 

My approach is to start with genuine clinical problems where additional decision support (through prediction models) is warranted, particularly cases where standard methods don’t suffice. Because of this problem-first approach my work is implemented in healthcare systems. My focuses are:

1. The role of causal inference in prediction, to enable prediction under intervention. This is directly beneficial to decision makers - standard prediction models consider correlation only, so do not allow ‘what-if’ (prediction under intervention) questions like ‘what if I stopped smoking’ to be answered - but these are exactly the questions decision-makers have. I am developing methods to answer exactly these questions - e.g., through the Causality in Healthcare AI (CHAI) hub UKRI funding, but also following this through to direct application - e.g., through the CHARIOT project where we are building a prediction under intervention model for cardiovascular disease prevention. I also study the role of causal inference in improving generalisability and transportability of models, and to elucidate fairness and bias issues.
2. Dealing with missing data when making predictions, and the inverse problem of adaptive observation - determining what should be measured, and when, at an individual level.
3. Ensuring models remain up-to-date across space and time using dynamic modelling and updating.

These three broad areas are motivated by improving the pathway to implementation for prediction models in healthcare. Clinical application areas include primary prevention of long term conditions (especially CVD), and treatment in cancer. I work mostly with routinely collected electronic health record data.

I have wide ranging experience of course leadership and curriculum development in the undergraduate, postgraduate, and CPD contexts, having developed and delivered courses across all three. I have taught in two universities, both to ‘specialist’ students (e.g. mathematicians and MSc students) and to ‘service’ students (e.g. skills courses for the NHS). I use a range of innovative teaching approaches, including blended learning.

I am Programme Director for the MSc Health Data Science.

Royal Statistical Society, CStat.

SFHEA

Statistics, Epidemiology, Data Science

PhD, Statistics, Lancaster University (2010)

MMORSE (Masters in Maths, Operational Research, Statistics and Economics), University of Warwick (2006)