Dr Matthew Smith
Statistician Knowledge Transfer Partnership KTP Associate
United Kingdom
I am a statistician specialising in causal inference and survival analysis. My work focuses on target trial emulation, pharmacoepidemiology and the use of real-world data to answer causal questions.
I am currently a Knowledge Transfer Partnership (KTP) Associate, based jointly at the London School of Hygiene & Tropical Medicine and GSK. The KTP is funded by UKRI Innovate UK, and I work with Professor Ruth Keogh and Professor Jonathan Bartlett at LSHTM alongside colleagues in GSK's Statistics and Data Science Innovation Hub. There I help embed target trial emulation across therapeutic areas including oncology, vaccines, HIV, hepatology, and respiratory medicine. My role covers methodological development, study design and quality review, training, and tools for analysts. The aim is to turn methodological research into routine industry practice, using real-world evidence to inform clinical development, regulatory strategy and trial design.
My methodological interests include:
- target trial emulation and immortal time bias
- double-robust estimation, including targeted maximum likelihood estimation
- net survival and excess hazard models
- competing risks
Before my current role, I was a postdoctoral Research Fellow in Statistics in the Inequalities in Cancer Outcomes Network at LSHTM. I was also co-investigator on the MRC-funded ROBEST project (Better Methods, Better Research award), which developed and promoted double-robust causal inference methods for applied researchers in public health, epidemiology and clinical sciences.
I hold an honorary appointment as Senior Research Fellow in Statistics at the Institute of Health Informatics, University College London, where I co-supervise PhD students.
I completed my PhD at LSHTM in 2020, on socioeconomic inequalities in survival among patients with non-Hodgkin lymphoma. I also hold an MSc in Medical Statistics from LSHTM and a BSc in Mathematics.
Affiliations
Teaching
In my current role, I develop and deliver training in target trial emulation for statisticians, epidemiologists and data scientists at GSK. I also give invited seminars on the topic to academic audiences.
I currently supervise two PhD students:
- Aasiyah Rashan (UCL) is researching the relationship between mechanical ventilation and health outcomes in intensive care.
- Thomas Gachie (University of Sheffield, external supervisor) is researching double-robust machine learning methods and their application to evaluating policies that affect mental health outcomes.
I have contributed to postgraduate teaching on the MSc Medical Statistics at LSHTM. This included co-organising the Causal Inference and Missing Data module and leading practical sessions for modules including Generalised Linear Models and Analysis of Hierarchical and Other Dependent Data. I have also supervised MSc summer projects and been a personal tutor to MSc students.
Research
My research focuses on developing, evaluating and applying causal inference methods for observational data, especially target trial emulation. I am particularly interested in using real-world data to answer clinical questions that inform treatment decisions, trial design and regulatory strategy. I also study the design choices that decide whether these analyses are credible, including whether time zero is aligned and how immortal time bias is avoided.
I also work on double-robust estimation, particularly targeted maximum likelihood estimation (TMLE). I evaluate how well these estimators perform and extend their use to longitudinal and survival data. I have published accessible introductions to causal inference for applied researchers, contributed practical guidance on implementing TMLE and target trial emulation with real-world data, and co-developed the eltmle command for Stata.
Selected Publications
