Professor of Medical Statistics and Director of CSM
- Director of UK EQUATOR Centre
Gary's research interests are primarily focused on methodological aspects surrounding the development and validation multivariable prediction (prognostic) models (design and analysis) and he has published extensively in this area. He is particularly interested in the sample size considerations and the role of big data in developing and evaluating prediction models. He is also interested in the systematic review and appraisal of prognostic studies and developed the CHARMS Checklist for conducting systematic reviews of prediction modelling studies. Gary is the principal investigator of a five-year Cancer Research UK Programme grant (2019 to 2024) to improve statistical methodology around studies of prognosis (with a focus on machine learning and artificial intelligence) and initiatives to improve peer review and study reporting.
Along with Doug Altman, Karel Moons and Hans Reitsma, UMC Utrecht, the Netherlands, he led an international collaboration to produce the TRIPOD consensus guidance on issues to report when developing or validating (prognostic and diagnostic) prediction models. This guidance is currently being extended for prediction models developed using 'big data' or individual participant data from multiple studies. Gary and Karel Moons are leading an initiative to develop similar guidance for studies using artificial intelligence and machine learning (TRIPOD-AI). Gary is also involved in other guidance for reporting artificial intelligence/machine learning studies including CONSORT-AI/SPIRIT-AI (for reporting AI intervention studies), STARD-AI (for reporting AI based diagnostic test accuracy [DTA] studies), and DECIDE-AI (bridging the development-implementation gap). He is also involved in developed risk of bias tools for machine learning diagnostic test accuracy studies (QUADAS-AI) and prediction model studies (PROBAST-AI)
Between 2012 and 2019, Gary was involved in the development of PROBAST, a risk of bias tool to evaluate prediction model studies (using regression or machine learning/artificial intelligence methods). Two papers were published in the Annals of Internal Medicine describing the PROBAST checklist and and accompanying Explanation & Elaboration paper.
Gary is also a member of other reporting guidelines including the GATHER working group, which developed guidance for reporting global health estimates, recently published in the Lancet and PLoS Medicine, and the AGReMA statement for reporting mediation analyses, published in JAMA. He is also a steering group member of the international STRATOS Initiative, which aims to provide accessible and accurate guidance in the design and analysis of observational studies.
Gary has more than 300 peer-reviewed articles, editorials and commentaries, including over 40 research articles in the ‘Big 6’ general medical journals, and is the first or senior author of more than 100 articles. He is principal investigator and co-applicant on a number of methodological and clinical research grants (>£16 million).
He is also a Statistical Editor (‘hanging committee’) for the British Medical Journal, an Associate Editor for Statistics in Medicine, Research Integrity and Peer Review, and on the Editorial Board of the Journal of Clinical Oncology. In 2016 Gary co-founded and is Editor-in-Chief of the journal Diagnostic & Prognostic Research. Gary served 3-year term as board member of the NIHR HTA Commissioning Board between 2017 and 2020, and currently serves on the NIHR Doctoral Fellowship selection panel.
Gary teaches the personalised / stratified medicine course on the Masters of Science in Public Health: Comparative Effectiveness Research at the University Paris Descartes, France. He also teaches on the
- EQUATOR-OUCAGS course 'An introduction to Medical Research'
- EQUATOR Publication School
- Statistical Methods for Risk Prediction & Prognostic Models (Keele University).
Gary also supervises DPhil/PhD/MD students and manages a team of statisticians and researchers working on applied statistical methodology (primarily clinical prediction and prognosis). Gary is also a member of the NDORMS Graduate Studies Committee.