George A Tomlinson

I am a research scientist contributing to the design and planned analyses of clinical research studies. These contributions involve choices of outcomes, timing and numbers of assessments of outcomes, analytic methods and presentation of results. These choices are based on integration of subject-area knowledge for the specific project, specific expertise in biostatistical methods, and broader principles of biostatistics and epidemiology. As a scientist in a hospital-based research institute, my primary focus is the advancement of medical research through the application of sound methods.

I engage in research in several areas of biostatistics and research methods: I contribute to some of the GRADE papers on strength of evidence. With several graduate students, I am involved in research into extensions of joint models for biomarkers and clinical outcomes, methods for Bayesian G-computation in analyses of observational data, and extensions to both multivariate and longitudinal outcomes in network meta-analyses. I often collaborate with colleagues on methodological questions related to the kinds of clinical research in the first thread; recent examples include work on potential issues with interim analyses and early stopping of clinical trials where benefits are not constant over time, intention-to-treat analyses and per-protocol analyses in non-inferiority trials, and comparisons of machine learning and more traditional methods for clinical prediction models.