Seminar Series: Dr. Guangyong (GY) Zou

Distribution-free methods for the design and analysis of randomized controlled trials

Guangyong Zou

Professor
Department of Epidemiology and Biostatistics
Schulich School of Medicine & Dentistry
Western University

Scientist
Robarts Research Institute
Western University

Biostatistician Director
Alimentiv Inc.

Short Biography:
Guangyong (GY)  Zou holds a Bachelor in Agronomy, a MSc in Soil Science, a MSc in Statistics, a PhD in Plant Science, and a PhD in Biostatistics. He is currently a Full Professor of Biostatistics in the Department of Epidemiology and Biostatistics, a Scientist in Robarts Research Institute at Western University, and Director of Biostatistics at Alimentiv Inc, a global contract research organization specializing in inflammatory bowel disease (IBD) clinical trials. He has over 25 years of experience as a biostatistician for clinical trials in a wide range of diseases. His research focuses on developing novel statistical methods for biomedical research. Many of these methods have become very popular in medical research and have been highly cited. For example, a single authored paper by Dr Zou in 2004 is No. 1 on the list of the 100 most cited papers in the 100 years of the American Journal of Epidemiology (https://academic.oup.com/aje/pages/100-years).

Abstract:
Randomized controlled trials (RCTs) have been established as a gold standard for evaluating effects of interventions. Design and analysis of RCTs usually rely crucially on distributional assumptions for the outcome variables. Yet making appropriate assumptions is not an easy task. Without making assumptions, we have successfully developed a general approach for estimating treatment effects and covariate adjustments in RCTs. We quantify treatment effect by the win probability that a participant in the treatment group has a better outcome than (or wins over) a control participant. The win probability can be defined for binary, categorical, counts, and continuous outcomes. Our approach entails two steps. We first use ranks to transfer each observation into a win fraction to reflect the fraction of times that the observation wins over all observations in the comparison group, followed by applying linear regression to the win fractions for treatment effect estimation and sample size planning of RCTs. In this talk, I will demonstrate how to take this approach to the analysis and planning RCTs using real examples.

Area of Research:
Biostatistics; Cluster randomization trials; Confidence interval estimation; Nonparametric methods; Sample size estimation

Date: Friday, September 19
Time: 1:30 pm - 2:30 pm
Location: PHFM 3015 (Western Centre for Public Health and Family Medicine) or Zoom (request link by email    epibio@uwo.ca)