Statistics Colloquium: Survival Data Analysis Using Average Hazard with Survival Weight
Presented by: Hajime Uno, Associate Professor, Department of Biostatistics, Harvard T.H. Chan School of Public Health
Date: Wednesday, September 17, 2025, 4:00 PM, AUST 434
Meeting Link: Link
Coffee will be served at 3:30 in the Noether Lounge (AUST 326)
Abstract:
Hazard ratios from Cox proportional hazards models have been routinely used to quantify intervention effects on time-to-event outcomes in comparative studies. However, the well-recognized limitations of this approach have prompted interest in alternative analytical frameworks. The average hazard with survival weight (AH) is a summary measure of the event-time distribution, interpreted as a person-time incidence rate and robust to nuisance random censoring. While AH itself is not a between-group contrast measure, it can be used to construct intervention effect measures that are more interpretable and robust. In this talk, we present the AH-based framework for survival data analysis, including applications to two-sample comparisons in randomized trials, regression analysis, stratified analyses, and the evaluation of long-term treatment effects. By leveraging AH within comparative analyses, researchers can obtain summary measures that better capture and communicate the magnitude of treatment effects.
Bio:
Hajime Uno, PhD is an Associate Professor of Medicine at Harvard Medical School and an Associate Professor in the Department of Biostatistics at the Harvard T.H. Chan School of Public Health. He serves as Principal Biostatistician and Director of the Statistical Programming Core in the Division of Population Sciences at the Dana-Farber Cancer Institute. Dr. Uno received his PhD in Biostatistics in Japan and completed postdoctoral training at the Harvard School of Public Health under the mentorship of Professor L.J. Wei. Since then, he has worked extensively on methodological and clinical research projects. One of his notable methodological contributions is a modification of the concordance index (C-index) for evaluating risk prediction models with censored time-to-event data. His 2011 Statistics in Medicine paper describing this method was ranked among the ten most-cited articles in the journal over 2012–2013, and the method—commonly known as “Uno’s C”—is implemented in the SAS/PHREG procedure. As of July 2025, the paper has been cited more than 1,500 times. His current research interests focus on improving the practice of survival analysis in clinical research and promoting alternative approaches that enhance the quality of informed treatment decisions.