Frontiers in Biostatistics Seminar

The Frontiers in Biostatistics seminar features speakers whose work in biostatistical methodology has relevance in oncology research applications. The series aims to highlight topics of broad interest to the Department, and focuses on inferential approaches to analysis, clinical trials designs, biomarker evaluation, and other topics in translational biostatistics.

Want to be on the mailing list? Click here to sign up

Coming Up

December 10, 2019
1:00PM


Jessica Franklin
Assistant Professor of Medicine at Harvard Medical School
Biostatistician, Division of Pharmacoepidemiology and Pharmacoeconomics, Brigham and Women’s Hospital

Comparing Nonrandomized Studies with Randomized Trials to Evaluate Causal Inference Methods in Real-World Data

Randomized controlled trials (RCTs) remain the gold standard for establishing the causal relationship between medications and health outcomes. However, for some clinical questions RCTs may be infeasible, unethical, costly, or generalizable to only a very narrow population. In these cases, observational studies from routinely collected “real-world” health data (RWD) are crucial for supplementing the evidence from RCTs. To explore the validity of observational studies, there have been several efforts to compare the results of observational studies to those of RCTs investigating the same clinical question. Empirical evaluations such as these could improve understanding of the performance of methods as applied to RWD, as well as understanding of when nonrandomized studies are likely to fail and why. However, nearly all of these comparisons have relied on retrospective reviews of the literature, which provide flawed measures of the validity of real-world studies as a whole and provide little insight into which clinical questions can be answered with confidence and with which methods. In this presentation, I will discuss the flaws of past research and present a potential path forward towards building an empirical evidence base for the use of causal inference designs and methods in RWD, which we are pursuing in the RCT DUPLICATE initiative.


Past Seminars

September 24, 2019
12:00PM

Nikesh Kotecha, PhD
Vice President, Informatics
Parker Institute for Cancer Immunology

Systems Immunology in IO: A view from the Parker Institute

Abstract: The introduction of immunotherapies has revolutionized the treatment of cancer and ushered in a corresponding explosion of research into cancer, the immune system, and their interaction. This talk will introduce the Parker Institute for Cancer Immunotherapy, its mission to accelerate the development of breakthrough immune therapies, highlight the informatics opportunities and challenges presented in this space and our approaches to address them.


October 15, 2019
1:00PM

Miguel Hernan
Kolokotrones Professor of Biostatistics and Epidemiology
Department of Epidemiology
Department of Biostatistics
Harvard School of Public Health

Observational studies – How do we learn what works?

 Randomized experiments are the preferred method to quantify causal effects. When randomized experiments are not feasible or available, causal effects are often estimated from non-experimental or observational databases. Therefore, causal inference from observational databases can be viewed as an attempt to emulate a hypothetical randomized experiment—the target trial—that would quantify the causal effect of interest. This talk outlines a general algorithm for causal inference using observational databases that makes the target trial explicit. This causal framework channels counterfactual theory for comparing the effects of sustained treatment strategies, organizes analytic approaches, provides a structured process for the criticism of observational analyses, and helps avoid common methodologic pitfalls.