Chief, ImmunoTechnology Section Vaccine Research Center, NIAID Co-Sponsored with the CFAR Seminar Series
Immunological and Virological Mechanisms of Vaccine-Induced Protection From SIV Acquisition
Austrian Auditorium CRB
"Targeting quiescent cancer cells"
"Immunological and Virological Mechanisms of Vaccine-Induced Protection From SIV Acquisition"
Virology Seminar - Speaker: Alexander Gill - 209 Johnson Pavilion
"The Impact of Unmeasured Confounding in Observational Studies: Analysis of Bias"
Yun Li, PhD Research Assistant Professor Department of Biostatistics School of Public Health University of Michigan
Abstract: Unmeasured confounding almost always exists in observational studies which can lead to bias in effect estimation and undermine the validity of the results. Clustering within groups occurs frequently in these studies. This research intends to understand the impact of unmeasured confounding on effect estimation with focus on two methods: instrumental variable analyses and linear mixed models. Instrumental variable analyses have often been used to control for unmeasured confounding and groups of patients (e.g., clinical centers, physician practices, or service areas) have often been proposed as instruments. Linear mixed models have frequently been employed where patients within each cluster are more similar to each other than those in other clusters. That is, patient heterogeneity exists which often arise from unmeasured factors attributable to patient differences between clusters. In this research, we introduce the framework of between- and within-cluster unmeasured confounding to assess assumptions required for group-based instrumental variable analyses and linear mixed models. We then derive closed-form expressions of asymptotic bias of estimators from both methods in the presence of within- and/or between-cluster level unmeasured confounding when the number of independent units, i.e., groups, goes to infinity. Simulations demonstrate that the asymptotic bias formulae approximate bias in finite samples quite well under a range of parameter settings for both methods, particularly when the number of groups is moderate to large. We also show that linear mixed models alleviate the amount of bias when unmeasured between-cluster level confounding exists while instrumental variable analyses can reduce the amount of bias when unmeasured within-cluster level confounding exists. When unmeasured confounding exists at both patient and group levels, whether instrumental variable analyses are advantageous in reducing bias over linear mixed models depends on the extent of unmeasured confounding at the within-cluster levels relative to that at the between-cluster level. We provide practical guidance in the use of group-based instruments in IV analyses and linear mixed model. We illustrate the method to estimate the effect of erythropoiesis-stimulating agents on hemoglobin levels.
Greetings from the Penn Biotech Group!
We are a Wharton-led, Penn-wide club focused on creating opportunities for its members to get real hands-on consulting experience on life-sciences related projects! Members work on cross-functional and interdisciplinary teams that simulate real world project groups and include students from Wharton, School of Arts and Sciences, School of Medicine, School of Engineering and Penn Law. Each year we source around 40 projects from Fortune 500 companies and several startups dealing with immediate business issues involving market strategy, business development, marketing, operations and finance.
This is a great experience to put on your resume and give you an edge in a growingly competitive job market!
Please see below for more information or visit www.pbgconsulting.org . If you would like to continue receiving information from us, please sign up to our mailing list by clicking here.
1. PBG Kick-off Event- February 4th, Location TBD
Happy New Year! PBG is starting off 2014 with a kick-off event for all members and we hope you can make it! Members new and old are invited to come and learn more about our Wharton-led, Penn-wide club and get a preview for upcoming spring projects.
We hope to see you there!
To make sure that you receive information about the time and place of our kickoff, be sure to sign up for our mailing list!
The New Era for U.S. Global Health Diplomacy A Lecture by Jack Chow (C’82), Former U.S. Ambassador and Deputy Assistant Secretary of State for Health and Science
Date: Tuesday, February 4, 2014 Time: 4:30 PM Location: Jon M. Huntsman Hall, Room 340
Professor Chow will discuss the progress of global health diplomacy, the forward vision of such efforts, the context of such diplomacy within the evolution of foreign policy, the implications for the US, as well as the outlook for private sector efforts.
Currently serving as distinguished service professor at Carnegie Mellon University, Jack Chow has been involved with health public policy both domestically and abroad. Chow was the first U.S. diplomat of ambassador rank appointed to a public health mission. He has worked with esteemed politicians including U.S. Secretary of State Colin Powell, Representative Silvio O. Conte, and U.S. Senator Arlen Specter.