Selected Talks
“Bayesian Logistic Regression for Medical Claims Data”, Ivan Zorych, Department of Statistics, Columbia University : bayesian_logistic.pdf
at THE 33rdANNUAL MIDWEST BIOPHARMACEUTICAL STATISTICS WORKSHOP
MAY 24 – 26, 2010 • BALL STATE UNIVERSITY ALUMNI CENTER, MUNCIE, INDIANA
JSM 2013: Ivan_Zorych_JSM_2013_draft
| Title: | Large-Scale Penalized Regression for Propensity Score Estimation in Observational Health Care Data |
| Author(s): | Ivan Zorych*+ and Patrick Ryan and David Madigan |
| Companies: | Department of Statistics, Columbia University and Jenssen Research and Development LLC and Columbia University |
| Keywords: | propensity ; balance ; observational data ; regression ; penalized ; OMOP |
| Abstract: | Propensity score adjustment is a useful tool to deal with confounding in observational data. There are different views on building the propensity score models. Here we compare two approaches. One is based on an empirical covariate selection procedure and another one relies on a Bayesian logistic regression and utilizes all covariates, sometimes many thousands, that are available from an observational database. We investigate balancing properties of both procedures using a set of drug-outcome pairs on several observational databases provided by the Observational Medical Outcome Partnership (omop.fnih.org). |
JSM 2012: Zorych_Ivan_JSM_2012
| Title: | Active Surveillance on Medical Observational Databases |
| Author(s): | Ivan Zorych*+ and Patrick Ryan and Martijn Schuemie and David Madigan |
| Companies: | Columbia University and Johnson & Johnson and Erasmus Medical Center and Columbia University |
| Keywords: | post-marketing safety ; surveillance ; observational data ; adverse reactions ; longitudinal medical data |
| Abstract: | Post-marketing drug safety surveillance can be enhanced by utilizing observational health databases. Application of statistical methods to such longitudinal medical data is a novel and challenging task. We consider a set of statistical methods and their operational characteristics on several real and simulated databases. Evaluations were carried out on sets of selected drugs and selected outcomes of interest. |
JSM 2011: JSM Aug 2-2011 Ivan Zorych
| Title: | Bayesian Logistic Regression for Medical Claims Data Using CPU and GPU |
| Author(s): | Ivan Zorych*+ and Patrick Ryan and David Madigan |
| Companies: | Columbia University and Johnson & Johnson Pharmaceutical R&D, LLC and Columbia University |
| Address: | Room 1005 SSW, New York, NY, 10027, |
| Keywords: | Bayesian regression ; coordinate descent ; logistic ; lasso ; GPU ; CUDA |
| Abstract: | Bayesian logistic regression for medical claims data is a novel statistical approach that possesses the advantages of regression analysis such as being resistant to confounding by co-medication and adjusting for masking effect. Mapping medical claims data into the form appropriate for the regression analysis is an essential step. We consider several ways to represent claims data in the form appropriate for regression. We investigate Bayesian regression models with either Normal or Laplace priors. Analysis of each condition of interest requires fitting a separate regression model. Fitting such a model is a challenging computational task because each dataset contains millions of reports and thousands of covariates. Our numerical approach to logistic regression relies on coordinate descent algorithm. We consider two implementations of this algorithm, traditional central processing unit, CPU, version and a parallel implementation that utilizes graphics processing unit, GPU. The performance of our approach will be illustrated on the simulated and real data. |