Statistical Computing Challenges at FDA
View Presentation View Presentation
*Paul Schuette, FDA 

Keywords: Statistical Computing, Scientific Computing

Statistical computing at the US FDA spans a variety of needs, based largely upon on the requirements of each center. This talk will focus on two statistical computing challenges experienced by most centers: infrastructure together with reproducibility and repeatability. Infrastructure challenges include hardware, software, data issues and personnel. We will highlight some initiatives to deal with these challenges. Repeatability and reproducibility challenges include replicating sponsors’ submitted results, conducting independent analyses, being able to repeat results and findings over time. We recommend some strategies for sponsors and FDA statisticians to begin address these concerns including the adoption of Good Programming Practices (GPP), and appropriate documentary procedures. Finally, we illustrate the potential of an openFDA application which analyzes adverse event report data to change how statistical analyses are conducted and disseminated.