Speaking Different Languages – Clinical Statistical Modelling in a World with Choice
Clinical Statistical Reporting in a Multilingual World (CSRMLW) is a collaborative working group project between the R Consortium and PHUSE aimed at empowering statisticians to make informed decisions about the implementation of statistical analyses when multiple languages yield different results. In addition to the possibility of software errors, the source of these discrepancies may be due to differences in modeling assumptions, default parameters, or the choice of estimator or algorithm to compute a particular quantity. Resolving some differences may require adjudication by experts, while others may not be critical. The primary objective of CSRMLW is to provide guidance on the types of questions an analyst should ask to identify the fundamental sources of discrepant results. By doing so, the analyst is empowered to make informed decisions on how to perform the analyses of interest most appropriately.
Applications of the framework under development within the project initially focus on R vs SAS differences in four classes of statistical models: linear model, mixed model, survival analyses, and Cochran–Mantel–Haenszel analyses (CMH). During this session, participants should expect a project update from the project leads, interactive breakouts sessions devoted to each of the four classes of models, and guidance on how they might contribute to the project in the future.
Schedule – 90 mins:
- Background/Intro – 15 mins
- 4 Use Cases – Each one 15 mins (5 min brief presentation, 10 min group discussion)
- Closing/Next Steps – 15 mins
Michael Rimler is the Head of Technical Excellence and Innovation at GlaxoSmithKline, a group that aims to accelerate the programming function’s path to industry pre-eminence through innovation and facilitating excellence in data engineering, data analysis, and sharing of insights.
In addition to leading innovation activities within Clinical Programming, Michael serves as the primary business lead for the integration of R into GSK’s clinical reporting process. In this role, he oversees activities driving the use of R for independent QC, developing standard reporting tools in R, and exploring opportunities for using R for the generation of trial results.
Externally, Michael is a sub-team lead for TransCelerate’s Modernization of Statistical Analytics project, co-chair of the 2022 PHUSE US Connect, and the co-lead of the PHUSE working group on multilingual clinical reporting which brings us here today.
Mike Stackhouse is a 2020 UC Berkeley School of Information Master of Information and Data Science (MIDS) program graduate, where he worked on projects involving computer vision, natural language processing, cluster computing, and deep learning. Previously, Michael was a senior manager of statistical programming at Covance, where he led U.S. innovation activities for the FSP department. Michael and his team at Atorus have been actively developing and releasing open-source R packages, such as pharmaRTF and Tplyr.
Mike is a Lead of PHUSE’s Data Visualisation & Open Source Technology Working Group, home to the Clinical Statistical Reporting in a Multilingual World project, which he co-leads alongside Michael Rimler.
The R Adoption Series
This is a new series of webinars focused on the adoption of R. Each session will include a case study and often include panels or discussions to enable those starting their journey to ask questions.
R Consortium will keep this page updated with information on future webinars in the R Adoption series. If there is some information that you are looking for specifically and you don’t see it here, feel free to email us at firstname.lastname@example.org.