PAGE. Abstracts of the Annual Meeting of the Population Approach Group in Europe.
PAGE 28 (2019) Abstr 8913 [www.page-meeting.org/?abstract=8913]
Oral: Methodology - New Tools
Alison Margolskee (1), Fariba Khanshan (1), Andrew Stein (1), Yu-Yun Ho (2) and Michael Looby (3)
(1) Novartis Institutes for Biomedical Research, Cambridge, MA, USA; (2) Novartis Pharmaceuticals, East Hanover, NJ, USA; (3) Novartis Pharma AG, Basel, Switzerland
Introduction: As pharmacometricians, we sometimes jump into complex modeling before thoroughly exploring our data. This can happen due to tight timelines, lack of ready-to-use graphic tools or enthusiasm for complex models. Exploratory plots can help to uncover useful insights in the data and identify aspects to be explored further through modeling or in future studies. Exploratory plots can even quickly answer questions without the need of a complex model, improving our efficiency and providing timely impact on project strategy. The Exploratory Graphics (xGx) tool is an open-source R-based tool, freely available on GitHub . Intuitively organized by datatype and driven by analysis questions, the tool aims to encourage a question-based approach to data exploration focusing on the key questions relevant to dose-exposure-response analyses.
Methods: PK (single and multiple ascending dose), and PD (continuous, time-to-event, categorical, count, and ordinal) data were simulated and formatted according to a typical PKPD modeling dataset format. Lists of key questions relevant to dose-exposure-response exploration were compiled, and exploratory plots were generated to answer each question. The graphs were created following good graphics principles to ensure quality and consistency in our graphical communications .
Results: Examples of the key analysis questions include:
For each datatype in the simulated dataset, plots were generated to answer these key questions. The plots along with the codes to produce them were compiled into a user friendly interface. The tool is intuitively organized by datatype and driven by the analysis questions. Since the graphs were generated based on a typical modeling dataset format and hosted online, they can be easily accessed and applied to new projects.
Conclusion: Exploratory plots were generated, built around typical key questions particularly relevant to dose-exposure-response exploration and compiled into a user friendly interface. The Exploratory Graphics (xGx) tool can help underscore the role of purposeful data exploration for quantitative scientists. Through a question-based approach, xGx helps uncover useful insights that can be revealed without complex modeling and identify aspects of the data that may be explored further.