Use of R is a hot topic among statisticians and programmers in the pharmaceutical industry. At the recent PhUSE conference in Barcelona there was a clear uplift in interest in the language and a number of sessions explored introductory principles and examples of how R can be used in practice. Cytel's Namrata Deshpande presented on the use of R beyond Statistics through a case study of the development of a user friendly tool deploying non-statistical packages in R to enable clinical decision making. The talk won first prize in the Trends and Technology track at the PhUSE conference. In this blog, we'll discuss some of the aspects presented and share Namrata's slides for download.
As previously discussed on the blog, R is on the rise with 76% of analytical professionals reported to be using the tool. Even in the pharmaceutical industry, where SAS has traditionally been more widely used, uptake of R is increasing. The reasons for R's popularity are clear. It is up to date, with statistical packages for the latest developments in the field becoming available very quickly. It has strong graphical capabilities, and is supported by a range of operating systems. Finally, it has no licensing costs, making it an attractive option. However, strong programming skills and knowledge are usually required to deploy its capabilities. In the case study presented by Deshpande, she illustrates how non-statistical R packages were applied and leveraged to create a user friendly solution which harnessed the power of R for a group of clinician users who didn't have programming knowledge.
The Case Study
A group of clinicians within one of Cytel's clients frequently needed to conduct a routine exploratory analysis. The process was time consuming as the analysis had to be conducted manually, generally using functions in Excel. An automated approach was required to save time and reduce errors.
Any proposed solution needed to take into account the fact that the users would be non-statisticians.The tool was also intended for widespread use throughout the team and among 3rd party vendors, making the use of licensed software costly and impractical. With these considerations in mind, the Cytel team decided to build a solution using functions and packages freely available in R.
The Cytel team developed a function in R to accept the data, and analyze it. Then,they created an intuitive Graphical User Interface (GUI) using the package ‘gWidgets’. This allowed the non-statistician user to easily browse and select the data file and modify their options for the analysis.The report creation had also previously been a time consuming process for the clinician team. Using another R package ReporteRs, functionaility was created to allow the users to generate a report at the end of analysis, delivering all the required outputs and graphs in a clear and consistent format.
Non Statistical R Packages Used
GWidgets is anR package authored by John Verzani. The ' Widget' of the name refers to the ability of the package to create interactive controls for a Graphical User Interface ( GUI). For instance, the package provides the capability to add labels, drop down boxes, buttons and 'handler' functions to manage interdependent actions on the GUI. It also provides many more functional and cosmetic capabilities to simplify tool development.
ReporteRs is a package authored by David Gohel et al. It enables the user to generate well presented reports from R in PowerPoint, Word, and .html formats.
Great improvements in productivity were delivered using the developed tool, with an average time for analysis taking less than 30 minutes versus 16 hours using the previous methods.
To download Namrata's slides from PhUSE click below.
Liked this blog? Click the button below to sign up to receive the latest news, trends and best practices in biostatistics, programming and data management direct to your inbox.