Keywords: Exploratory analysis, data science, interactive visualizations, R, Shiny, dashboards, markdown
Traditional techniques such as pre-planned analyses and static table/figure/listings that have been suited to the confirmatory paradigm are often used as the default approach in clinical research. The data generated in modern drug development are growing rapidly in quantity and complexity. More non-conventional endpoints and modern digital technologies are being used to collect data throughout various stages of drug development, as drug developers are tackling complex and challenging diseases. In this context, the traditional approach often fails to do justice to the data collected and in many cases does not address questions of interest. We adopt a streamlined data science approach (tidy – interactively transform, visualization and model – communication), empowered by modern tools such as R/Shiny, that is able to turn complex raw data into understanding, insight and knowledge in an efficient and effective fashion. We will illustrate this approach using a case study of exploratory analysis, using data from a natural history study of patients with retinitis pigmentosa and featuring interactive visualizations with Shiny and Spotfire, interactive reporting using R/Markdown, and dashboards, etc.