All Times EDT
Keywords: R, Python, Airflow, Clinical Trials
Clinical trials take almost twice as long as planned to recruit patients (Tufts, 2013). During clinical development (first in human to registration), the recruitment period for clinical trials represents the largest opportunity for reducing the development time. Beyond time reductions, the ability to accurately forecast clinical trial timelines has many opportunities to increase efficiency in areas of resource and drug supply planning. Our Data Science team is tasked with helping optimize the operationalization of clinical trials.
Leveraging both internal and external data on the performance of clinical trials, we build applications to help in the planning and monitoring of clinical trials. We focus on creating automatic workflows of data ingestion and curation, analytics, and deliver a novel user experience to help bring insights that drive better planning and decision-making.
The presentation will focus on applications we developed to optimize the geographic footprint of clinical trials to have a high confidence of meeting timelines while minimizing costs. We’ll discuss the software and technologies we leverage to deliver insights from raw data. Specifically, there will be more detailed discussions of
• automating the data pipeline • development of machine learning/deep learning algorithms • delivering analytics through restful web services
The goal of the presentation is to provide an overview of a production analytics pipeline leveraging R, Python, and Apache Airflow and to provide an understanding of the design decisions.