We’ve recently published a piece regarding analytics and data science during a clinical trial, written by Nina Anderson at inSeption Group. If you are interested in reading the full article and seeing sources, you can click here to see the PDF, but we’ve included a short synopsis here for you to get started.
Anderson points out that the usage of advanced analytics can assist pharmaceutical sponsors with a multitude of challenges, but that advanced data analytics is critically underutilized for a variety of reasons, including concerns of in-house expertise, return on investment, regulatory scrutiny and other hurdles. Through better understanding of high quality data, Anderson states, the industry can move towards more efficient and well-managed clinical trials and better quality results.
The article explains how the need for quick turnaround in conducting clinical trials often hinders the ability to collect and analyze high quality data. Data must be cleaned and monitored, which is a time-consuming process, partly because of an antiquated method of trial leadership that insists on source document verification (SDV).
As Anderson cites in a real life example, source document verification does not necessarily result in the data in that document being accurate. The end result in this example was multiple mis-stratifications across separate databases. Anderson points out that algorithms designed to catch such discrepancies in real time could have flagged the issue sooner.
The article goes on to explain how part of the issue with data contextualization and analysis is that the data engineers who program visualization tools have little understanding of clinical data, and as a result might not use proper data sets for visualizations. Ideally, Anderson writes, “A more effective approach would comprise having a subject matter expert (SME) in clinical operations who also understands programming, including its available tools and technologies.” Unfortunately, this isn’t an easy option for pharma companies, due to the lack of such SMEs or the cost of licensing fees.
To this end, Anderson does describe free and open source software platforms that can produce effective visualizations, with no need for licensing fees. Posit (formerly RStudio) is one example of such software. Once data has been organized for analysis, organizations can now manually view the output before trusting the data and its conclusions.
Anderson sums up the piece by pointing out the benefits of using the underutilized advanced data collection and analytics, including stopping potential fraud. A pharma company looking for approval through a clinical trial can be better served finding ways to leverage those analytics.