THIS STORY IS AWAITING CUSTOMER APPROVAL. CONSIDER IT COMPANY CONFIDENTIAL. DO NOT DISCLOSE.
In rare disease biotechnology, the fusion of science and statistics drives innovation at scale
Statistical approaches to experimentation and the automation of data workflows have increased institutional capacity at CSL Behring – an organization dedicated to delivering maximum value in under-resourced therapeutic areas.
_____________________________________________________________________________________
Challenge: In biotechnology labs around the world, a reliance on tools like Excel has for years limited scientists’ ability to collect and utilize large data sets. Furthermore, single factor approaches to experimentation have traditionally constrained capacity in R&D and risked creating a culture of firefighting at the commercialization stage in any new product’s life cycle.
Solution: Recognizing that to go beyond the “digital transformation” buzzwords they would need to build capacity in the statistical space, CSL Behring’s Plasma Product Development (PPD) department brought on a team of statisticians and data scientists to guide the cultural shift and develop tools for driving its adoption. Statistical software platform JMP® has since played a key role as a sophisticated yet approachable tool used by experts and novices alike. “It’s about managing our data analysis capabilities and innovation at scale in a meaningful way,” explains Senior Statistical Expert and Data Manager Peter Vogel, PhD.
Results: A two-pronged approach blessed by PPD leadership involves the upskilling of scientists around statistical methods such as design of experiments (DoE) as well as the development of globally accessible and scalable applications by Vogel and his colleagues – both of which can be done in JMP. Though only in place as CSL Behring’s tool of choice for less than three years, he says, the tool has already dramatically increased organizational capacity in R&D. “Having the ability to do innovation at scale – that is really unique,” he adds. “JMP enables us to process a lot of data in a reasonable timeframe with very customized and extendable options.”
_____________________________________________________________________________________
Biotherapeutics leader CSL Behring, a member of the CSL Group of companies, develops cutting-edge therapies in immunology, hematology, cardiology, and transplant procedures. With 25,000 employees globally – many of whom are scientists specifically dedicated to advancing scientific knowledge in fields where patient needs remain unmet – the company’s portfolio includes major breakthroughs in treating hemophilia, hereditary angioedema, and primary immune deficiencies, among others.
Facing new growth opportunities in the age of digitalization, CSL Behring’s leadership recognized that digital transformation, including a shift in the underlying culture, would be essential to maintain the pace of innovation. The result was a capacity-building project that empowered R&D scientists to adopt statistical best practices while also supporting their efforts through a nimble team of statistical experts tasked with deploying scripted applications across the enterprise.
When Peter Vogel, PhD, was brought on as Senior Statistical Expert and Data Manager at CSL Behring’s Marburg, Germany site, he quickly identified the need to support the Plasma Product Development (PPD) organization through customized add-ins and the creation of a statistical upskilling program for scientists across R&D.
In the early days of his tenure, he recalls, most experimental data resided in Excel spreadsheets. As these spreadsheets often consisted of unstructured data, the system did not allow for easy processing and analysis by scientists and lab technicians. “The typical question of colleagues that came to me was: I have this set of data in Excel, but how do I analyze it efficiently? How do I use it to generate meaningful insights?” Vogel recalls. “They were unable to produce even low-level insights due to basic data processing hurdles inherent in the existing systems.” Quite often, he explains, the solution was to transfer their data from Excel to JMP® statistical discovery software.
JMP helps boost scientists’ data agility
“We recognized that, to process the increasingly larger quantities of instrument data needed for meaningful scientific innovation, we would need customized solutions that empower scientists – not limit them,” Vogel says. Consequently, the R&D organization adopted JMP as the unified data analysis platform for domain and statistical experts alike.
To enable new JMP users across R&D to quickly familiarize themselves with the tool, Vogel and his colleagues created a JMP Basics class that is now available on-demand. In addition, they host regular live sessions to answer questions, showcase new developments and add-ins, and exchange knowledge about best practices. In their effort to increase the data literacy across R&D, they regularly also offer a 20-hour Design of Experiments (DOE) class based on JMP that empowers scientists to construct efficient statistical designs for the typically complex scientific questions they face day-to-day as well as to analyze and interpret results in the respective context.
Now, in conversation with JMP users across R&D, Vogel says, “scientists often highlight how JMP is an easily accessible toolbox and that it often provides them with a better scientific understanding of their data. Data visualization and analysis are all in one place, making for a system that is easy and intuitive to use.” As more and more scientists regularly utilize JMP, he adds, there are “many examples” wherein scientists are performing more complex and insightful analyses than before – all with fewer resources and in a more reproducible way that can easily be shared with colleagues.
With more JMP users across R&D, several scientists reached out to Vogel and his colleagues seeking customized automation solutions to streamline data flows and analyses. Fortunately, this requirement had been planned for when the PPD organization selected JMP as its enterprise data analysis software of choice in PPD.
The key to managing innovation at scale
JMP, Vogel explains, had been selected not only for its intuitive graphical user interface – which allows scientists to ask meaningful questions to their data without the need to create code – but also for its background scripting extension. “With this capability, we can develop automated applications that can easily be utilized by scientists across R&D globally,” Vogel explains.
Over the course of the last 2-3 years and along with colleagues in the Global Digital Core (GDC), Vogel and his colleagues created a number of add-ins that provide users with easy access to previously inaccessible or time-consuming-to-access data sources, automated solutions for repetitive data transformations, and access to innovative statistical data analyses. In the field of data access, for example, they automated the parsing of data from PDF documents, thereby relieving scientists and lab technicians of many hours of tedious and repetitive manual data transfer and check.
In a second case, the team developed what they call a “2D Fingerprint Data Comparator” that allows users to characterize and compare large samples based on, for example, chromatographic and spectroscopic data at once. Previously, Vogel explains, scientists could only compare summary statistics (e.g., peak areas) or visually overlay data from two or three samples to spot differences. This visualization was limited to very small sample sizes, and it was virtually impossible to compare more than five or six samples at a time. With the JMP add-in, however, teams can now easily characterize and compare 40 or 50 samples at a time, thereby determining similarities between (groups of) samples – an efficiency which in turn helps to characterize the underlying processes and process changes.
“This application is really innovative in my eyes,” he says. “Empowering scientists to easily access data from many samples at once, to easily pre-process the data as they need it for the scientific question at hand, and to run innovative statistical analyses is really unique. This application allows us to quickly gain very precise insights that were previously inaccessible.”
Moreover, he adds, “We regularly discuss the current version of each add-in with the scientists and lab technicians who use it on a daily basis: “Is there something missing? What features are particularly helpful? How can we further improve the useability of the add-in?” And based on our agile approach, we will include these features in the next version of the add-in to support users on their scientific endeavors.”
While the development and initial distribution of add-ins to users is straightforward, it is “virtually impossible with standard methods and a growing and global user base” to ensure that all users of an add-in receive the latest version of the add-in. To this end, he and his colleagues used the UpdateManager to design an add-in that allows JMP users across CSL Behring to access all add-ins developed by the GDC. This innovation, known within CSL Behring as the Innovator’s Toolkit, enables users to receive automated notifications about add-in updates as well as helpful information and news. This add-in, Vogel says, will go live in the near future and become the next step in PPD’s digital transformation and data analysis journey.
Early indicators accelerate the move towards greater analytics adoption
The JMP user base at CSL Behring has expanded quite rapidly over the last three to four years – a fact that, in Vogel’s view, demonstrates “both how we internally advertise and use it; and that JMP is a good software in terms of what it enables us to do.” In combination with overwhelmingly positive results from the rollout of upskilling opportunities such as the 20-hours DoE class, this reception illustrates how scientific innovation is supported and fueled by statistics and the wider digital transformation.
Analytics enablement efforts, together with JMP adoption, have also been positively recognized by CSL Behring leadership and ensure the continuation of the digital transformation journey in PPD. And with the global rollout of the UpdateManager – allowing all JMP users at CSL Behring globally to access and leverage the JMP add-ins in the Innovator’s Toolkit – PPD will take the next step on their digital transformation journey.
