Can Predictive Modelling Fix Fragmented Pharma Development Data?

A collaboration between Centrix Pharma Solutions, a UK-based CDMO, and the University of Sussex is targeting one of pharmaceutical development's persistent inefficiencies: the failure to fully use the data that development programs already generate.¹ The partnership, structured as a Knowledge Transfer Partnership and supported by Innovate UK, will apply data analytics and predictive modelling to early-stage product development. The £375,000 ($510,000) project pairs formulation development expertise with academic research in data science to build tools intended to make development decisions more consistent and less dependent on individual judgment.

Kate Thorpe, Head of Innovation and Business Partnerships, University of Sussex, stated in a press release,¹ “Knowledge Transfer Partnerships bring together academic expertise and real-world industrial challenges. Our ambition is not only to solve today's challenges but also to establish new ways of working that continue delivering value and innovation long after the project has concluded.”

Why Does Fragmented Data Slow Down Development Programs?

Pharmaceutical development generates substantial formulation, analytical and process data across every program.¹ Challenges arise when that information tends to sit in disconnected systems and separate projects, which limits its usefulness for spotting trends or informing future work. As a result, decisions such as formulation selection, dosage form optimization and development pathway planning often rely on manual review and the experience of individual scientists rather than on the aggregate dataset available to an organization.

That gap is relevant industry wide.¹ When development decisions rest heavily on manual analysis, the risk of rework, delays and suboptimal outcomes increases, and those costs eventually show up in timelines and budgets across the industry. For professionals evaluating how to structure early-stage programs, the initiative is a case study in whether systematic data analysis can be embedded into routine decision-making rather than applied only after problems emerge.

The project will focus initially on early-stage development, working to build methodologies that identify development risks sooner, reduce unnecessary experimentation and support more confident decisions as programs move toward clinical evaluation.¹ A dedicated Knowledge Transfer Partnership Associate will be recruited to lead the work and serve as a link between the academic and industrial sides of the collaboration, with the goal of establishing data science capability that remains in place after the project concludes.

“Much of the progress so far has come from academic research, which often seeds future industry innovation, as well as from emerging startups and the steady improvement of open-source models,” says Pauric Bannigan, chief science officer and cofounder of Intrepid Labs in an article from our October 2025 issue on predictive modeling.

What Does Predictive Modelling Look Like in Practice?

Peer-reviewed research offers a window into how predictive modelling is already being applied in early drug discovery.² One study used information-theoretic models to analyze thousands of biological screening samples, identifying which molecular features most strongly influenced compound activity. The approach achieved high classification accuracy, though researchers cautioned that skewed datasets, where active compounds are rare, can inflate apparent performance if accuracy is the only metric considered. The findings underscore a broader lesson for data-driven development that predictive tools are only as reliable as the datasets and evaluation methods behind them.

In a PharmTech article published in October 2025 on turning real-world data into development insights through AI, Gustavo Ferrer, MD, CEO of Dr. Ferrer BioPharma and Moxie Health Group, says, "Predictive modeling identifies promising candidates thousands of times faster than traditional methods."

What Does this Signal for the Development and Manufacturing Sector?

Chris Davison, CEO, Centrix, stated in the press release,¹ “The pharmaceutical industry generates an enormous amount of valuable data throughout product development, but there is still significant untapped potential to use that information more effectively. We recognized an opportunity to combine our pharmaceutical expertise with advanced data analytics to make better-informed decisions throughout the development process.” He continued, “The Knowledge Transfer Partnership gives us access to specialist academic expertise while embedding those capabilities within our business. This project represents an exciting opportunity to apply advanced data science techniques to a real-world pharmaceutical development challenge. Beyond the immediate project, we're creating a long-term capability that will benefit both our team and our clients, helping us deliver more efficient development programs.”

For professionals tracking how development organizations are building analytical capability, the initiative reflects a wider industry question.¹ Can structured, predictive use of development data be scaled into standard practice, or will it remain dependent on individual partnerships and funding cycles? The answer will shape how quickly data-driven development approaches move from pilot projects to routine practice across the sector.

References

1. Centrix Pharma Solutions. Centrix Pharma Solutions and University of Sussex partner to advance data-driven pharmaceutical development. East Essex, UK: Centrix Pharma Solutions. Press Release. August 26, 2026.
2. Saied H, Alfahad O, Aljaffer AA, AlOmari AK, Saleh MA, Malaekah E. From data to discovery: The rise of information-theoretic predictive models in drug development. Sci Rep. 2026;16(1):12857. Apr 17, 2026. doi:10.1038/s41598-026-45644-5