Raw data only becomes useful for fine-tuning a drug manufacturing process through effective, appropriate analysis, says the team behind a new “information-centric” analytics framework.
The FoReSight framework—detailed here—is based on the idea data are symbols representing the properties of objects or events while information is the actionable, useful output of the analysis of data.
And this distinction is vital for effective process analytics, says lead author Marcus O’Mahony, PhD, from the PMTC, Chemical Sciences, Bernal Institute at the University of Limerick in Ireland.
“Data doesn’t have any immediate usefulness. It is often uncontextualized. Siloed. It is a record that just sits there until it is accessed, transformed, and presented as information. Data analytics turns data into information which is an actionable state of play for the data, which means decisions can be made with the information.”
“We started from the point of wanting to use data and data analytics more effectively to support decision-making in pharma and biopharma manufacturing. We found we had to talk about the information needed and how that information intends to be used rather than the data,” he tells GEN.
With this in mind, O’Mahony and colleagues developed a framework to help biopharmaceutical manufacturers turn data into actionable, trustworthy information.
Three stages
The approach has three stages: framing the information needed and its intended use; assessing the hazards around data access, data transformation, and its presentation as information; and finally ensuring there are controls to support intended information use, unintended information use, and that periodic risk review is specified appropriately.
For example, in biopharmaceutical manufacturing, the approach could be applied to information provided by a real-time dashboard for making operational changes and decisions, or to support the use of information provided by an advanced process model for indicating parameter changes to improve yields and process performance.
The overall aim is to ensure information generated at each stage of a process can be trusted by all the technicians and engineers involved.
O’Mahony said, “It is trustworthy because there is a collective transparent effort from end user to developer to approver. We even created some values and principles of the framework to help engender this collective trust.”
He added, “It’s advantageous as it proposes a methodology that is well-aligned with the most recent international guidance on quality risk management ICH Q9 (R1) as well as the FDA’s draft guidance on Computer Software Assurance.”
Another advantage is the potential to use the framework in combination with innovative process control technologies, according to O’Mahony.
“Applications discussed in the FoReSight group included enabling automated reporting such as for annual product reviews, using data analytics to support ongoing process verification efforts, process optimization, and support for process investigations.
“The effectiveness of the framework to support GenAI and LLM applications is still to be determined but if you can specify a piece of information, how you intended to use it and the hazards involved in turning it from data into information, it may turn out to be effective.”
Rethinking data
The next step is to determine how the information-centric approach can be of most benefit to the biopharmaceutical industry.
O’Mahony says,“We are considering ways this new framework could help drive improvements in digital skills and digital maturity levels, drive sound investments in data infrastructure, shape data governance policy, as well as its potential to support GenAI tools.
The framework—and in particular the focus on useful, trustworthy information rather than just data—could also inform how the next generation of bioprocess engineers are trained.
“At the University of Limerick, we are developing a four-year BSc/MSc Immersive Bioscience and Biotherapeutics program (iBio) in conjunction with major industry partners, which embeds data competencies throughout the four years. We intend to leverage some of the FoReSight learnings to facilitate data and bioanalytics integration into this curriculum,” O’Mahony said.