Biopharma is embracing artificial intelligence (AI), according to the European industry group, IFMPA, which points to process development and visual inspection as examples of areas where use of such technology is becoming commonplace.
The biopharmaceutical industry, like other sectors, has already been convinced of the merits of AI, says Sergio Cavalheiro Filho, the Geneva-based organization’s regulatory affairs manager.
“AI, including through machine-learning models for process prediction and control—for example, digital twins, which are in silico models of physical systems—is already adding real value to pharmaceutical manufacturing.”
Biopharma has a reputation for being conservative when it comes to the adoption of new technologies or processes. The argument is that the potential benefits of any novel approach or system will be offset somewhat by the effort needed to convince regulators of its merits.
Similarly, for approved biopharmaceutical manufacturing processes, drug companies need to justify any modifications to regulators, which can be a complex, expensive process.
With AI, the dynamics are somewhat different. While there is still a degree of reticence, many biopharmaceutical companies are already embracing the technology in manufacturing, according to Cavalheiro Filho.
“Companies are adopting these processes in a stepwise way to ensure regulatory compliance, and broader adoption is therefore dependent on globally harmonized regulatory guidance that is predictable and aligned with established quality control frameworks,” he tells GEN.
Digital twins and QA
IFPMA set out its support for AI in a position paper this month, citing the technology’s ability to derive value from historical manufacturing data as an example of how the drug industry is already benefiting.
“The digital twin translates historical experience or designated process development data with the same or similar product or their manufacturing process into actionable information.
“Digital twins can be used either as part of the control system or outside the established control system for improving planning and scheduling of production runs, with a different resulting level of risk,” the industry group writes.
IFPMA also notes the growing use of deep learning algorithms, a type of machine learning that uses a hierarchical network of nodes—hence the name “deep”—to detect patterns in complex data, in quality assurance, particularly in visual inspections.
“Deep learning-powered algorithms and computer vision can help reduce false reject rates, minimize manual re-inspection, prevent delayed product release, and reduce the loss of product that complies with the required specifications,” the authors write.
Looking forward
Regulation will determine whether biopharma adoption of AI accelerates, according to Cavalheiro Filho, who adds that clarification in areas like risk management would be a positive for the industry.
“A harmonized risk management framework for AI models in the pharmaceutical sector can support manufacturers to adopt AI for increased agility and efficiency and advance the safe and effective use of AI in pharmaceutical manufacturing,” he says.
