
PDA 2025: Data Governance and AI's Impact on Drug Manufacturing
Key Takeaways
- AI is revolutionizing drug discovery, development, and manufacturing by leveraging vast amounts of unstructured data, enhancing safety, quality, and efficiency.
- Effective AI implementation in bio/pharma requires rigorous data preparation, adhering to the FAIR principles, ensuring data is findable, accessible, interoperable, and reusable.
Data integrity and quality are paramount for drug discovery, manufacturing efficiency, regulatory compliance, and patient safety.
The bio/pharmaceutical industry is on the cusp of a profound transformation, driven by an exponential surge in data generation and the sophisticated capabilities of artificial intelligence. With approximately 400 exabytes of data, equivalent to 18,000 trillion books, produced globally each day—much of it unstructured—the challenge and opportunity for leveraging this information are immense, said
During his Monday presentation on “
How can bio/pharma operations maximize data value?
At its core, AI is a mathematical tool, Manzano explained in his talk. However, AI’s efficacy hinges entirely on the quality and management of data. Data in the pharmaceutical context extend far beyond traditional numbers and lists, encompassing diverse unstructured images, sounds, and continuous values collected from various systems.
Regulators are keenly aware of this, frequently referencing data and metadata in AI guidelines, emphasizing their critical role in transforming raw information into actionable knowledge. This focus is not without reason, Manzano said, emphasizing that more than 25% of warning letters issued by FDA since 2019 have cited data accuracy issues, a complex problem that continues to challenge the industry. This concern is further underscored by
The journey to effective AI implementation is heavily weighted toward data preparation, consuming an estimated 80% of an AI project's time. This rigorous preparation ensures that data are findable, accessible, interoperable, and reusable, known as the FAIR principles, which are essential for quality AI outcomes. As more data is integrated, the potential for knowledge extraction and sophisticated AI models increases, yet this also adds layers of complexity to data and model management, Manzano said in his talk.
What AI-driven innovations are the industry seeing in bio/pharmaceutical manufacturing and development?
The practical applications of AI across the bio/pharmaceutical lifecycle are profound. Consider the management of human plasma, a highly variable raw material for which donations cannot be refused and establishing full product genealogy is exceptionally difficult, Manzano highlighted. AI can be applied here to integrate batch record information with data from other systems, such as programmable logic controllers and online controls, which transforms these diverse data into knowledge that can guarantee expected yields. Through AI, crucial process variables, such as pH in plasma fractionation, can be identified and adjusted to achieve target outcomes. This capability moves beyond trial-and-error, enabling predictive models that recommend optimal process parameters based on raw material characteristics and desired yields (3,4).
Beyond manufacturing, AI is also advancing quality control in novel therapies. For instance, the Parenteral Drug Association, in collaboration with the European Medicines Agency, is utilizing AI for quality control in advanced therapy medicinal products, allowing for the prediction of batch success an hour in advance (5). This collaboration provides critical lead time for decision making without replacing human labor or traditional lab testing, enhancing efficiency and patient safety for those awaiting these crucial treatments, Manzano said.
"Everyone in this room, we are working for pharma, but we are [also] working for patients, so we have to [always] have in mind that everything we do is because there is a patient waiting,” he emphasized in his talk.
The industry's embrace of digitalization is progressing, Manzano concluded, pointing to a recent survey indicating an average maturity score of 3.5 out of 5. That is a notable increase from 2.6 in 2019. While challenges such as cybersecurity compliance and budget constraints remain top concerns for the industry, the perceived benefits are clear, Manzano asserts: respondents anticipate significant improvements in process efficiency and quality through digitalization, especially in areas such as production, research, and quality control. AI, when applied with scientific rigor and supported by high-quality, well-managed data, offers a significant opportunity to navigate the inherent complexities and variability of bio/pharmaceutical operations, ensuring that the ultimate focus remains on the patient.
References
1. Longhi, S.; Ventura, S.; Macedo-Ribeiro, S.; et al. When Artificial Intelligence Meets Protein Research [Version 1; Peer Review: 2 Approved, 1 Approved with Reservations]. Open Res Europe 2025, 5, 185. DOI:
2. Falvey, et al.
3. IDBS. Enhancing Process Performance with AI in Bioprocessing. Blog Post. May 22, 2025.
4. Cheng, Y.; Bi,X.; Xu, Y.; et al. Artificial Intelligence Technologies in Bioprocess: Opportunities and Challenges. Bioresour. Technol. 2023, 369, 128451. DOI:
5. Manzano, T. PDA Europe and EMA Unite. PDA Letter. May 27, 2025.
Newsletter
Stay at the forefront of biopharmaceutical innovation—subscribe to BioPharm International for expert insights on drug development, manufacturing, compliance, and more.