Insights from FDA’s Senior Policy Advisor on Emerging Technologies and AI/ML
The Parenteral Drug Association (PDA) Interest Group “Data Integrity and Data Governance” (aka the DIG) lead by Kir Henrici, The Henrici Group, and Ulrich Köllisch, GxP-CC, recently held a Virtual Meeting titled “Data Governance and Data Integrity in Context with Digitalization, Emerging Technologies, and Artificial Intelligence (AI)/Machine Learning (ML)”. This event was a significant forum for discussing intersection of pharmaceutical digital transformation, data governance, and regulatory oversight. A key highlight of the event was a presentation by Dr. Krishna Ghosh, Senior Policy Advisor at the FDA, who shared insights on the challenges and opportunities of integrating new technologies into pharmaceutical industries.
The following summary captures the key points from Dr. Gosh´s presentation, emphasizing FDA’s evolving approach to emerging technologies, ensuring data integrity, and addressing the challenges of regulatory compliance.
The Key Highlights from the meeting:
1. Navigating AI and ML in Pharmaceutical Manufacturing: FDA Insights and Industry Challenges
The application of AI and ML in pharmaceutical manufacturing is growing rapidly. These technologies promise improved efficiency and transformative opportunities for drug development and production. However, as noted by Dr. Krishna Ghosh, the path forward is nuanced. While the potential is enormous, the validation and regulatory framework for AI/ML applications must remain a top priority. Dr. Ghosh emphasized that the existing regulatory framework is not obsolete but instead adaptable. She explained, “The regulations are there…you just need to interpret them in this modern language,” underlining that the standards already in place can accommodate these advanced models.
No New Regulations, but Guidance for Implementation
When asked about the potential for new FDA regulations specifically governing AI/ML, Dr. Ghosh clarified that there are no plans for dedicated regulations targeting these technologies. Instead, the FDA is relying on general guidance documents to help stakeholders navigate emerging technologies within the existing regulated framework considering the following key point:
– Stakeholder Responsibility; Assessing Risks: Pharmaceutical manufacturers are responsible for assessing and managing the risks associated with AI/ML technologies. The primary areas of focus are:
o Patient Safety: AI and ML should enhance patient outcomes, not jeopardize safety. This means ensuring that any technology used to influence production or quality decisions undergoes rigorous validation.
o Product Quality: Quality must be maintained regardless of the level of automation involved. AI/ML should consistently meet predefined quality metrics to be used in decision-making processes.
o Data Integrity: Trustworthy AI/ML models require impeccable data governance. Ensuring data quality, integrity, and traceability is crucial, as data forms the basis for both model training and operational insights.
– FDA Oversight Through Drug Approval: The FDA maintains oversight during the drug approval process by reviewing the methods and technologies used. Any technology that poses risks or raises questions may influence the FDA’s approval decisions. Dr Ghosh highlighted that while AI/ML models can be highly effective, such as in visual inspections of sterile injections, they are not yet allowed to fully replace human decision-making during final inspections. This reflects the FDA’s cautious but open-minded approach to ensuring technologies are dependable before full autonomy in critical areas.
– Communication with Authorities: The FDA encourages manufacturers to maintain an open dialogue with regulatory bodies. The FDA is receptive to engagement and wants to support stakeholders in adopting new technologies safely. Communicating with the FDA early and often can help clarify expectations and reduce risks
2. Data Governance: ALCOA+ (+) Framework
In the next point, Dr. Ghosh addressed the necessity of robust data governance as industries move from traditional paper-based systems to digital formats. She emphasized data governance needs to also ensure proper management of metadata. “Metadata is critical… it should travel or have a repository under the data management plan,” she pointed out, underscoring the importance of managing the data during the entire life cycle.
Dr. Gosh mentioned in this context the update of the ALCOA +(+) Framework which adds additional attributes to ensure robust data governance in today´s complex digital environments:
Complete: No data should be omitted, including failed or incomplete batches. Every step of the process must be documented.
Consistent: Data must be captured and presented in a logical, chronological order to ensure reliability.
Enduring: Data and its metadata must be preserved throughout its lifecycle, following retention policies.
Available: Data must always be readily accessible in its original format during its entire retention period, allowing easy inspection or review.
Traceability (The new “+”): This critical addition emphasizes linking all data back to its raw source, allowing each step in the data’s journey to be reconstructed. This ensures full transparency of the data history.
3. Data Lakes, Warehouses, and Governance
Another key point of discussion was the evolution of data storage systems, from data warehouses to data lakes and now lake houses. While data lakes offer flexibility for storing structured and unstructured data, Dr. Ghosh pointed out that they also come with challenges: “Data lakes can quickly become swamp lakes without proper governance.” She advised the industry to adopt data lake houses, which combine the flexibility of data lakes with the governance and structure of traditional data warehouses, enabling more secure and traceable data handling.
Conclusion:
The webinar provided valuable insights into the complexities of integrating digital technologies into pharmaceutical manufacturing while maintaining adherence to regulatory standards. Dr. Krishna Ghosh’s expertise highlighted the challenges and opportunities that lie ahead as the industry navigates data integrity and governance in an era of rapid technological advancement. Her discussion reinforced the FDA’s commitment to maintaining rigorous oversight while adapting to new technologies, ensuring that both innovation and safety remain at the forefront of the pharmaceutical industry’s digital transformation journey.