Artificial intelligence (AI) and machine learning (ML) are pervasive in every industry, and the pharmaceutical industry is no exception. AI and ML are already impacting drug development and manufacturing, but these innovations pose unique regulatory challenges.For example, if an ML algorithm can uniquely modify a CGMP-compliant manufacturing process to increase efficiency, how will a pharmaceutical manufacturer ensure that the updated machine-made process is CGMP-compliant? When researchers use AI to identify ideal candidates to participate in drug trials, how do they consider bias in the data underlying AI decision-making? Announced[1] Seek input from pharmaceutical industry stakeholders on how to address these issues. Stakeholders should use this opportunity to provide input to FDA as it develops the applicable regulatory landscape.
In the following, we describe real-world examples of AI/ML in drug development and manufacturing, describe the overarching principles that guide FDA’s current thinking on the regulatory landscape of AI/ML, and highlight the key FDA recommendations to industry stakeholders. Focus on the question.
What are AI and ML?
The FDA describes AI as “the branch of computer science, statistics, and engineering that uses algorithms and models to perform tasks and exhibit behaviors such as learning, decision-making, and prediction.” ML is a subset of AI that allows “models to be developed by ML training algorithms through analysis of data without explicitly programming the model.”[2]
Applications of AI and ML in drug development and manufacturing
An FDA discussion paper highlights the potential benefits and applications of AI and ML in drug development and manufacturing. for example:
- AI/ML creates a “digital twin” of an individual, enabling predictive modeling of that individual’s response to a particular drug prior to use.
- AI/ML can optimize existing pharmaceutical manufacturing processes, maximize efficiency and minimize waste. Continuous, real-time sensor data enables manufacturers to detect changes and deviations in the manufacturing process that indicate the need for equipment maintenance.
- AI can monitor product quality. For example, AI can perform quality control inspections of product packaging by analyzing the product for visible deviations from images pre-programmed into AI-based software.
- AI/ML can advance logistics and prevent supply chain disruptions by predicting product demand and optimizing inventory.
- Post-production, AI can collect and monitor consumer complaints and adverse events, and identify trends in reported product issues, so the fundamental Identifying the cause may be expedited.
Overarching principles for the use of AI/ML in drug development and manufacturing
While the FDA is still in the early stages of developing a regulatory approach to AI and ML in drug development and manufacturing, the FDA has outlined three key principles to guide thinking on these issues. (2) data quality, reliability and representativeness; (3) model development, performance, monitoring and validation; ”
First, the FDA believes that human-driven governance is a priority for building trustworthy AI/ML, and the future AI/ML regulatory framework will require AI/ML to be implemented in drug development. transparency and documentation will likely be required for all decisions and process changes. and manufacturing space. This approach is consistent with the FDA’s approach to AI and ML in the medical device space. FDA is proposing to require manufacturers to monitor changes in software implemented by AI/ML and provide regular updates to FDA on these changes.[3]
Second, FDA is concerned about the quality and reliability of the data underlying the AI/ML process, particularly as AI/ML can “amplify existing biases that exist in the underlying input data.” is holding The FDA’s focus on identifying and managing bias means that future regulatory frameworks will require documentation and explanation of how bias in the underlying data AI/ML was managed in the drug development process. Indicates that it is likely to be required.
Finally, FDA has specified the importance of regular monitoring and documentation to ensure AI/ML models are explainable, reliable, and verifiable, but FDA also , also states that “a risk-based approach may guide the level of evidence and recordkeeping required.” Validation and validation of AI/ML models in specific usage situations. The FDA’s future regulatory framework will impose less stringent requirements for simpler, more transparent modeling, and require more documentation and auditing for complex models (such as artificial neural networks). may become.
FDA Asks Pharmaceutical Industry Stakeholders for Feedback
In light of the three overarching principles described above, FDA will identify specific areas that require feedback and discussion with stakeholders to inform regulatory activity. These include:
- How can the pharmaceutical industry ensure accountability, transparency and trust in AI/ML systems that may not be easily explained or understood due to their complexity?
- How can drug developers using AI/ML prevent amplification of errors and biases in underlying data sources and ensure patient data privacy?
- Parties that use cloud applications (particularly from third-party hosts) to store product manufacturing data must ensure the integrity, quality, and security of the data, especially when these issues impact the manufacturer’s CGMP obligations and monitoring requirements. How can we ensure that?
- How does a pharmaceutical company store data generated for regulatory compliance (such as data supporting future quality decisions such as product recalls) in a way that enables retrieval and analysis to support decision making? can you?
- If ML algorithms modify and adapt processes based on real-time data, how can parties ensure that processes controlled by ML algorithms comply with regulatory obligations?
Conclusion
Stakeholders should strongly consider using an invitation from FDA to participate in discussions as FDA develops its approach to AI/ML in drug development and manufacturing. Industry players are in the best position to educate the FDA on these evolving issues. FDA requested comments electronically or in writing. August 9, 2023. You can submit your feedback here. We will continue to monitor regulatory developments in this area and report back as information becomes available.
