Advancing Healthcare Innovation: FDA Explores AI/ML in Drug Development and Devices

May 24, 2023 | News |

Advancing Healthcare Innovation: FDA Explores AI/ML in Drug Development and DevicesThe Food and Drug Administration (FDA) is actively exploring the use of artificial intelligence (AI) and machine learning (ML) in drug development and medical device creation. In collaboration with various centers, including the Center for Drug Evaluation and Research (CDER), Center for Biologics Evaluation and Research (CBER), and Center for Devices and Radiological Health (CDRH), the FDA aims to engage stakeholders in a discussion on leveraging AI/ML technologies to improve the regulatory landscape in these areas. As technology rapidly advances, with powerful data collection tools and advanced computing capabilities, drug development, and use may transform. Recognizing the unique opportunities and challenges presented by this evolving ecosystem, the FDA is committed to working with partners to fully harness these innovations' potential for the public's benefit. The discussion paper titled “Using Artificial Intelligence and Machine Leanrning in the Development of Drug and Biological Products” fosters dialogue and mutual learning among developers, manufacturers, regulators, and academic groups, to explore the specific applications of AI/ML throughout the drug development process. It is important to note that this paper does not represent an FDA guidance or policy. It serves as an initial communication to facilitate engagement and encourage stakeholder feedback, which will help shape the regulatory landscape in this field.

The Landscape and Current Potential Uses of AI/ML

The discussion paper explores the landscape of current and potential applications of AI/ML (Artificial Intelligence/Machine Learning), specifically within the development of drugs and combination products. One notable advantage is the ability to accelerate the development process, enabling safe and effective products to reach patients more swiftly. Additionally, AI/ML can contribute to broader access to medical devices and combination products, thereby improving health equity by ensuring that innovative solutions are available to a wider population. AI/ML also holds promise in improving the quality of manufacturing processes, ensuring that drugs and combination products meet stringent standards. By leveraging AI/ML technologies, manufacturers can enhance efficiency, reduce errors, and streamline production, ultimately leading to higher-quality products. Furthermore, the discussion paper highlights the role of AI/ML in improving the safety of drugs and combination products. AI can help identify potential risks, analyze safety data, and predict adverse events through advanced algorithms and machine learning models, enabling proactive measures to mitigate harm and enhance patient safety.

By analyzing vast amounts of data, AI/ML algorithms can identify patterns, discover new insights, and uncover innovative approaches for addressing medical needs. This can lead to the creation of groundbreaking devices, drugs, and combination products that were previously unexplored. Personalized treatment approaches also stand to benefit from AI/ML advancements. By analyzing patient data, including genetic information, medical history, and treatment outcomes, AI/ML algorithms can assist in tailoring treatments to individual patients, optimizing efficacy, and minimizing adverse effects. The paper provides specific examples of AI/ML applications throughout the development process of drugs; however, the same principles can be applied to combination products and medical devices. AI/ML can be applied from early discovery and research stages to postmarket surveillance and advanced manufacturing. The examples presented are not exhaustive, but highlight the potential impact of AI/ML in this domain.

Additionally, the text emphasizes the importance of shared learning and collaboration in leveraging AI/ML effectively. Finally, it acknowledges that regulatory clarity is crucial in implementing AI/ML technologies and encourages further exploration of areas where regulatory guidelines can provide guidance. While the potential utility of AI/ML extends to clinical practice, the primary focus of the discussion paper is on its applications within the development process of drugs and combination products.

Considerations for the Use of AI/ML in Drugs and Medical Device Development

Considerations for the use of AI/ML in drug development are explored throughout the paper. It emphasizes that AI/ML has been applied to various drug development activities and has the potential to accelerate the process and improve clinical trial safety and efficiency. However, assessing the specific risks and harms introduced by AI/ML is crucial.

One concern is the potential for AI/ML algorithms to amplify errors and biases in underlying data sources. Extrapolating findings beyond the testing environment may raise concerns regarding generalizability and ethical considerations. Additionally, the limited explainability (Explainability is how to take an ML model and explain the behavior in human terms)

Furthermore, specific guiding principles have been published for developing Good Machine Learning Practices (GMLP) for medical devices that utilize AI/ML. While these standards and practices were not initially tailored for drug development, their utility and applicability in this field are being explored. The FDA is considering approaches to provide regulatory clarity for using AI/ML in drug development, guided by an expanding body of knowledge and an understanding of the opportunities and challenges involved. Stakeholder feedback will help inform future regulatory activities. The FDA aims to initiate discussions with stakeholders and seek input on three key areas of AI/ML in drug development:

  • Human-led governance, accountability, and transparency
  • Quality, reliability, and data representativeness
  • Model development, performance, monitoring, and validation.

A risk-based approach will be employed, with measures tailored to the level of risk posed by the specific context of AI/ML use.

Next Steps for Stake Holder Engagement

The text outlines the following steps and engagement plans in regards to using AI/ML in drug development. The release of the discussion paper is part of a broader effort to communicate and explore relevant considerations with stakeholders. The FDA has included a series of questions for feedback and plans to conduct a workshop to facilitate further engagement. In addition, the FDA will provide various mechanisms to engage with stakeholders, sponsors, and developers to address questions and concerns related to AI/ML before conducting studies. These mechanisms include formal meetings, as well as programs like Critical Path Innovation Meetings (CPIM), The Innovative Science and Technology Approaches for New Drugs (ISTAND) Pilot Program, Emerging Technology Program, and Real-World Evidence Program, which facilitate communication and discussion on relevant AI/ML methodologies and technologies to enhance efficiency and quality in drug development. Engagement with patients and the public is also emphasized to ensure patient-centered approaches and policies regarding AI/ML in drug development. Building on the discussion paper, the FDA will continue to seek feedback and engage a broad group of stakeholders to discuss further considerations for utilizing AI/ML throughout the drug development life cycle. The goal is to foster ongoing discussions and collaborations with stakeholders in the future.

The FDA's active exploration of artificial intelligence (AI) and machine learning (ML) in drug development and medical device creation demonstrates a commitment to innovation and collaboration. By engaging stakeholders and seeking feedback, the FDA aims to shape the regulatory landscape and fully harness the potential of AI/ML technologies to benefit public health. Integrating AI/ML can accelerate drug development, improve manufacturing processes, enhance safety, drive the development of novel devices, and enable personalized treatment approaches. However, considerations regarding errors, biases, and transparency in AI/ML algorithms must be addressed by developing standards for trustworthy AI. The FDA's engagement plans, including workshops and programs, provide avenues for communication and collaboration with stakeholders, ensuring ongoing discussions and collaborations to leverage AI/ML effectively throughout the drug development life cycle. Through regulatory clarity and a patient-centered approach, the FDA aims to navigate the implementation of AI/ML in a manner that maximizes its potential while ensuring ethical and responsible use, ultimately advancing the future of healthcare.


Trevor Klemann – Associate Medical Device Consultant