Amgen

Lucía Cuadros

Regulatory science, reliance, and AI are moving Regulatory Affairs beyond compliance management and toward the design of evidence, decisions, and trust.

For years, Regulatory Affairs was viewed primarily as the function responsible for assembling dossiers, interpreting requirements, and ensuring compliance. That definition is no longer sufficient. Today, decisions about medicines, advanced therapies, and connected medical devices may depend on real-world data, predictive models, assessments performed by other authorities, evolving algorithms, and overlapping legal frameworks.

Regulatory professionals still answer a fundamental question of what the law requires. Increasingly, they must also determine what evidence is credible, where it is fit for purpose, and how much uncertainty can be accepted. These questions define the growing role of regulatory science.

Regulation by Method: The Rise of Regulatory Science

The FDA describes regulatory science as the field focused on developing tools, standards, and approaches to evaluate the safety, effectiveness, quality, and performance of regulated products. The European Medicines Agency (EMA) similarly applies scientific disciplines to product evaluation and regulatory systems.

Regulatory Affairs does not simply become regulatory science. However, its greatest value increasingly depends on scientific thinking: defining questions, assessing evidence quality, understanding uncertainty, comparing alternatives, and explaining decisions in proportion to risk.

A regulatory dossier is more than a collection of documents. It is a structured argument about benefit, risk, quality, and control. Two teams may review the same clinical results, but one may focus only on submission requirements while the other identifies study limitations, weak comparators, or analytical gaps, contributing to deeper regulatory judgment.

Complexity Beyond the Checklist

Regulatory complexity can no longer be managed through checklists alone, particularly as artificial intelligence becomes part of evidence generation and decision-making.

AI applications vary significantly in regulatory importance. A tool used to summarize documents requires different controls from one used to select patients, analyze medical images, predict toxicity, monitor manufacturing processes, or detect safety signals.

Both EMA and FDA emphasize risk-based approaches. AI systems must be assessed according to their intended use, data quality, influence on decisions, and consequences of error. Key considerations include data governance, traceability, human oversight, generalizability, lifecycle management, and ongoing performance monitoring.

For Regulatory Affairs, this requires earlier involvement in development. Professionals must help define acceptable AI applications, determine documentation requirements, and coordinate with clinical, quality, pharmacovigilance, manufacturing, data protection, and market access teams. The regulatory role is no longer limited to identifying applicable requirements. It is about building evidence, accountability, and controls that demonstrate when AI models can be trusted.

Reliance: Trust Without Surrendering Judgment

The World Health Organization (WHO) defines reliance as the practice where one regulatory authority gives significant weight to assessments performed by another trusted authority while maintaining its own independence, responsibility, and accountability.

Reliance is not copying another approval or transferring regulatory responsibility. It requires understanding the original assessment, confirming product similarity, and determining whether conclusions apply within a different regulatory context. A product approved elsewhere may share the same active ingredient but differ in formulation, manufacturing site, specifications, storage requirements, indication, product information, or risk management approach. The value of reliance lies in identifying what can be reused and what requires additional assessment.

A strong reliance strategy should clarify four questions. What was assessed? By whom and against which standards? What remains equivalent? What requires further review? Comparison matrices, gap analyses, version tracking, and change controls transform reliance from an administrative shortcut into a scientific decision-making process.

For the industry, this means moving beyond a static global dossier. Reusable evidence must be supported by disciplined cross-jurisdictional comparisons and transparent change history.

Artificial Intelligence: From Productivity Tool to Evidence Component

AI is increasingly moving from an efficiency tool to a component of regulatory evidence. When AI influences clinical decisions, endpoint selection, safety assessments, manufacturing controls, or benefit-risk evaluation, it becomes part of the scientific method and requires appropriate governance.

EMA promotes lifecycle-based oversight, where scrutiny depends on patient risk, regulatory impact, context of use, and the degree of influence the model has on decisions. FDA similarly emphasizes that AI credibility must be demonstrated through evidence appropriate to the model’s intended purpose and risk level. Its credibility framework highlights key steps: defining the question of interest, establishing the model’s context of use, assessing risk, developing a credibility plan, executing validation activities, documenting results, and determining whether the model is suitable for its intended use.

The principle is clear: evidence requirements must match risk and influence.

The growing use of AI in regulatory documentation also reinforces the need for transparency. AI-assisted work requires understanding the tool used, its version, purpose, limitations, and level of human oversight. Writing effective prompts is not the same as governing AI responsibly.

Regulatory professionals do not need to become programmers, but they must be able to evaluate data quality, challenge assumptions, understand limitations, and determine whether AI outputs are reliable and appropriate for regulatory decisions.

Capabilities That Will Shape the Future of Regulatory Affairs

The future regulatory professional will need broader capabilities:

Evidence literacy: Understanding study design, biostatistics, real world data, modeling, and validation to determine whether conclusions are supported by evidence.

Regulatory systems thinking: Connecting development, authorization, manufacturing, surveillance, health technology assessment, and access decisions.

Data and AI governance: Managing data lineage, version control, bias, privacy, cyber security, explainability, and lifecycle changes.

Comparative assessment skills: Understanding which regulatory reasoning can be reused through reliance and which elements require local evaluation.

Scientific communication: Translating complex evidence into clear, transparent, and auditable narratives for regulators, clinicians, engineers, lawyers, and patients.

Ethical accountability: Ensuring automation does not replace human responsibility for decisions involving safety, equity, and trust.

Continuous learning: Developing competencies that evolve alongside scientific and technological innovation.

It is too early to predict how many regulatory roles AI may replace. However, repetitive tasks will likely become increasingly automated, making human judgment even more valuable. The differentiating skills will be context, critical interpretation, accountability, and the ability to explain decisions.

A Profession Built on Judgment and Trust

The future of Regulatory Affairs is not about collecting more rules or adopting every new technology. It is about improving the quality of decisions. Regulatory science provides the methodology. Reliance enables the responsible use of knowledge from trusted authorities. AI creates value only when governed effectively. The regulatory professional of the future will become an architect of trust, knowing when to reuse evidence, question assumptions, apply AI appropriately, and recognize when uncertainty is too significant to ignore.

Regulatory Affairs is expanding beyond dossier management to make judgment, accountability, and public health protection explicit, defensible, and reproducible.