Global Regulatory Landscape of Artificial Intelligence in Medicine

Artificial Intelligence (AI) is profoundly transforming the pharmaceutical research and regulatory systems. Whether in drug discovery, clinical trial design, or post-market drug surveillance, AI technologies are reshaping traditional processes. However, at the same time, the uncertainties of AI - algorithm bias, data quality, model transparency - also pose unprecedented challenges for regulatory agencies.

The global drug regulatory system is shifting from "technology adaptation" to "systematic regulation". Full life cycle governance systems for AI pharmaceutical applications are being constructed, which are centered on vertical industry policies and supported by horizontal data and ethical frameworks.

AI Medical Regulation in the U.S.

The FDA is undoubtedly a pioneer in global AI pharmaceutical regulation. Its regulatory logic reflects an evolution from "exploratory guidelines" to "institutional frameworks".

Regulatory System for Medical Devices

Since 2019, the FDA has first proposed the concept of "Pre-Specified Change Control Plan" in the "Proposed Regulatory Framework for Modifications to Artificial Intelligence/Machine Learning (AI/ML)-Based Software as a Medical Device (SaMD)". This mechanism allows manufacturers to declare the learning scope and update strategies of the AI model before regulatory approval, achieving controlled learning. Subsequently, the FDA, in collaboration with the MHRA and Health Canada, released "Good Machine Learning Practice for Medical Device Development: Guiding Principles" in 2021, becoming a common reference for global AI medical device reviews.

These principles emphasize data diversity, model interpretability, performance monitoring, and clinical revalidation, establishing repeatable and traceable review standards for AI medical products.

Drug Review and Research

In 2024, the FDA established an AI Steering Committee within CDER to coordinate AI applications in drug research, review, manufacturing, and drug surveillance. The "Considerations for the Use of Artificial Intelligence To Support Regulatory Decision-Making for Drug and Biological Products" released that year first proposed how regulatory agencies themselves can use AI for risk identification and decision support.

In 2025, the FDA has further released two key draft guidelines:

• Promoting the Use of Artificial Intelligence in the Development of Drug and Biological Products

The AI model must have traceable training data, a verification process, and a performance evaluation mechanism.

• Artificial Intelligence/Machine Learning (AI/ML)-Enabled Medical Devices: Considerations for the Total Product Life Cycle

It is necessary to regularly revalidate the AI models and establish risk control documents.

AI-Enabled Self-Regulation of the FDA

The FDA not only regulates AI but is also actively using it.

• Adverse event analysis

AI for adverse event analysis is used to automatically identify potential signals in adverse reaction reports, improving the efficiency of drug alerting.

• Review assistance

The review assistance natural language processing model helps reviewers quickly analyze drug instructions and application materials.

• Manufacturing supervision

In the FRAME program, AI is regarded as the core of advanced manufacturing technology and is used for intelligent quality control and predictive maintenance in manufacturing supervision.

Summary

The U.S. AI medical regulatory system is evolving from "technical guidelines" to a "legal framework", demonstrating a highly adaptive and cross-departmental collaborative regulatory paradigm.

AI Medical Regulation in EU

AI regulation in the EU is guided by ethics and risks, forming a complete system that combines both horizontal and vertical approaches.

High-risk Classification of Medicines under the AI Regulation

The AI Act, which will come into effect in 2024, classifies medical and pharmaceutical-related AI systems as "high-risk categories". This means that all related products must:

• Undergo conformity assessment
• Submit high-quality training data and explanations of algorithm transparency
• Establish mechanisms for human supervision and risk control

AI medical products need to meet the same compliance requirements as medical devices. Manufacturers must register in the EU database and be subject to joint supervision by ECHA and EMA.

EMA's Longitudinal Regulation

In 2024, EMA released a "Reflection Paper on the Use of Artificial Intelligence (AI) in the Medicinal Product Lifecycle", systematically outlining the application principles of AI in drug research and development, production, review, and post-marketing supervision.

The document emphasizes:

• AI models should be used under the dual conditions of "scientific rationality and ethical compliance".
• The quality and interpretability of the dataset are the core of compliance.
• The regulatory authorities have the right to require enterprises to submit model verification reports and explanations of data sources.

Furthermore, within EMA, the "Scientific Explorer" AI system has been deployed to review literature searches and technical comparisons, thereby enhancing regulatory efficiency.

International Coordination and Technical Tools

The EU actively promotes international alignment of AI regulation through multilateral platforms such as IMDRF and ICMRA. For instance, it participates in the standardization of AI medical device terminology under the IMDRF framework, and promotes the collaborative development of AI risk assessment methods for pharmaceuticals under ICMRA.

Summary

The EU's regulation is centered on "risk level + ethical priority", forming a closed-loop governance system from the legislative to the operational level.

AI Medical Regulation in UK and Japan

UK Innovation Promotion through Regulatory Sandbox Mechanisms

The MHRA started operating AI-Airlock from 2024, allowing companies to use real-world data in a regulated environment to test AI medical devices. The sandbox mechanism has lowered the entry threshold for innovative enterprises and also helped regulators accumulate data on the risks of AI products. Meanwhile, the MHRA established a "Digital Technology Group" to explore the application of AI in drug surveillance and fraud detection.

Japan's Guideline-Driven, Demand-Oriented Governance Model

Japan has incorporated AI healthcare into its "Social 5.0" national strategy. The regulation mainly relies on guidelines and industry self-discipline, such as "Social Principles of Human-Centric AI" and "Guidelines for AI/ML-Enabled Medical Devices".

The MHLW clearly defined that AI-assisted diagnostic systems fall under the category of medical devices and set requirements for algorithm updates, verification frequencies, and the boundaries of doctors' responsibilities.

Summary

Both the UK and Japan adopt an "innovation-friendly" regulatory model, achieving flexible management through controlled trials and dynamic evaluations.

AI Medical Regulation in China

The AI medical regulatory system in China exhibits the characteristics of "forward-looking policies, leading standards, and digitalized supervision".

Regulation System

In recent years, the NMPA has issued a series of key policies:

• Guideline for the Registration Review of Artificial Intelligence Medical Devices

Propose requirements for algorithm annotation, training data quality, and model updates.

• List of Typical Scenarios of Applying Artificial Intelligence in the Regulation of Medicinal Products (2024)

List 15 typical AI application scenarios in the regulation of medicinal products, including intelligent review, remote supervision, and drug alerting.

• Interim Measures for the Administration of Generative Artificial Intelligence Services

Clarify the algorithm content, data security and ethical boundaries.

These documents form a policy loop covering the "research and development - registration - regulation" of AI medicine.

YY/T 1833 Series

China has taken the lead in establishing an AI medical device standard system:

• YY/T 1833.1: Definitions of AI Terms
• YY/T 1833.2: Data Set Quality Control
• YY/T 1833.3: Algorithm Verification and Traceability
• YY/T 1833.4: Software Update and Maintenance Management

This set of standards not only provides technical basis for supervision, but also offers compliance guidelines for enterprises.

Regulatory Technology and Collaboration Platform

The AI Medical Device Innovation Cooperation Platform was established by NMPA's CMDE, bringing together enterprises, research institutions and regulatory experts to jointly conduct algorithm verification and real-world data application research.

Meanwhile, the regulatory authorities themselves are also promoting "intelligent review" and "smart supervision" systems, aiming to apply AI in the approval process and risk monitoring.

Summary

China has initially established an AI medical governance system characterized by "standard setting as the priority, policy support, and digital regulation", and is gradually aligning with international standards.

Global Trends

Overall, the global AI healthcare regulation is showing three major trends.

• Risk classification and full life cycle management have become mainstream.

AI Medical products are generally classified as high-risk categories, and the regulatory focus has expanded from pre-market approval to post-market monitoring.

• Data governance and model transparency have become core regulatory indicators.

All countries require enterprises to disclose the sources of training data, the logic of algorithms, and performance verification.

• Adoption of RegTech

Various regulatory agencies are leveraging AI-assisted review and monitoring to achieve the goal of "regulating AI with AI".

Disclaimer: The above content is compiled based on existing public information and is for reference only.

Our Services

Proregulations continuously tracks and deeply analyzes the latest guidelines from global regulatory agencies such as the FDA, EU EMA, and NMPA, deeply understanding the unique regulatory challenges brought by AI to drugs and medical devices. We are committed to seamlessly integrating forward-looking compliance requirements into the entire life cycle of our clients' products, from algorithm design, product development, clinical validation, to post-market supervision.

• Regulatory consulting and Interpretation
• Product launch strategy planning
• Gap analysis
• Document preparation, pre-review and submission
• Risk management
• Regulatory communication support

Proregulations is skilled at monitoring global regulatory trends and predicting review directions, providing customers with professional, rapid responses and strategic support, helping their innovative products always meet the regulatory requirements of target markets. If you are interested in our services, please contact us.

Related Service

U.S. New Drug Application (NDA)

EU Active Substance Master File (ASMF) Production

China Drug Product Registration

U.S. Medical Device Registration with the FDA

EU Medical Device Regulation(MDR) Registration

China Medical Device Registration and Device Classification