NMPA Issues World’s First Industry Standard for Medical Devices Using/Adopting BCI Technology

On September 14, 2026, China’s National Medical Products Administration (NMPA) issued YY/T 2029–2026, “Medical Devices Incorporating Brain-Computer Interface Technology — Quality Requirements and Evaluation Methods for EEG Datasets Used for Artificial Intelligence Algorithms.” The standard establishes quality requirements and evaluation methods for electroencephalogram (EEG) datasets used by AI algorithms in brain-computer interface (BCI) medical devices.

The standard will take effect on September 1, 2027. While the official standard document has not yet been published, its release is expected soon, providing further detail on the requirements and evaluation methods manufacturers will need to consider.

Completing China's BCI Regulatory Architecture

The new standard builds on a series of BCI-specific standards developed by the NMPA since 2025: 

• September 2025 (YY/T 1987–2025): Standardized core BCI terminology.
• September 2025 (YY/T 1996–2025): Established performance testing methods for closed-loop implantable neural stimulators.
• January 2026: Fast-tracked development for upcoming standards, including paradigm design for motor rehabilitation.

While the earlier standards addressed terminology and device-level performance, YY/T 2029—2026 moves into the data layer underlying AI-enabled BCI systems. It establishes a dedicated framework for assessing the quality of the EEG datasets on which these algorithms rely.

What Does YY/T 2029—2026 Require?

The standard governs EEG datasets across the entire lifecycle — from early-stage R&D through to commercial manufacturing and quality control — for any BCI medical device relying on AI to post-process EEG signals.

It targets four core pillars of dataset management:

• Data collection: Requirements and evaluation methods for the collection of EEG data.
• Data annotation: Quality considerations for the annotation of datasets used by AI algorithms.
• Data storage: Requirements relating to the storage and management of EEG datasets.
• Quality control: Criteria and evaluation methods for assessing and maintaining dataset quality.

For manufacturers whose EEG datasets and data-management processes were developed under other regulatory frameworks, these requirements create a new reference point for assessing existing practice — especially when those datasets form part of a product's technical documentation for NMPA registration.

“Recommended” Does Not Mean Irrelevant

Although YY/T 2029–2026 is designated as a recommended standard (YY/T), foreign companies should treat compliance as an operational expectation rather than an optional guideline.

AI-enabled BCI medical devices are typically categorized as high-risk Class III devices, requiring centralized NMPA registration and review by the Center for Medical Device Evaluation (CMDE). During technical review, CMDE examiners routinely evaluate submissions against published recommended standards.

Therefore, aligning datasets with YY/T 2029–2026 will serve as critical technical evidence in the submission, while non-alignment will require extensive justification. 

What Medical Device Companies Should Do Before September 2027

Foreign neurotechnology manufacturers planning to enter or register products in China before or after the September 1, 2027 effective date should take four key steps:

1. Define the scope: Determine whether the device uses AI to post-process EEG signals and therefore falls within the standard’s scope.
2. Audit existing datasets: Review data collection, annotation, storage, validation, and quality-control practices against the new requirements.
3. Close the gaps: Update data-management procedures, technical protocols, and documentation where current practices fall short.
4. Build dataset quality into registration documentation: Consider how relevant dataset quality requirements and metrics should be reflected in Product Technical Requirements (PTR) and supporting documentation.

Final Thoughts

YY/T 2029–2026 signals a fundamental evolution in China’s regulatory oversight: attention is moving downstream from general algorithm performance to the health and integrity of the underlying training data. For international developers, the immediate priority is verifying whether existing data governance can withstand CMDE scrutiny before the 2027 deadline.

Navigating this transition demands a proactive strategy that bridges technical data management with CMDE expectations. At Cisema, we help global neurotech and medtech innovators successfully enter China's market by translating complex regulatory shifts into actionable compliance. Whether interpreting new standards, conducting gap analyses, drafting Product Technical Requirements (PTRs), or assembling complete NMPA submission files, our team ensures your AI and BCI innovations move smoothly from global development to market approval.

For companies seeking seamless regulatory support for medical device registration in China, contact Cisema today.

Further Information

Explore Cisema's services for medical device registration in China.

References

National Medical Products Administration Announcement on Issuing the Medical Device Industry Standard YY/T 2029–2026, “Medical Devices Incorporating Brain-Computer Interface Technology — Quality Requirements and Evaluation Methods for EEG Datasets Used in Artificial Intelligence Algorithms” (Announcement No. 79 of 2026)