Ensuring GDPR Compliance in AI Research Using Patient Data

March 6, 2025

Vladislav Zaimov is an experienced Telecommunications specialist with expertise in enterprise telecommunications and the risk management of vulnerable networks. Today, we’ll be discussing important considerations under GDPR when re-using patient data in AI systems for scientific research.

Can you explain the significance of the GDPR’s purpose limitation principle regarding the use of patient data in AI systems for scientific research?

The GDPR’s purpose limitation principle is crucial because it ensures that personal data is processed only for specific, legitimate purposes, as initially intended. When re-using patient data to train AI systems for scientific research, an organization must assess whether this secondary use is compatible with the original purpose. This principle helps maintain the integrity and trust in data handling practices by preventing unauthorized or unexpected repurposing of sensitive personal data.

What factors determine the compatibility of secondary use of patient data according to GDPR? Can you elaborate on the importance of the link between the initial and secondary purposes? How does the context and the reasonable expectations of data subjects impact the compatibility assessment? Why is the nature of the data, especially its sensitivity, crucial in this assessment? What are the potential consequences for data subjects that must be evaluated? What types of safeguards should be in place to ensure the protection of patient data?

Several factors influence the compatibility assessment under GDPR. Firstly, a strong link between the initial and secondary purposes increases the likelihood of compatibility. If the secondary use aligns closely with the initial purpose, it’s more likely to be justified. The context and reasonable expectations of the data subjects are also vital; if data subjects were informed about possible secondary uses when their data was collected, compatibility is more justifiable. The nature of the data, particularly its sensitivity, is crucial—the more sensitive the data, the narrower the scope for compatibility. Evaluating the consequences for data subjects, both positive and negative, is necessary. Finally, appropriate safeguards like encryption, pseudonymization, transparency, and opt-out options must be implemented to protect patient data.

How does the GDPR define ‘scientific research,’ and what are its implications for using patient data in AI systems? What does Recital 159 suggest about the interpretation of ‘scientific research’? What does the European Data Protection Board (EDPB) recommend regarding ethical and methodological standards?

The GDPR doesn’t explicitly define ‘scientific research,’ but Recital 159 suggests a broad interpretation that includes technological development, fundamental and applied research, and both privately and publicly funded studies. Consequently, AI research for scientific purposes is often covered under this broad definition. The EDPB recommends that scientific research adhere to established ethical and methodological standards to qualify as compliant. This means that the AI system development should follow strict research methods or protocols.

Despite the EDPB’s guidance delay, what practices should organizations adopt to ensure GDPR compliance when using patient data for AI training in scientific research?

In the absence of specific guidance from the EDPB, organizations should adopt thorough compatibility assessments and ensure robust safeguards are in place. They should not assume automatic compatibility but instead evaluate each case individually. Implementing strong data protection measures and documenting their compliance efforts is also essential to meeting GDPR obligations.

What are the potential outcomes of a compatibility assessment for secondary use of patient data? What steps must be taken if the secondary use is found to be incompatible with the initial collection? How should organizations proceed if the secondary use is compatible with the initial collection? What documentation is required to meet the accountability principle?

If a compatibility assessment finds the secondary use incompatible, the organization cannot proceed unless it obtains explicit consent from data subjects or relies on specific legal provisions. If the secondary use is deemed compatible, the organization can rely on the initial legal basis but must still comply with other data protection principles. Organizations must inform data subjects about the further processing and their rights, and conduct a Data Protection Impact Assessment (DPIA) if needed. All steps taken must be documented to fulfill the accountability principle.

Regarding Article 9 of the GDPR, what exceptions allow for the processing of health data for secondary use in AI systems? What role does explicit consent play in processing health data for secondary use? How can reasons of public interest in public health justify secondary use? What are the requirements for scientific research exceptions under Union or Member State law?

Article 9 of the GDPR specifies exceptions for processing sensitive health data. Explicit consent from data subjects is one such exception. Public interest in public health, under Union or Member State law, can also justify secondary use. Furthermore, scientific research exceptions require adherence to legal standards and appropriate safeguards to ensure protection.

What are some best practices for organizations to ensure GDPR compliance while using patient data for AI system development and training in scientific research?

Best practices include conducting detailed compatibility assessments, implementing strong data protection measures such as encryption and pseudonymization, and ensuring transparency with data subjects about how their data will be used. Organizations should also document their processes to demonstrate accountability and compliance with GDPR.

How can organizations inform data subjects about the potential secondary use of their data?

Organizations can inform data subjects through clear and concise privacy notices at the time of data collection, outlining the possibility of secondary use. They should update data subjects on any changes and provide options for opting out if they do not consent to the secondary use of their data.

What challenges do organizations face in conducting compatibility assessments for secondary use of patient data in AI research, and how can these challenges be addressed?

Challenges include interpreting the principles of compatibility, ensuring robust safeguards, and balancing the benefits of scientific research with the rights of data subjects. These can be addressed by keeping up with evolving regulatory guidance, investing in data protection technologies, and maintaining transparency and communication with data subjects to build trust.

Do you have any advice for our readers?

My advice is to stay informed about data protection laws and invest in strong data governance practices. Understanding and implementing GDPR principles, performing thorough compatibility assessments, and ensuring transparency with data subjects will help organizations navigate the complexities of using patient data in AI research.

Subscribe to our weekly news digest.

Join now and become a part of our fast-growing community.

Invalid Email Address
Thanks for Subscribing!
We'll be sending you our best soon!
Something went wrong, please try again later