Artificial intelligence and medical research databases: ethical review by data access committees

Area of Science:

• Bioethics
• Data Science
• Artificial Intelligence Ethics

Background:

• Ethics review committees (e.g., IRBs) face challenges in evaluating big data and AI research due to novel ethical considerations.
• Existing committees may lack expertise in assessing collective risks and benefits or may inappropriately exempt de-identified data research.

Purpose of the Study:

• To address the ethical review weaknesses in big data and AI research, particularly concerning de-identified data sharing in medical research databases.
• To propose data access committees as a viable alternative for ethical oversight.

Main Methods:

• Analysis of ethical issues in de-identified data sharing within medical research databases.
• Evaluation of the capabilities and limitations of data access committees in ethical review.
• Proposal for enhancing the ethical review function of data access committees.

Main Results:

• Ethics committees exhibit functional weaknesses in reviewing big data and AI research, especially concerning de-identified data.
• Data access committees possess relevant technical expertise and governance knowledge for overseeing big data projects.
• Data access committees can effectively conduct ethical reviews if they integrate diverse ethical expertise.

Conclusions:

• Data access committees are well-positioned to conduct ethical reviews for medical research databases.
• Enhancing data access committees' review capabilities requires incorporating professional and lay ethical expertise.