The integration of AI into all aspects of society is moving at lightspeed and is certain to impact the ethical review of clinical research. Currently, there are no federal regulations specific to AI for Institutional Review Boards (IRBs) to follow, but the development of guidance and best practices is likely to evolve over the next few years as IRBs encounter AI applications in all aspects of protocol review. Many research projects that include AI may be exempt from IRB review, either because the study involves secondary use of data without interaction with a human subject or the data is de-identified.
However, The Secretary’s Advisory Committee on Human Research Protections (SACHRP) recently opined on this issue and noted that the “AI/ML [machine learning] and BD [big data] research expose the limits of the traditional concept of identifiability that serves as the basis for privacy protections under the Common Rule.” As technology advances, what was considered private in the past may no longer be, so IRBs may need to evaluate the data sources used in research more closely and consider whether adequate protections are in place before establishing that a protocol is exempt from IRB review.
Once a protocol is determined to require IRB review, there are a variety of areas where AI might be incorporated into the research. For example, AI may be incorporated into software as part of a medical device. It may be used as part of drug and biologics development. It could be used to help select what participants might be enrolled in a clinical trial, or it may be integrated into digital technologies as a study outcome measure.
The IRB is tasked with evaluating how AI will impact the IRB’s analysis of benefit and risk, particularly harm related to the validity and bias of models used to train AI, as well as how the privacy of the individual will be protected in the research. The IRB is also tasked with ensuring that the informed consent form and process includes information on the use of AI as part of the research if AI is being used to make treatment decisions during the study and is clear on any secondary use of data after the study is complete. IRBs may consider having an individual with expertise in AI/ML available for consultation for complex protocols utilizing AI to ensure that participants are adequately protected.