Résumé
This whitepaper outlines the development of a novel, streamlined, and compliant Institutional Review Board (IRB) review process for AI Human subjects Research (AI HSR). Our proposal centers on collaborating with cross-functional teams across the institution in the risk-to-benefit ratio assessment of AI tools and their risk mitigation efforts. This includes establishing a shared platform and common language around human subjects research and AI risks. By aligning IRB risk-to-benefit assessment with the institution’s broader stakeholders and ancillary committees, we ensure that AI technologies are integrated sustainably, consistently, and ethically.
A fundamental premise of this approach is that scientific evidence is the foundation for trust in healthcare. As clinicians begin developing AI tools for research and their practice, the healthcare organizations for which they are employed may take on the role as legal manufacturers of AI technologies. Therefore, institutions must work closely with their IRBs to ensure risk management activities are applied appropriately and in a timely manner.
Healthcare providers rely on scientifically validated research to make informed medical decisions, whether a human or a machine makes those decisions. Such decisions directly impact the health and well-being of individuals and their families. These healthcare providers making their products available inside and outside their institution could harm their own and other patients if their innovations are not properly developed and risks mitigated. Ensuring trust in healthcare requires that the innovators in research hospitals adhere to rigorous scientific and ethical standards during the development of digital health tools. Therefore, under the oversight of an IRB and in close collaboration with the broader Human Research Protection Program, a systematic risk management framework is applied in the evaluation of AI tools under development.