The IRB Role & Practical Tools for IRB Review of Research Proposals Involving Artificial Intelligence (AI)

AI models, and Machine Learning (ML) in particular, are excellent tools that can be used by a learning healthcare system to immediately impact healthcare delivery and patient outcomes. To achieve high impact, researchers must be able to effectively communicate to ethics professionals the essential components of the specific system, so that IRBs (and IACUCS when such algorithms are used in animal research) can accurately calculate the risk/benefit profile of the model, and thus document their criteria for approval. However, this task is daunting at best, and often what is offered by way of system description is beyond the scope of most IRB members' expertise. There is some guidance from federal agencies, but additional discussion is necessary to identify true risk and possible benefit of such models so that the field can assess appropriate review pathways for AI and ML-centered research. Through in-depth discussion and case study review, this presentation will offer tools to identify AI/ML models used in research and their risks and benefits, and present practical strategies for regulatory review so that such research is implemented with the highest of ethical standards and the precision of good science. Case studies will be discussed to allow attendees to use materials later for training IRB members. Learning Objectives: Discuss various AI models and technologies that employ them Examine strategies to use when reviewing protocols with AI and ML components Identify risks and concerns regarding privacy and confidentiality Gain an understanding of US, EU and other international AI ethics standards that apply to IRB review The views presented here are those of the speakers. They do not necessarily reflect those of any organisation. Francis gratefully acknowledges the work Ukraine Clinical Research Support Initiative (UCRSI), of the International Science Council’s CODATA International Data Policy Committee (IDPC), and the EOSC-Future / RDA Artificial Intelligence & Data Visitation Working Group (AIDV-WG). Francis’ EOSC Future / RDA Ambassador for Ethics & Law and the EOSC-Future projects are co-funded by the European Union’s Horizon Programme call INFRAEOSC-03-2020 - Project ID 101017536. CODATA is funded by the International Science Council (ISC) and Horizon Europe as well as by other funding organisations. The Research Data Alliance (RDA) is funded by Horizon Europe as well as by other funding organisations (see here). Francis is grateful for the support provided by CODATA and the EOSC-Future and the RDA communities, structures, and secretariats