Reliable, Explainable, Accountable, Safe, Open Network Artificial Intelligence Consortium of Research (REASON-AI)

About This Grant

Background and Significance: Although artificial intelligence (AI) has been developed and applied in medical disciplines and settings for decades, recent acceleration of its development has yielded both attention and greater adoption. AI’s essence is learning from complex data, and applying lessons learned in many ways. AI technology is embedded into diagnostic devices and strategies, but numerous challenges with AI abound. Rapid evolution of the field has challenged legislators, policy makers, and regulators to maintain safety while facilitating innovation. Many AI algorithms and tools have not delivered on the promise of AI-assisted care. This project will create the Reliable, Explainable, Accountable, Safe, Open Network Artificial Intelligence Consortium of Research (REASON-AI CORE). Goals are to create and build a community of practice for sharing recent research, operations, policy developments, opportunities, innovations, and partnerships in AI; to assemble, synthesize, assess, and disseminate evidence about AI and its impact; and to facilitate processes and pathways for translation of AI from research to practice. Strategic Plan: Our key approaches will provide research summit conference series; Cyberseminar series; "Let’s Chat about AI" longitudinal discussion; engagement of interested parties; systematic review; subject-matter expertise consultation; training and professional develoment; guidance for infrastructure, science, & operations; an expert advisory panel meeting series; and selection of pilot projects to advance AI in the VA. Innovation: Our conference series will present new findings and developments. “Let’s Chat about AI” will provide a new forum for discussion. The systematic review will provide a summary of core literature never previously assembled or reported. Pilots will provide the VA with new and essential approaches for research, implementation, iteration, and evaluation. Training will develop the AI workforce. Our hope is that this work will lead to an expansion of AI experience as scientists and partners continue to pursue funding and other resources and support for research and operations as a result, in part, of REASON- AI’s facilitation. The work that REASON-AI will conduct, as described in this proposal, has not been conducted previously in the VA. Coordination, Collaboration, & Partnerships: The primary partner is the VA National Artificial Intelligence Institute, created in 2019. Additional partners will include the VA Informatics and Computing Infrastructure, the Office of Information and Technology, and the Office of the Chief AI Officer. We are also working closely with several QUERI investigators as co-investigators on this project, to provide important connection to QUERI in terms of building impact and pathways to implementation. Management Plan: The Multiple Principal Investigators are Michael Weiner (Contact PI), Michael Matheny, and Matthew Samore. Collaborating sites include academic affiliates (Vanderbilt University and University of Utah), as well as VA sites in Indianapolis, Bedford, and Ann Arbor.