National Center for Foundational Artificial Intelligence for Rehabilitation (FAIR Center)

Artificial intelligence (AI) is transforming society by enabling advances in computer vision, natural language processing, and several areas of biomedical research. Some modern AI models have become so powerful they have been dubbed “Foundation Models.” Unfortunately, rehabilitation researchers and people with conditions that limit mobility have yet to see much benefit from modern AI, and no foundation model exists for rehabilitation. Our Center for Foundational Artificial Intelligence for Rehabilitation (the FAIR Center) will establish a vital research program to enable rehabilitation scientists to apply state-of-the-art AI to diagnose, monitor, and improve the outcomes of rehabilitation. We have created a large-scale, high-quality dataset of movements and rehabilitation outcomes, the FAIR Dataset, and tools to automatically integrate data from many research studies, which is vital for the field and the research we propose. The FAIR Center will: 1. Develop and validate a Foundation AI model for Rehabilitation (the FAIR Model) and use the model to address important rehabilitation research questions. The FAIR Model will be trained and tested with our FAIR Dataset and leverage a state-of-the-art generative machine learning architecture. We will apply the model to (a) develop a video-based metric to predict ACL injury, (b) personalize gait retraining for individuals with knee osteoarthritis, and (c) predict surgical outcomes for children with cerebral palsy. 2. Disseminate, support, and enhance the FAIR Dataset and FAIR Model through easy-to-use web interfaces, training materials, and open-source code. 3. Engage thousands of rehabilitation scientists, engineers, and people with lived experience in a community that uses data science for rehabilitation research via training and community-building programs. 4. Establish a cohesive, vibrant, and sustainable Medical Rehabilitation Research Center through the leadership of an experienced executive team, external advisors, and people with experience living with mobility-limiting conditions. By providing high-quality software, data, and AI models, the FAIR Center will enable collaboration of unprecedented scale between bioengineers, clinicians, computer scientists, people with lived experience with mobility-limiting conditions, and others focused on rehabilitation. Our training efforts will create a new generation of rehabilitation scientists who are fluent in the strengths and challenges of AI. Our Center will be run by a tightly integrated clinical and engineering team, enabling us to appreciate the goals of people with lived experience, recruit participants to our studies, and rapidly create and share valuable new technology. Together with the FAIR Center community, we will achieve the potential of AI to understand and improve human movement, and increase ability for people with osteoarthritis, cerebral palsy, and many other conditions.