Collaborative AI for precision neurorehabilitation: a roadmap

 You're putting the cart before the horse. AI doesn't become useful until 100% recovery protocols are created! Using AI to deliver the failed status quo faster doesn't get survivors to full recovery!

Collaborative AI for precision neurorehabilitation: a roadmap

• Sook-Lei Liew,
• R. James Cotton,
• Etienne Burdet,
• Sergi Bermúdez i Badia,
• Pablo Celnik,
• James H. Cole,
• Ruishan Liu,
• Surjo R. Soekadar,
• Carolee Winstein &
• Nicolas Schweighofer 

We are providing an unedited version of this manuscript to give early access to its findings. Before final publication, the manuscript will undergo further editing. Please note there may be errors present which affect the content, and all legal disclaimers apply.

Abstract

Precision rehabilitation seeks to improve care for individuals by identifying personalized treatments that enhance both the efficacy and efficiency of care. Rapid advances in artificial intelligence (AI) and data-driven methodologies stand to greatly enhance precision rehabilitation methods, particularly when paired with clinical expertise and patient-driven goals. Although AI has been successfully integrated into healthcare in fields such as precision oncology, the dynamic, multi-faceted, complex nature of rehabilitation require a different approach than standard predictive models. Here, we argue that precision rehabilitation may benefit from collaborative AI, in which humans and AI work together with AI distilling complex data that humans can use as part of their nuanced decision-making process. We review the current landscape of precision rehabilitation and explore how collaborative AI, and AI in general, can advance the field. We begin by outlining a roadmap for a general collaborative AI system in rehabilitation, noting four key challenges that must be addressed. Next, we examine three existing precision rehabilitation frameworks that have been developed independently by several of the co-authors, which share four common key elements of the roadmap. We then describe examples of how AI has already been applied to specific aspects of complex rehabilitation scenarios, as these may be integrated into larger collaborative AI models. We follow this with a discussion of how to make AI-based precision rehabilitation a reality, with an emphasis on the large data required for these models to make accurate predictions, as well as potential ethical issues. Finally, we conclude with recommendations for future directions. Ultimately, collaborative AI promises to transform rehabilitation by leveraging vast, diverse datasets to create individualized digital profiles, which can first be used to simulate the effects of rehabilitation interventions in silico, maximizing impacts before real-world implementation. By merging personalized rehabilitation strategies based on empirical evidence with the depth and complexity of clinical knowledge and reasoning, collaborative AI holds promise as a powerful tool to help clinicians advance patient care and improve long-term outcomes.