“The true gift of the wayfinder’s journey is not arrival at a destination; it is who we become along the way as we fulfill our potential.”


This Viewpoint piece from the Journal of American Medical Association focuses on the next-generation artificial intelligence for diagnosis, and specially addresses “wayfinding”. The term wayfinding refers to the act of finding one’s way to a particular destination or, in short, navigation. The authors explain that there are several elements of wayfinding: orientation, path selection, route monitoring, and destination recognition. Wayfinding is therefore a strategy to reduce the cognitive load in the navigation of a complex journey. 

The only time that I have heard of this term used in healthcare is in the context of hospital wayfinding, an innovative way to improve navigation in a health system that is mainly focused on the patient experience. The authors’ contention of introducing wayfinding in artificial intelligence and diagnostic tests is an interesting concept: patients present with signs and symptoms leading to information gathering/organization/prioritization, information integration/interpretation, and formulation of next steps in the workup. 

These authors suggest shifting the role of diagnostic artificial intelligence from prediction of labels to wayfinding (defined as interpreting context and providing cues that guide the diagnostician). However, it may be a bit unrealistic to expect such lofty cognitive elements for the current artificial intelligence in medical diagnostic testing. The current artificial intelligence in healthcare is at the relatively early Renaissance period of music, and will need time to mature into a more sophisticated strategy with wayfinding (think Beethoven’s Ninth Symphony or at least Mozart’s Clarinet Concerto in A major, K.622). 

The authors are absolutely spot on in that their wayfinding expectation of artificial intelligence is what eventually will need to be in place for any clinician. It is certainly possible that the right wayfinding strategy can reduce the number and increase the appropriateness of diagnostic tests in the first place. This navigation strategy of a continual and dynamic refinement process for diagnosis, however, does mandate a close synergy between the artificial intelligence and clinical medicine domains with a cohort that possesses both domain expertise to facilitate this cognitive collaboration as well education of both domains for everyone. 

Read the full paper here.