AI Doctor

CHAPTER 12Which Health AI Applications Are Ready for Their Moment?

A KEY (AND VERY LEGITIMATE) question about artificial intelligence (AI) in Healthcare is whether it's ready for prime time.

If all of the hospitals, physicians, nurses, patients, and insurance companies lined up and said “give me AI now,” could the applications we've discussed deliver consistently and with good form? That's the question we'll tackle in this chapter. Is the use of natural language processing (NLP) on clinical notes for point‐of‐care decision support ready to deliver consistently if used in the management of cancer patients? Is the use of deep learning to find ischemic stroke on computed tomography (CT) scans of patients in the emergency room (ER) consistently and accurately predictive? Will decision support models developed with large language models be useful to physicians at the point of care?

In the end, the path for the adoption of AI in healthcare will be through the in‐demand use cases where AI performs at an acceptable level. That intersection tells us where we'll see the initial wave of application adoption and, as the needs of users evolve and the technology matures, we'll see the next use cases gain traction. Professor Pete Szolovits, Head of the Clinical Decision‐Making Group at MIT's Computer Science and Artificial Intelligence Laboratory, told me that one of the key barriers will be the lack of cross‐functional expertise in medicine and computer science under the same roof. This ...