While generative artificial intelligence will "completely" reshape primary care, there's still a "huge gap between hype and reality" for healthcare AI, a Mount Sinai Health System leader told Becker's.
For instance, most primary care visits will one day be transcribed by AI, which will know who's talking and put it into context via the patient's EHR before automatically creating a summary, said Thomas Fuchs, chair of AI and human health at the New York City-based health system.
"All conversations will be AI-supported in the future. Primary care will change completely," he said. "But I think that's clear to see for everyone already."
Mount Sinai is piloting generative AI for chatting with patients, triaging messaging for physicians, and creating safe chat environments for providers, he said. The health system uses a HIPAA-compliant GPT model from Microsoft.
It is also employing large language models to extract patient phenotypes from deidentified EHR data for research in areas such as genomics and pathology, as well as teaching medical students about their optimal usage.
But while healthcare AI has "enormous potential," there is also a significant amount of noise to sort through, Dr. Fuchs said. He gave the example of pathology, which has only one AI solution that's been approved by the FDA.
"So if you contrast that with the hype and the 12,000 health tech startups and so forth, there's a huge gap between hype and reality," he said. "There's, of course, a long way to go. We're just scratching the surface with what's possible with generative AI and large language models. But we are, as an institution, committed to use it where it makes sense."