Anthony Chang, MD, chief intelligence and innovation officer of Orange, Calif.-based Children's Hospital of Orange County, told Becker's he uses ChatGPT in the clinic "every hour."
The pediatric cardiologist cited the example of a patient who had started on an antiseizure medication he wasn't familiar with. So he asked ChatGPT more about its mechanics and interactions. He double-checked its answer but said it was "spot on."
"It's really, really going to change the face of clinical medicine in a very big way," he said. "When these large language models combine with all of medical literature, all of electronic records and all of medical imaging, to essentially build this big portfolio of medical knowledge, it's going to be the smartest doctor on the planet."
Since he became the nation's first health system chief intelligence officer, in 2015, Dr. Chang first witnessed a gradual growth in healthcare artificial intelligence — then an explosion of it in more recent years.
He had returned to school four years earlier, to get a master's degree in biomedical data science and AI from Palo Alto, Calif.-based Stanford University School of Medicine. He may have been ahead of his time in that regard as well, as he said only two of 100 students in the program were clinicians.
He wanted to understand both the clinical and AI sides and how they intersect.
"I'm a violinist, and I occasionally get to help conduct the orchestra if the conductor is late," he said. "And I always feel like when you're a musician and a conductor, you feel the music a little bit differently than if you were just a conductor who was not a musician. I want to feel that intellectual exercise that programmers do."
But he expects more clinician data scientists in the years to come. He estimates there are only about 100 of them now, but he said "tens of thousands" are needed to really make an impact on the field.
"A cohort of duly trained and educated physician data scientists will be really important to drive the agenda of what questions are important, and what policy should be," he said. "As I always say, you can't put a neurologist and a general surgeon in the operating room and expect them to do neurosurgery."
He also predicts that by the end of the decade, there will be more than 100 people in positions like his at health systems around the country. "This is a paradigm that's not going to go away," he said. "It's going to be coupled with a myriad of issues like ethics and regulatory and legal and equity. This is important enough that it will need a separate person and a team."
He also envisions AI continuing to be used to improve efficiencies on the nonclinical side of hospital and health system operations, with data employed for resource allocation and staffing decisions, on top of reducing infections, managing surges and predicting adverse outcomes in patients to lower risks. It will also play a big role in extended reality, digital health and wearables.
"This pandemic really showed that we were totally inadequate in solving some of the decades-old problems in healthcare," he said. "For things like scheduling and allocating resources and figuring out what to do with overflow, those big problems cannot be handled by humans and their brains alone anymore. And shouldn't be. We've stressed a lot of health professionals out of their jobs, basically. So we need help, and this is a really great resource and it's coming in a very timely fashion."
He also anticipates a coming "revolution" in medicine and medical education from generative AI.
"It's super information," he said. "Knowledge is what you do with the information, which is still a human domain. This is information packaged in such a way that you can really be smarter and have better knowledge. So it's the ultimate source of information. But we do have to be careful because it's not foolproof.
"In a busy practice, I see 12 to 15 kids in half a day. I don't have the kind of time just to sit back and say, 'Let me look this up and let me research this.' I call this like the stinger missiles in a war conflict. It gives me a weapon I can use to hopefully make a difference with a patient's outcome."
Dr. Chang hopes to be an inspiration for younger physicians and medical students to become "bilingual" in medicine and data science. He jokes that he speaks the language of AI "with a heavy accent" because he learned it later in his career and at a time when it hadn't quite matured.
"I'm very excited for the future generation of clinicians who can speak this language but also for patients to finally get the precision care and the population health-focused solutions that we all need," he said. "Without AI, I don't see another solution on the horizon. So that's why I'm very excited about this."