Singh, M.D., is the chief health AI officer at UC San Diego Health.
In the 1960s, Dr. Warner Slack built the first computer capable of asking patients about their medical history. Surprisingly, when it came to directly interviewing patients, the computer turned out to be more accurate and detailed than physicians. In the face of this technological breakthrough, Dr. Slack was asked whether computers would replace doctors. His response: “Any doctor who can be replaced by a computer deserves to be replaced by a computer.”
Computers have changed health care in many ways. The medical questions you answer using your phone today probably contain some of the same questions Dr. Slack’s research participants answered on a vacuum-tube-powered computer in the 1960s. Today, this information is a core part of the way health care is delivered.
ADVERTISEMENT
Artificial intelligence is on a similar path. AI didn’t enter health care until recently, especially a modern form of it known as deep learning. Many of the advancements over the past decade have focused on making it more useful. It has only recently begun to find its way into clinical practice, although its adoption is growing quickly. AI is already playing a key role in making health care safer and more personal.
Today, if you arrive at UC San Diego Health with a fever and an elevated heart rate, an AI tool monitoring emergency room patients will alert your medical team that you may have sepsis, a potentially deadly reaction to a bloodstream infection. Since rolling out this sepsis tool, more patients have received timely treatment for sepsis, resulting in fewer deaths. This adds to a growing body of research from around the country showing AI’s potential to save lives.
AI is also helping make health care more personal. While computers have made care more accessible, our reliance on them means that clinical staff spend more time typing on a computer and spend less time face-to-face with patients. This problem was the focus of the book “Deep Medicine” by Dr. Eric Topol, executive vice president of Scripps Research and another pioneer in the use of AI in health care. While many solutions have been thrown at this problem, generative AI may be able to help.
Generative AI is a type of AI focused on generating text, images and video. Popularized by OpenAI’s ChatGPT, there are now several variations of this technology, many of which you can download and run on your own computer. UC San Diego Health was among the first health systems to give doctors the ability to use generative AI to draft messages to their patients. While initial feedback from patients has been positive, the jury is still out on whether the drafts are good enough to be a useful starting point. Later this year, we are launching an evaluation of ambient AI scribing tools that use audio from visits, with permission, to draft the documentation normally prepared by hand. Early research from The Permanente Medical Group shows that AI scribing tools can save time spent typing, with some patients reporting better interactions with their physicians. But work still needs to be done on how to support the tools’ use without over-relying on them. And there are many uses where generative AI simply hasn’t been explored, such as in the review of clinical records to identify barriers to care or preventable medical errors. We are actively evaluating these uses in the hopes of making care safer and more equitable.
At UC San Diego Health, the Joan and Irwin Jacobs Center for Health Innovation aims to be the hub that will incubate uses of AI across the health system. We will do this in coordination with the health system’s robust AI governance process to ensure that AI is used ethically for the communities we serve. UC San Diego Health has publicly committed to this as a signatory on a U.S. White House statement on responsible AI. We are also participating in national partnerships to ensure that we share with and learn from others.
Even though computers have not and may not replace doctors, they have brought us into the AI era. If Dr. Slack were with us today, I’d tell him that any health care task that can be made safer with AI deserves to be made safer with AI.
Editor’s note: This OpEd was originally published in the San Diego Union Tribune.
