At the recent Digital Health Investment Symposium, I had the opportunity to learn about topics like precision medicine, and facilitate a panel on artificial intelligence (AI), one of the hottest topics in healthcare IT. This new technology has generated both awe and frustration over the years, and I was eager to pick the brains of some subject matter experts.
Not So Fast
I introduced one of my burning questions by bringing up a man known as one of the fathers of AI: Herbert A. Simon. In 1957, Simon famously predicted that within 10 years, a computer would be able to routinely beat the human chess champion. Well, 1967 came and went, and humans were still on top. In the end, it took nearly 40 years for Simon’s vision to come to pass.
This story illustrates our tendency to overestimate the power and potential of innovation in the short term. This seems particularly true when it comes to AI. How often do we see an article in the news about something else AI can do, only to read further and learn that the technology isn’t ready for general adoption?
Despite some unfulfilled AI dreams, many of us can’t help but get excited at what it might someday do for many industries, including healthcare. In this spirit, I invited our panelists to step into Simon’s shoes. My question: “If you had a crystal ball and could see three to five years into the future, what AI use cases would you expect to be working and making an impact?”
The symposium attendees and I were extraordinarily lucky to hear from the two panelists in the room. One was Lonny Northrup, Senior Health Informaticist at Intermountain Healthcare, an enterprise and process architect who worked for years on implementing data innovations. The other was Jason Wiesner, Medical Director at Sutter Health. A diagnostic radiologist by training, Wiesner also co-founded teleRAS, a 24/7 subspecialty teleradiology company. I knew that between the two, we would get an accurate and exciting picture of the future of AI.
As most providers can testify, there’s a lot of work to be done on things as simple as fundamental workflow issues. “We’re overwhelmed with unread worklists and more patient records than we can read,” said Wiesner, “and there are even more demands on radiologists and other folks who interpret medical imaging.”
Wiesner is confident AI will be able to help fairly quickly. He anticipates a tool that will be able to sort through patient information, pinpoint the patients most likely to be abnormal, and triage accordingly. “Some people would call those innovations table stakes,” Wiesner said, “but those tools are desperately needed today, and they could be adopted like wildfire.”
Northrup said he believes AI will have the greatest impact in population health and patient engagement. He described tools that could suggest to doctors which actions would be best for their patients, both clinically and socioeconomically, which could prevent conundrums with medical devices and directly help patients. “We all want to be healthier, but we don’t all do the things that will help us to be healthier,” Northrup pointed out. “These solutions can actually help us to do things we haven’t been successful at doing.”
In describing the work Intermountain has done with medication adherence for complex cardiovascular patients, he pointed out that the technology “connects the patient to their care team 24/7,” and helps them “receive care in the ways they want to receive care.” These tools can also be used for patients with type 1 diabetes and are being developed for those with COPD and other conditions. “The results are patient satisfaction scores from 85 to 100 percent, reduced readmissions, improved hospital and clinical utilization, improved clinical outcomes, and reductions in HbA1c results, hypertension rates, and more,” he noted.
While most of the technology currently used may not facilitate provider-patient interactions until weeks after a patient leaves the hospital, Intermountain looks forward to AI-inspired patient engagement tools that can help manage care immediately. “As soon as a patient walks out the door, the clinical team will receive meaningful feedback,” Northrup said. Getting information on a daily basis about whether patients are taking medications, in addition to measures like their weight, can enable providers to “intervene much more quickly and increase patient wellness.”
If you’d like to learn more about the near future of AI and how this technology can improve healthcare, stay tuned for another post about findings from the panel.