Congressional Perspective: How AI Can Improve Health IT

There are both promises and challenges in applying breaking technology to federal health IT, noted Rep. Michael Burgess during the Digital Health CXO Tech Forum Tuesday.

In reviewing the ongoing revolution in electronic health records, Burgess looked back on his experience as a physician to recognize the prior challenges in consolidating relevant health information from patients who have been admitted to various regional hospitals.

“Now with electronic health records, that problem doesn’t exist anymore,” Burgess noted of the oft-fragmented nature of paper records.

Still, this has left remaining questions around regulation — especially as relating to patient privacy. The greatest quandary in this area, Burgess detailed, is navigating regulatory laws that may vary across the state and federal level.

“One of the big things we’re grappling with, and I suspect will be grappling with for some time, are the issues around privacy and the fact that different states are coming up with different privacy regiments,” he said.

This is especially complicated for health networks that span multiple states and would require considerable financial and human capital investment to adopt a wholesale electronic health records system. The possibility of an eventual federal standard for EHR privacy is another roadblock in the digital transition process.

“Before I make that investment, before I do that time and training, are you going to do something at the federal level?” Burgess asked.

Another challenge Burgess discussed were coalescing concerns around both privacy and liability, especially who bears responsibility in the case of a health data breach.

“The question of liability remains, but privacy is still the biggest obstacle,” he said.

Despite the questions around EHR privacy legislation and broader implementation, Burgess remained optimistic about the possibility for artificial intelligence to improve patient care for some of the most widely managed conditions.

“I think of a disease like type I diabetes, and it is one that lends itself to artificial intelligence as one of its principle management tools,” he said. “It’s a minute-by-minute protocol, and artificial intelligence is probably better suited to monitoring that because you’re not going to have your doctor testing your glucose literally minute-by-minute all through the day.”

Still, Burgess recognized that medical applications of artificial intelligence will require legislative oversight of their own — a pattern that will follow for all emerging technologies that are applied to patient care. This will require consistent attention on the part of lawmakers, as well as a certain degree of technical literacy on Capitol Hill.

“There are clearly things we haven’t even thought of that will be problems, and the policymaker will of course still have to be involved,” he concluded.