Data Mining Unlocks Hidden Resources for Physicians on Rounds

Informaticists with iPads are accompanying doctors on rounds at Georgetown University Medical Center, and they're using complex search technology to improve patient care.

Doctors making rounds at Georgetown University Medical Center have a new digital health tool at their disposal that can speed up diagnoses and treatment plans, improve clinical outcomes, and even perk up patient engagement.

Rounding teams at the Washington, D.C. hospital can now include an informaticist armed with an iPad and AI-based text-mining technology that scans medical literature for the latest information on patient care, including disease symptoms, medications, and suggested treatments. The resource gives physicians an instant, on-demand link to clinical decision support while they meet with patients.

"Doctors are busy on rounds and often don't have the time to look things up at all," says Jonathan Hartmann, director of clinical information services and data discovery at the health system, which has been partnering with Linguamatics since 2012 on text mining capabilities.

The support addresses several care gaps for physicians, whether they're checking up on patients in the hospital's general population or treating critically ill children in the pediatric ICU. Just as important, it allows them to access information with minimal interruption to the conversation they might be having with the patient, the patient's family, or others attending rounds.

"Their time is limited," says Hartmann. "They have quite a lot to do without looking anything up."

But an informaticist with an iPad changes that dynamic. Hartmann and his colleagues are far more skilled at digging up data than the traditional doctor, who might pull out his smartphone or tablet and type a word or phrase into a traditional medical search engine. An AI-supported platform moves beyond the simple keyword search, traveling down alleyways of relationships and side roads of context, going far beyond the description of a disease to pull up symptoms, side effects, and associated research.

Healthcare organizations have been excited about the prospect of AI for many years, particularly in automating and improving the clinical decision support process. Physicians are under constant pressure to stay on top of their game—with patients' lives in the balance. It's estimated that medical knowledge now doubles every 73 days, and those coming out of medical school now will have learned only about 6% percent of what they need during those first three years.

Forget the good old days of the kindly doctor dispensing homegrown advice with a lollipop or George Clooney look-alikes solving a medical mystery in the nick of time after a night in the library. Physicians need a resource that can allow them on-demand access to the knowledge they need to treat the patient in front of them. And while many can now pull out the smartphone and access a CDS platform on the fly, they still need to know what they're looking for.

Hartmann says Georgetown University Medical Center may be the only healthcare system in the world sending informaticists on rounds with doctors (a practice that has been curtailed by the pandemic). He notes that physicians often have their own unique way of searching for things, or they favor one search engine over another, while informaticists are focused on the search itself.

"We have expertise in searching and can do complex searches quickly," he says.

The value is there, particularly in the pediatric ICU. Hartmann says there have been instances where these searches have helped to save lives, and one physician in particular has credited the process with helping him treat a young patient with three different health concerns, a challenging task when taking into account how treatments and medications might interact.

The potential for this platform is even bigger. AI-enhanced data mining technology is only now being integrated into the electronic health record system, giving physicians and researchers access not only to literature but unstructured data contained in clinical notes and scientific papers, even—potentially— information stored on medical devices, wearables, and mobile platforms. The platform might then be used not only by physicians treating their patients, but by researchers looking to identify genes associated with certain diseases or pharmacists tackling the molecular challenges or drug targeting and repurposing.

The challenge, of course, is in cost. There's a reason Georgetown University Medical Center is unique in sending informaticists out on rounds. Most health organizations can't afford to dedicate specialists to those duties or free up the technology for clinical care. In time, it's expected that AI and data mining will become more fully integrated in the EHR, as the technology becomes more malleable and commonplace.

For now, data mining and AI are more commonly associated with research. Hartmann says the examples of the technology saving lives or dramatically improving treatment are few and far between.

"Not many people realize the value of it," he says. "They do not realize how [this] could help physicians."

And that's where the technology needs to go next. Hartmann envisions that integration into the EHR, perhaps through an "info button" that a doctor simply pushes on a tablet or laptop to launch the process.

"We have all of this information available," he says. "The big problem is getting the knowledge into the physician's head quickly."