How systems use AI to treat more patients, increase revenue

Health systems are using artificial intelligence to get patients in and out of the hospital quicker, increase capacity and hone staffing levels.

Cleveland Clinic, OhioHealth and UCHealth are among the many systems using predictive analytics and machine learning to try to run hospitals more efficiently, cut down on unnecessary expenses, increase revenue and improve the patient experience, executives said. Much of the cost savings stem from reallocating nurses to different departments based on demand, and revenue increases come from treating more patients.

Related: Mayo Clinic, Phillips collaborate to speed up cardiac MRI with AI

“Hospitals are trapped in a way of doing most things manually,” said Rohit Chandra, chief digital officer at Cleveland Clinic. “Given the challenges healthcare faces in terms of access, cost and efficiency, health systems need to make sure they are leveraging these technology investments as aggressively as possible.”

The technology is predominantly used to revamp manual, burdensome administrative tasks associated with staffing and scheduling. Health system executives said workforce applications are inherently less risky than clinical uses.

Cleveland Clinic, OhioHealth and UCHealth each used AI to build out virtual command centers, which forecast bed availability, patient demand and staffing.

Cleveland Clinic, for instance, uses Palantir Technologies' software to create a digital replica of its main hospital campus that shows a real-time display of its patients, how many nurses are staffing each unit and operating room staffing schedules. The technology analyzes patient data to estimate patient discharge times, automatically pulls scheduling information to forecast nurse staffing projections and identifies gaps in operating room scheduling.

As a result, Cleveland Clinic has been able to transfer about 30 more patients to its main campus each week, an increase of 7%, Chandra said. In addition, AI has helped improve employee communication, emergency care management and hone staffing strategies, he said.

“Before this tool, we had the equipment equivalent to duct tape and bailing wire,” Chandra said. “The complexity of running a hospital in this day and age means that operations and workflow need to be optimized by algorithms, not by humans.”

UCHealth uses its AI-powered command center to better match providers with patients based on clinician specialties and patients’ care needs, as well as predict capacity levels.

The technology has helped reduce patients' average length of stay at the Aurora, Colorado-based UCHealth University of Colorado Hospital by half a day and boost admissions by 6.4%, said Jamie Nordhagen, senior director of patient flow and capacity management at UCHealth.

Predictive analytics from software developer LeanTaaS have helped appropriately staff hospital departments and expedite the discharge process, said Nordhagen.

The algorithm creates a color-coded system that forecasts which departments will be the busiest over the next 24 hours and offers surge capacity recommendations. The system also lays out how staff roles change as departments get busier.

“That has labor, cost and staffing implications and it means more patients have access to care,” Nordhagen said.

OhioHealth, meanwhile, has used artificial intelligence to improve care coordination and streamline the discharge planning process, said Jean Halpin, chief operating officer of OhioHealth Grant Medical Center and the systemwide vice president of customer experience. For instance, the algorithm notifies staff which paperwork needs to be filled out before a patient is discharged.

The AI software developer Qventus created an inpatient AI-backed tool that uses a machine learning algorithm to estimate when and where a patient should be discharged based on patient data and hospital capacity and detect any gaps in treatment plans.

Columbus, Ohio-based OhioHealth has used the technology at its Grant Medical Center to reduce unnecessary extended hospital visits by 1,400 days, Halpin said. That has led to cost savings, more efficient care transitions across hospital departments and an increase in the number of emergency patients OhioHealth can treat, she said.

“Patients are happier because they are admitted sooner and discharged when they’re ready, and our care teams are able to focus more on patient care and less on administrative tasks,” Halpin said.

Still, there have been some AI implementation obstacles. Health system executives said they had to explain to employees why they were implementing the tools and show them the data was accurate and reliable. An open, transparent dialogue helped overcome some initial skepticism, the executives said.