Across industries, CIOs cited artificial intelligence as the most "problematic" technology to execute, followed by digital security and the internet of things, according to recent Gartner survey.
This sentiment is widespread among healthcare C-suite leaders. While hospitals are collecting massive amounts of data to improve patient safety and identify care inefficiencies, this information is often deposited in disparate dashboards. Not only do healthcare administrators and frontline clinicians lack time to analyze cumbersome data reports, but most executives view machine learning technologies aimed at organizing data — like AI — as mere hype rather than practical solutions.
Demystifying AI can help hospitals and health systems express their data in a frictionless manner, Mudit Garg, CEO and co-founder of Qventus, said during a Dec. 5 webinar sponsored by Qventus and hosted by Becker's Healthcare. Qventus is a data analytics company offering hospitals and health systems AI and machine learning solutions to improve operations.
Transforming AI from a buzzword into a functional solution first requires defining AI, Mr. Garg said.
"Fundamentally, AI is where intelligence can be exhibited by machines. If there is a goal we want to achieve and a machine can solve it — and it can maximize its own chances at achieving that goal — that's AI," Mr. Garg explained.
AI is different than machine learning, Mr. Garg continued. Machine learning is an application of AI comprising algorithms, with algorithms' performances improving over time with exposure to more data. AI is a program that can sense, reason, act and adapt to help clinicians make the best decisions.
What an AI and machine learning-equipped solution can do for hospitals is aggregate and read disparate data to identify patterns and predict the likelihood of certain outcomes. These AI solutions are "like an air traffic control" center, Mr. Garg added, supporting decision-making across a system, from improving patient flow in ambulatory settings to enhancing patient safety.
Mr. Garg pointed to El Camino Hospital, a 433-bed facility in Mountain View, Calif., as a forerunner in AI use among hospitals and health systems. El Camino Hospital turned to AI because of a problem with patient falls, Cheryl Reinking, RN, chief nursing officer of the hospital, told webinar attendees. While El Camino Hospital collected fall data in its Epic EHR system, that data became static over time.
"It's not continuously updated to push out information to our clinical staff regarding changes in the patient condition," Ms. Reinking explained. The system didn't "allow care plans and interventions to be changed," she added.
Within three months of implementing an AI machine learning tool, Ms. Reinking said El Camino Hospital witnessed a 39 percent reduction in falls. This resulted from using data, which was gathered from disparate silos in El Camino Hospital's EHR, as the basis for forming real-time messages to clinicians about patients' changing fall risk.
In addition to improving patient safety, AI and machine learning can help hospitals identify opportunities to improve operational efficiencies nonmoving, dashboard-based data can't, according to Karim Botros, senior vice president and chief strategy and innovation officer at MetroHealth in Cleveland. MetroHealth's frustration with a high no-show rate among its ambulatory network — where 19 percent of nearly 1.3 million patients treated annually failed to show up — drove the system toward an AI solution.
"Our traditional way of dealing with this was to overbook 19 percent," he said. "On a daily level, that could work across all sites. But for a specific provider, who had 10 patients guaranteed to show, an additional two patients is frustrating for everyone involved. For another provider with 10 patients that never show up, overbooking by two patients" didn't solve the issue.
To address this inefficiency, MetroHealth began piloting an AI and machine learning tool Nov. 8 to improve primary care scheduling. The program mapped which patients were most likely to not show up for an appointment when accounting for life changes, past behavior and other outside information. After a month, MetroHealth's main campus saw an 82 percent accuracy rate when predicting expected no-shows. This allowed MetroHealth to fill appointment slots with new patients in less than 24 hours. The pilot will last another couple months.
Two reasons why MetroHealth is prioritizing AI and machine learning are reimbursement headwinds and rising labor costs, Mr. Botros said. These operational factors are also pressuring El Camino Hospital to prioritize AI, as is direct demand from patients and payers for AI and machine learning capabilities.
"Our customers are demanding more efficiency. They want to interact with a healthcare organization that is frictionless, where they can depend on efficiency and high quality. Our payers are also demanding that," Ms. Reinking said. "Strategically, it's very important for our organization to stay viable in the marketplace. We need to be able to offer healthcare that our community members want to use. Creating such processes is just the right thing to do."
Click here to view the webinar recording.
To view the webinar slides, click here.