Workflow, communication key to success of predictive analytics
Because predictive analytics in healthcare is relatively new, its success is often determined by its impact on provider workflow, a panel of experts who participated in FierceHealthIT's executive breakfast panel discussion "Using Predictive Analytics to Improve Care & Efficiencies" in New Orleans last week said.
Julio Silva (right), vice president of clinical systems and CMIO at Rush University Medical Center in Chicago, discussed how his facility accomplished just that--requiring that their physicians wear sensors over a 6-month period that mapped their use of ancillary tools and checked their utilization of test times for certain disease types.
The data gathered, he said, helped Rush to improve their clinical processes by creating alerts tailored to fit in with provider workflow.
"You have to incorporate data [as a part of that workflow] Silva said. "Providers want acceptable alerts, not interruptions."
Benjamin Horne (left), director of cardiovascular and genetic epidemiology at Salt Lake City-based Intermountain Medical Center, talked about his experience both in creating predictive models for providers to use and in convincing those same providers of their importance. He echoed Silva's thoughts about interruptions, but also said that once you convince physicians of the importance of such tools, they can't live without them.
"You can't do enough to get information to them fast enough," Horne said.
Overcoming early mistakes in the process also is crucial to success, according to Tina Buop CIO at Oakland, Calif.-based La Clinica de la Raza (right). She said that initially, her team assumed it had all the right tools because it was looking at internal clinical data.
"Use of predictive analytics isn't just about point-of-care information, though," she said. "It's also about identifying other trends so you can say 'in six months, you'll be here.'"
That's where communication comes in, according to Kaveh Safavi, M.D., managing director for Accenture Health Practice (left). He said organizations that want to use such tools should be building a culture around testing and management, but often fail to go from "insight to action" due to a lack of communication.
"Predictive analytics represent a great opportunity right now in healthcare," he said, "but we're still really in a state somewhere between adolescence and a mid-life crisis.
"Our conversations tend to lock us into clinical data," Safavi continued. "But you can't build systems by only looking at information from one or two sources; you need more than clinical data."
He added that perhaps patient reported or socially reported data could be an answer.
"The reality is, we need to understand the healthcare ecosystem as it is taking shape around us," said Matt Siegel, vice president of strategy and corporate development at Verisk Health (right). "And we need to be really clear about what we're trying to accomplish," before embarking on such efforts.
How predictive analytics data helps your patients--and your business
Rush CIO Lac Tran on using predictive analytics to improve outcomes, efficiency
Predictive tool helps ID at-risk trauma patients
Predictive tool pinpoints heart patient readmissions
Online nurse support improves vascular disease risk factors
Tool predicts survival odds for cardiac-arrest patients
Hospitals cut readmissions with predictive modeling