‘Computer, how is this patient doing?’ Using analytics in health care

Correction: This story has been updated to correct the spelling of Onur Torusoglu’s name.

St. Luke’s Health System is looking at new ways to use data analytics to help produce better patient outcomes.

“It’s not magic, it’s math,” said Onur Torusoglu, chief digital and analytics officer at the Boise-based nonprofit health organization.

Consolidation

With analytics employees sprinkled across St. Luke’s, the organization has recently consolidated the 80 to 90 employees who can use technologies such as SQL, R and Python to create analytic algorithms, Torusoglu said.

Altogether, St. Luke’s has 60 technology systems in a centralized repository using a cloud-based data warehouse to make the data available to people who can use it to help make business decisions, Torusoglu said.

Computers can constantly monitor patients in the hospital setting and alert staff if patients are developing a problem such as sepsis, Torusoglu said. Clinicians define thresholds to determine which patients are at risk, he said.

“A number of data points are updated on an almost instantaneous basis,” he said. “We capture key variables and analyze that on the fly. We can identify patients before they turn septic, which is a tremendously helpful tool for our physicians.”

While St. Luke’s has had the capacity in its electronic health records system to do this for some time, the organization has been developing this project for the past 12 to 18 months.

“We took our time to build the right teams and skill sets,” Torusoglu said. “We’re dealing with human lives. We didn’t want to just turn on this capability and see what happens.”

For example, if the thresholds weren’t set correctly, they could alert staff too often – or not soon enough.

“That fine-tuning is the tricky part today, to set those thresholds correctly so they serve patients and clinicians in the correct manner,” Torusoglu said.

Artificial intelligence

Other systems are using artificial intelligence and machine learning, Torusoglu said.

“We have an ability to utilize analytics in an oncology setting to help match patients with a given trial or other active cases we have,” he said. “Instead of sorting manually through hundreds of thousands of different oncology trials, we can directly match patients to trials on a much faster basis.”

Analytics are also beneficial after a patient is discharged, Torusoglu said.

“We have resources around case and care notification settings that would allow us to create follow-up conversations with patients, social workers or case managers” for issues such as medication adherence, he said.

Studies bear fruit

St. Luke’s isn’t alone. A number of health care organizations are looking to analytics to help create better patient outcomes. Analytics are associated with a higher chance of successful patient outcomes, according to a recent study, “How Successful Are Healthcare Organizations at Achieving ROI with Analytics?” by HIMSS Analytics, sponsored by  Dimensional Insight, a vendor in this field.

The organizations surveyed 109 senior health care executives involved with analytics and decisionmaking in clinical analytics, financial analytics and operational analytics. The survey found that organizations saw the most success with clinical analytics (77.7%) compared with financial analytics (73.5%) and operational analytics (70.3%). In particular, organizations see the highest measured success rate (75.4%) if they use clinical outcomes improvement as their primary metric, the survey found.

However, there are barriers toward implementation of analytics in health care, according to a 2018 study, “The Opportunities and Challenges of Data Analytics in Health Care,” by the Brookings Institution, a Washington, D.C.-based nonprofit public policy organization.

“These barriers include the nature of health care decisions, problematic data conventions, institutionalized practices in care delivery, and the misaligned incentives of various actors in the industry,” the study found.