By Jenna Larson
With its aim to be one of the nation’s best, the University of Iowa Hospitals and Clinics continues to receive national recognition.
A group at the UIHC was one of 20 teams to win the Computerworld Data+ Editor’s Choice Award, according to a recent press release.
Each year in the spring, Computerworld, an online magazine geared toward information-technology professionals, names roughly 20 teams nationwide as winners of the award.
“The award launched in 2013,” Computerworld Editor Ellen Fanning said. “We showcase organizations that are using data to analyze and predict trends.”
When the nominations opened in the spring, Computerworld editors and data specialists had to narrow the nominations to around 20 teams, Fanning said.
“We ask companies to submit a case study and to outline what their project is all about and how they were able to analyze trends,” she said.
Fanning said the UI team won the award because of its interest in creating a program that would reduce patient infection through data intake and prediction.
“The main thing that caught the judges’ eyes was the metrics and how they were able to reduce infections,” she said. “That really put them in the finalist category because it showed that they understood the mechanics.”
UI Clinical Associate Professor John Cromwell, the UIHC director of surgical quality and safety, worked with several colleagues to create the database.
“Surgery patients often have an issue in which many days after surgery, their intestines shut down and stop working,” he said. “Several years ago, we began working on a device that will tell us within a few hours after surgery which patients are going to develop this problem days later.”
Cromwell said he and his team used the device to focus on bigger problems in medicine, including hospital-acquired infections.
He and his collaborators wanted to know if they could identify the patients from the operating room to see who is at risk for infections through data available before and during the time the patient is in the operating room, he said.
In designing the data program, he said, he and his team trained the machine to know potential outcomes and protocols using available data from the hospital.
“We develop the algorithm, train [the machine] based on historic data, and then we take that and deploy it,” he said.
When a new patient is coming through surgery, the device will indicate whether the patient will develop a surgical-site infection, Cromwell said.
“What we did that was unique here is that this is the first time this has ever been used in the setting of the operating room,” he said.
He and his team are very proud of this accomplishment, especially because in the course of only three years, surgical-site infections have been reduced by 74 percent using the algorithm.
Jose Monestina, a data warehouse applications developer at the UIHC, helped Cromwell support the collection and storage of the data from the electronic medical records in order to design the algorithmic program.
There were a lot of hurdles that had to be dealt with in order to prevent exposure of data to the public, Monestina said, and dealing with the protected health-care information was difficult because the UIHC has to maintain Health Insurance Portability and Accountability Act regulations.
“Our hopes for this project is to open eyes of others and serve as an example that it is possible to use [electronic records] combined with predictive analytics to predict outcomes and improve health care,” he said.
One of the great things about this system is that as time goes on and patients change, the machine is able to adapt and recalibrate to learn the changes, Cromwell said.
“It is the evolution of machine learning into the health-care environment,” he said.