Surgical site infections affect between 160,000 and 300,000 patients each year. Using machine learning tools integrated with Epic, University of Iowa Hospitals & Clinics reduced surgical site infections by 74% during a three-year pilot study.
Because interventions to reduce infection risk during wound closure can be costly and invasive, targeting interventions toward patients for whom the benefits outweigh the risks is key.
While a patient is still in surgery, machine learning algorithms analyze the patient’s risk for a surgical site infection based on historical patient data and procedure documentation in Epic. If a patient is at risk for infection, a warning appears and suggests follow-up actions to the provider, such as negative wound pressure therapy, that can be taken during wound closure. The machine learning system then tracks the patient outcome and any intervention performed to show the effects of intervening for high-risk patients.
Selectively using these interventions makes it possible to “maximize the therapeutic effect, while minimizing the cost and potential risks to patients,” said Dr. John Cromwell, associate CMO and director of surgical quality and safety at UI Hospitals & Clinics.
Read the full article at Healthcare IT News.