Samaritan Hospitals Launch New Tool to Prevent Acute Kidney Injuries June 12, 2025 Samaritan Health Services is implementing a new alert system designed to help protect patients from serious kidney injuries during their hospital stay. This innovative tool, which launched during the first week of June, uses a predictive model to identify which patients might develop a hospital-acquired acute kidney injury (AKI) – a potentially reversible condition where damaged kidneys can increase the risk for other medical conditions and increase the risk of mortality. How Does the Tool Prevent Kidney Injuries? The predictive model continuously monitors hospitalized adult patients, except labor and delivery patients or patients with existing kidney disease. It analyzes 45 different data points including blood test results, vital signs and current medications. Every 15 minutes, the model calculates each patient’s risk of developing a hospital-acquired AKI in the next 24 hours, providing health care teams with real-time risk assessments. When clinicians order certain medications that can potentially harm the kidneys for patients categorized as high risk by the model, an alert (called an OurPractice Advisory) will appear. The list of medications that can trigger the alert was created by a group of Samaritan pharmacists, nephrologists and critical care clinicians. When the alert appears, clinicians can choose to prescribe an alternative medication or proceed with the original treatment plan if they determine that the therapeutic benefits outweigh the potential risks. While medication is a common contributor to AKI, the model that drives the alert takes a myriad of other factors that may increase the risk of AKI, into consideration, such as lab results and vitals. During testing, the alert demonstrated impressive accuracy and a relatively low activation rate. At Good Samaritan Regional Medical Center, the alert triggered approximately five times per day, and in 80% of those instances, patients either already had AKI or developed it later. At our other hospitals, which typically care for fewer, the alert activated less frequently—about once per day. The system also effectively minimizes false alarms, with only 3% of patients who did not develop kidney issues receiving unnecessary alerts. “This predictive model, along with the associated alert, is an example of how advanced data science tools can be employed at the point of care to help inform clinical decision making,” said Liam Finlay, director of Business Intelligence and Data Services. It is anticipated that this tool will decrease the incidence of hospital-acquired AKI. Additionally, it has the potential to provide other benefits such as reducing length of hospital stays and decreasing mortality. This was a collaborative effort with immense support from the HA-AKI Workgroup, including Wendy McCallum, Jim Herman, Tyler Mann, Jan Hull, Kristin Moser, Rachelle Collier, Dr. Brian Delmonaco, Dr. Tomer Pelleg, Dr. Blessing Osondu, Daniel Sherman, Dr. Tyler Andrea, Dr. James Crane, Dr. Divine Ribakare, Jackie Chandler, Charlene Gutt, Debbie Culley and Paula Stahl.
