Illinois Health System Implements AI-Based Sepsis Surveillance Tool

Illinois-based Blessing Health System has integrated an artificial intelligence (AI)-based sepsis surveillance tool into its clinical workflows, enabling the health system to identify patients with sepsis or septic shock more quickly than with standard methods.

Sepsis is a major challenge for most hospitals and health systems, affecting 1.7 million US adults each year, and killing 270,000, according to a Centers for Disease Control and Prevention (CDC) report. Of patients who die in hospitals, one in three has sepsis. Sepsis can also progress quickly to septic shock, resulting in severe organ problems and death.

Many patients who survive severe sepsis and septic shock will recover completely, but some people experience permanent organ damage, immune system disruption, or cognitive issues, in addition to potentially being at a higher risk for certain medical conditions.

These potential outcomes make effective sepsis surveillance and management a key part of supporting clinical decision-making and improving care. To support these goals at its hospitals, Blessing Health implemented an AI tool known as POC Advisor.

According to the press release, POC Advisor utilizes real-time clinical and EHR data to determine patient sepsis status. When sepsis is detected, alerts are routed directly to a central monitor, into the EHR, or to other point-of-care devices, which allows clinicians to act in a timely manner. The tool also utilizes natural language processing (NLP) to pull critical information that may indicate sepsis from clinical notes within the EHR.

"Early detection of sepsis is a challenge for many health systems, and we were determined to tackle this through more proactive monitoring," said Mary Barthel, MD, chief quality and safety officer at Blessing Health System, in the press release. "Using POC Advisor, we can now deliver real-time patient status updates to our clinicians so they can intervene immediately with an evidence-based approach to help save more lives and improve our sepsis bundle compliance. By combining advanced technology with our sepsis coordinator on the ground, we are better equipped to identify patients who develop sepsis earlier and optimize their care."

POC Advisor has been implemented at two of Blessing Health’s hospitals. It is the latest addition to the health system’s clinical surveillance suite, including infection prevention and pharmacy surveillance tools.

This deployment is part of a larger effort to leverage AI to forecast sepsis and sepsis-related outcomes.

In July, researchers from the University of Florida and the University of Washington developed an AI tool capable of identifying a patient’s likelihood of developing sepsis within 12 hours of hospital admission.

AI has also been used to predict more severe outcomes such as sepsis mortality.

In April, a study showed that machine-learning (ML) algorithms outperform conventional methods for predicting sepsis mortality rates using administrative data. Similar research last month indicated that ML models can accurately predict 30-day mortality in patients with sepsis-associated encephalopathy.