FDA Watch: AI-Based Sepsis Prediction Test Receives FDA Authorization

Despite being one of the most common medical conditions to occur in a hospital setting, patients who contract sepsis are often entering a clinical crapshoot. They can be treated fairly quickly and recover, or they can experience a massive organ failure that leads to death.

The speed of the sepsis diagnosis and delivery of treatment—primarily large infusions of fluid and broad-spectrum antibiotics—is the key. Each hour that treatment for sepsis is delayed increases the risk of mortality by 4 to 9 percent, according to the Sepsis Alliance.¹

Traditionally, sepsis—which is caused by a body’s hypervigilant and dysregulated response to an infection—is diagnosed through a combination of monitoring patient temperature, their level of physical discomfort, and mental decline. But it can manifest in other ways that may be harder to detect, particularly in a facility that may be short-staffed or overworked. As a result, the nationwide mortality rate from sepsis is more than 25 percent, even though that rate can be in the low single digits at some institutions, said Steven Q. Simpson, MD, a critical care physician and professor of medicine at the University of Kansas, in a phone interview with G2 Intelligence.

As previously reported in G2 Intelligence’s Lab Industry Advisor, there has been a rise in the use of artificial intelligence (AI) in diagnostic tests.² Following that trend, the Chicago-based company Prenosis’s method of using AI to more quickly diagnose sepsis has just received De Novo marketing approval from the U.S. Food and Drug Administration (FDA).³

Called the Sepsis ImmunoScore^™, the Prenosis test relies on close measurement of biomarkers and other clinical data to predict the presence of sepsis and its progression, informed by a proprietary biobank of 100,000 blood samples from more than 25,000 patients the company has closely analyzed.³

The company claims that it can diagnose the presence of sepsis within 24 hours of a hospital admission. Many cases are contracted from the hospital setting, with most patients being treated for something else when they contract sepsis. The Prenosis platform can also be integrated into many electronic medical records systems used by hospitals.³

“FDA authorization of a sepsis diagnostic tool with significant predictive power is a landmark event for people that could ever be at risk of sepsis at some point in their lives,” said Prenosis chief executive officer Bobby Reddy, Jr, PhD, in a statement.³

The company has not released any information as to when the test would be publicly available or its cost. Reddy did not respond to emails or a phone call seeking comment.

According to Simpson, who is also chairman of the board of directors for the Sepsis Alliance, the use of AI could make a significant difference in treating the condition. He noted that the sepsis care can be terribly uneven from hospital to hospital, with nearly a third of acute care facilities in the US lacking a formal program to spot and treat sepsis.

“There is no reason why a hospital that’s there to take care of those who might get sick…doesn’t have some sort of a plan for how to treat sepsis early,” he said.

Simpson said he was familiar with the ImmunoScore through his work as senior advisor to the Solving Sepsis initiative created by the Biomedical Advanced Research and Development Authority, part of the U.S. Department of Health and Human Services.

“I think that the best potential for diagnosing and establishing prognostic information about sepsis comes from an approach like Prenosis has taken—measuring both biomarkers and physical response over the course of a hospitalization,” Simpson said.

However, Simpson cautioned that sepsis is like any other medical condition, with detection and mortality rates often varying significantly not only by hospital, but by the race and financial demographics of patients. In other words, poorer patients and patients of color tend to be diagnosed with sepsis at a slower rate than patients who are White and more affluent, thereby negatively impacting their outcomes. While he noted that Prenosis will likely be able to fine-tune its test using data from its biobank, health equity issues should also be taken into consideration.

“AI needs to be incorporated (in sepsis detection), but with all the caveats that we have for any other use of AI,” Simpson said.

References:

1. 1. https://www.sepsis.org/protect-sep-1/

1. 1. https://www.g2intelligence.com/diagnostics-pipeline-more-ai-reliant-platforms-being-approved-by-the-fda/

1. 1. https://prenosis.com/news/prenosis-announces-fda-de-novo-marketing-authorization-of-immunoscore/

1. https://www.federalregister.gov/public-inspection/2024-08935/medical-devices-laboratory-developed-tests

**********

Other tests and related products approved by the FDA in the past month include:

New FDA Approvals 

New CE marks and global certifications

There were no significant EU or global approvals announced during the period.