FDA authorizes Prenosis software as first AI tool that can diagnose sepsis

The health technology company Prenosis on Wednesday announced its artificial intelligence-powered diagnostic tool for sepsis is the first to receive approval from the U.S. Food and Drug Administration. 

Sepsis occurs when the body has an extreme response to an infection, and it is often deadly. More than 350,000 adults with sepsis die during hospitalization or are discharged to hospice each year, according to the Centers for Disease Control and Prevention. Sepsis is notoriously challenging to diagnose, but early detection is key since patients can deteriorate quickly. 

Prenosis’ tool, called Sepsis ImmunoScore, uses 22 different parameters like temperature, heart rate and cell counts to help clinicians assess a patient’s risk of sepsis, the company told CNBC. Doctors and nurses often have to monitor each of these parameters individually. Prenosis’ solution uses AI to evaluate all of those markers at once. The tool generates an overall risk score and four categories that “correlate to a patient’s risk of deterioration,” according to a release.   

Sepsis ImmunoScore was built using Prenosis’ Immunix platform, which is based on a dataset of more than 100,000 blood samples from 25,000 different patients, the release said. 

The tool is integrated directly into electronic health records where clinicians create and manage patients’ medical records. Embedding Sepsis ImmunoScore within existing workflows means it will be easy to use and access, Prenosis said. Clinicians will also see a display that shows exactly which parameters were used to calculate the overall risk score. 

Prenosis’ solution was approved through the FDA’s De Novo pathway, which means the agency has not authorized anything like Sepsis ImmunoScore before. It’s a big win for the 10-year-old Chicago-based startup.

While Prenosis is the first company to receive FDA approval for its AI diagnostic tool for sepsis, several organizations have built and released similar solutions. For instance, Johns Hopkins University has built an AI system that aims to catch sepsis symptoms more quickly. A 2022 study published in Nature Medicine found that, in severe cases, the university’s AI model detected sepsis an average of six hours earlier than traditional methods.  

Epic Systems, a leading health-care software vendor, has also developed an AI-powered sepsis prediction tool, though the company’s model has been subject to substantial criticism in recent years. Epic’s sepsis model is used in hundreds of hospitals across the U.S., according to a 2021 study published in JAMA Internal Medicine. But researchers found that the model “poorly predicts sepsis,” and that “its widespread adoption despite poor performance raises fundamental concerns about sepsis management on a national level,” the study said. 

Epic pushed back on the findings, and the company published a blog post that said health-care organizations have seen sepsis mortality rates improve because of its technology. Even so, Epic reportedly overhauled its sepsis model the following year in an attempt to improve its performance, according to Stat News. 

Prenosis told CNBC that while it could have gone straight to market with Sepsis ImmunoScore, it did not want to try and sell the product without FDA approval. The company said the tool was ready around three years ago, but it wanted to work with regulators to be conscious of safety concerns and ensure that the technology does not cause harm.

Additionally, the FDA updated its guidance for companies in September 2022 and shared examples of device software functions where it “intends to focus its regulatory oversight.” The agency said software that “analyzes patient-specific medical information to detect a life-threatening condition, such as stroke or sepsis,” falls under this category. 

In other words, the agency suggests companies developing sepsis detection tools should seek official authorization. 

Prenosis said it worked to demonstrate the safety and efficacy of Sepsis ImmunoScore to the FDA for around 18 months. Now that the company has secured the agency’s approval, it will conduct additional studies to demonstrate the tools’ accuracy and impact on clinical decision-making. Prenosis will begin to sell the tool to hospitals in the U.S. and eventually to hospitals across the globe, the company said.