News: FDA authorizes first-ever AI sepsis diagnostic tool

CDI Strategies - Volume 18, Issue 15

Last week, the FDA granted marketing authorization of an artificial intelligence (AI) software tool for rapid diagnosis and risk assessment of sepsis, Prenosis announced in a press release. The AI sepsis diagnostic is called “the Sepsis ImmunoScore” and uses a combination of 22 biomarkers and patient parameters to provide a risk score and four discrete risk categories that indicate a patient’s risk of deterioration, in-hospital mortality, and escalation of care within 24 hours of patient assessment, MedPage Today reported.

The FDA authorized the tool with a “De Novo Classification,” meaning that the device provides reasonable assurance of safety and effectiveness for its intended use, but for which there is no legally marketed predicate device. It was designed using a biobank and dataset consisting of more than 100,000 blood samples from over 25,000 patients. In a 2021 study published in the ASCPT Journal, researchers collected samples from 1,400 adult patients in emergency departments suspected of sepsis to evaluate the machine-learning algorithm and found the tool accurately identified sepsis and other related secondary outcomes.

“Given that diagnostic performance alone does not address the entire dilemma facing healthcare providers today, we now present a more holistic analysis of a subset of NOSIS, a large, multi-center, novel data set comprised of three plasma proteins (procalcitonin [PCT], interleukin-6 [IL-6], and C-reactive protein [CRP]) and routinely measured EMR parameters,” researchers explained in the study.

They found a higher score accurately predicted less favorable outcomes regarding discharge time, 30-day mortality, and 30-day inpatient readmissions. Predicted risk scores for patients were higher for those with septic shock and infection and organ dysfunction compared with those with sepsis but without those conditions.

Editor’s note: To read MedPage Today’s coverage of this story, click here. To read the ASCPT Journal study, click here.

Found in Categories: 
News