By Diagnostics World News Staff
December 12, 2024 | Sepsis ImmunoScore, the first FDA-authorized AI diagnostic for sepsis, has high accuracy for the diagnosis of sepsis and for prediction of key adverse patient outcomes, reports a paper published last week in the New England Journal of Medicine AI (NEJM AI; DOI: 10.1056/AIoa2400867). In the paper, the authors write that the test showed both diagnostic and prognostic power.
Sepsis, an overactive immune response to infection, costs the U.S. healthcare system billions of dollars annually and claims the lives of at least 350,000 people each year—more than all cancers combined. The Sepsis ImmunoScore was built using Prenosis' Immunix platform, which enables precision medicine in acute care. Prenosis received FDA-authorization via the de novo pathway for Spesis ImmunoScore in April 2024.
While single biomarkers, such as Procalcitonin (PCT), have been valuable tools in diagnosing sepsis for over two decades, they have limitations when trying to predict the future. The study demonstrated the combination of multidimensional inputs—vital signs, lab data, and sepsis-specific biomarkers—into a comprehensive AI-driven tool provides more reliable diagnostic information and is predictive of future patient outcomes.
This study looked at all of the critical outcomes that Sepsis ImmunoScore predicts simultaneously: accuracy for sepsis diagnosis and prediction of critical outcomes, including mortality during hospitalization, length of stay before discharge, ICU admission, need for mechanical ventilation, and use of vasopressors.
"The Sepsis ImmunoScore is the first FDA-authorized sepsis AI algorithm that functions by embedding in the EMR and is approved to aid providers in the identification and assessment of patients with sepsis," said Nathan I. Shapiro, MD, MPH, a Professor of Emergency Medicine at Harvard Medical School and attending physician in the Department of Emergency Medicine at Beth Israel Deaconess Medical Center, and the paper's corresponding author in a press release.
"This study confirmed that AI, when validated, can be a reliable, comprehensive tool to assist providers with identification of patients at high risk of sepsis and prediction of sepsis-related adverse outcomes," said Akhil Bhargava, the study's lead author and Prenosis' Senior Data Scientist — Applied Machine Learning, in the same press release.
In the paper the authors highlighted several limitations to the study including findings that may not generalize to populations that differ from the five hospitals used in the study; missing data in the EMR or the use of ICD-10 codes that may have led to misclassification of certain elements such as comorbidities; and the fact that the study was observational and did not assess the impact of the Sepsis ImmunoScore on clinical decision-making or changes in therapeutic approaches. Further work is warranted, the authors note, to further establish the test’s generalizability to other settings.
However, the authors confidently conclude, “The Sepsis ImmunoScore demonstrated high accuracy for the identification and prediction of sepsis and critical illness metrics that could enable prompt identification of patients at high risk of sepsis and adverse outcomes, potentially improving clinical decision-making and patient outcomes.”