FDA approves new AI tool for sepsis detection

The US Food and Drug Administration (FDA) has approved the Sepsis ImmunoScore, a diagnostic tool based on artificial intelligence (AI) and machine learning, developed by Prenosis. This is the first approval for an AI tool dedicated to the early detection and prediction of sepsis, a condition resulting from a harmful immune response to infection.

Created by the Chicago-based firm Prenosis, the Sepsis ImmunoScore integrates into hospital electronic health records to provide immediate diagnostic and predictive insights. The tool analyzes 22 biomarkers and clinical data to evaluate the risk of developing sepsis within 24 hours of a patient’s admission to an emergency or hospital setting.

With its capability to generate a risk score and categorize patients into four distinct levels of risk, the Sepsis ImmunoScore aids healthcare providers in making informed treatment decisions swiftly, which is crucial for high-risk patients.

The FDA’s De Novo pathway facilitated this approval, highlighting the tool’s novelty and the moderate risk it presents. This pathway is often used for innovative devices that also offer a new approach to medical treatment or diagnosis.

The urgency for advanced diagnostic tools like this stems from the substantial impact of sepsis, which affects over 1.7 million adults in the U.S. each year, with about 350,000 resulting fatalities, according to the Centers for Disease Control and Prevention (CDC). Prompt diagnosis and treatment are critical, as the mortality rate increases significantly with delays in sepsis detection.

Sepsis diagnosis has historically been complicated due to the non-specific nature of its early symptoms, such as fever and increased heart rate, which can resemble those of other conditions. The ImmunoScore addresses these challenges by providing a quantifiable assessment of a patient’s risk for sepsis, supporting more targeted and timely interventions.

To develop this tool, Prenosis utilized its extensive biobank and dataset comprising more than 100,000 blood samples from over 25,000 patients. This research was crucial in identifying the biomarkers and patient parameters that are predictive of sepsis, enhancing the tool’s accuracy.

The approval of the Sepsis ImmunoScore by the FDA is a notable development in the field of medical diagnostics and represents a significant step towards the application of AI technologies in healthcare.