Cytovale, maker of the FDA-cleared rapid sepsis diagnostic IntelliSep, presented new real-world data at the Society of Critical Care Medicine’s 2026 Critical Care Congress showing a 19% relative reduction in sepsis-related mortality after the test was rolled out across four additional hospitals in the Franciscan Missionaries of Our Lady Health system. The one-year pre- and post-implementation review covered more than 34,000 emergency department patients with suspected infection, making it one of the larger real-world datasets yet published on an AI-assisted sepsis diagnostic.
An Eight-Minute Test Built on Machine Learning
IntelliSep is not a blood culture or a biomarker panel in the traditional sense. The device uses microfluidics to physically deform white blood cells drawn from a small blood sample, then applies a machine-learning algorithm to interpret how those cells behave under mechanical stress — a signal Cytovale says correlates with the body’s immune response to infection. The result is delivered in about eight minutes, far faster than the hours it can take to get back culture results or full lab panels. The FDA authorized IntelliSep as what the agency described as the first sepsis diagnostic specifically cleared for use in the emergency department.
Why Speed Matters in Sepsis Care
Sepsis remains one of the leading causes of in-hospital death in the United States, and clinical guidelines emphasize that mortality risk climbs with every hour treatment is delayed. But sepsis symptoms overlap with many less dangerous conditions, which pushes emergency physicians toward two costly defaults: over-triaging patients who don’t have sepsis into aggressive treatment pathways, or under-triaging patients who do have it because early symptoms look mild. Cytovale’s pitch is that a fast, objective read on immune response can help sort patients into the right pathway sooner, before culture results come back.
The New Numbers
Across the four expansion sites, the study found a 19% relative reduction in sepsis-related mortality and a 22% relative mortality reduction among patients initially suspected of sepsis who ultimately turned out not to have it — a group that can still suffer harm from unnecessary aggressive treatment. The data also showed a 20% increase in emergency department discharges without a corresponding rise in seven-day return visits, suggesting clinicians felt confident sending lower-risk patients home rather than admitting them defensively. On the operational side, blood culture use dropped 30%, concentrated among patients ultimately found not to have sepsis, while cultures were still ordered in more than 90% of confirmed sepsis cases.
Consistency Across Hospital Sizes
What Cytovale and outside observers flagged as notable is that the results held up moving from large tertiary referral centers to smaller community hospitals within the same health system — an important test for any diagnostic tool, since technologies that work well in academic medical centers with abundant specialist staff don’t always translate to smaller facilities with different patient mixes and staffing levels. Cytovale has framed the multicenter, multi-site consistency as validation for its broader commercialization push beyond its original pilot sites.
Vendor-Funded Data and Open Questions
As with much of the sepsis-AI evidence base, the Franciscan Missionaries data was generated in partnership with a health system that had already adopted the technology and was presented by the manufacturer at a clinical conference rather than published as an independent, peer-reviewed multicenter randomized trial. Systematic reviews of AI sepsis-prediction tools published earlier in 2026 have generally found that machine-learning models outperform older clinical scoring systems on discrimination metrics, but researchers have also cautioned that pre-post implementation studies — where a hospital compares outcomes before and after adopting a new tool — can be confounded by other simultaneous changes in staffing, protocols or patient populations. Cytovale is not alone in this space: competitors including Prenosis, which also holds FDA marketing authorization for an AI-powered sepsis risk tool, are pursuing similar real-world evidence strategies.
What’s Next
Cytovale’s next challenge is scaling beyond health systems that adopted IntelliSep early and demonstrating the same mortality and resource-use gains in more varied hospital settings, ideally validated by researchers outside the company. With sepsis diagnostics increasingly viewed as a proving ground for whether AI-assisted tools can move beyond incremental workflow improvements to measurable reductions in mortality, 2026’s growing body of real-world data — vendor-generated or otherwise — is likely to shape how quickly emergency departments elsewhere adopt similar rapid, AI-interpreted tests.
