TREWS

By Bayesian Health
· United States
Bayesian Health — Real-time sepsis alerting integrated into the EHR.
Predictive
Critical Care
Production
Not Formally Regulated
Regulatory status
Not Formally Regulated
Deployment
Production
Clinical role
Adjunct (decision support)
Headquarters
United States
AI type
Predictive

Precise indication

Continuous screening of health record data in adults (18+) from emergency arrival or admission through the acute-care stay. It shows a ‘Sepsis Risk High’ flag in the record to help clinicians detect sepsis within 24 hours; it must not be the sole basis for diagnosis.

Regulatory status

Not directly applicable — a clinical alerting system, not a regulated diagnostic device in the FDA/CE sense.

Clinical evidence

Vendor-supported

Adams et al., Nature Medicine 2022 (founder co-author): 5 US hospitals, 6,877 sepsis patients; alert confirmed within 3 h linked to 3.3-point lower adjusted mortality, not randomised.
FDA 510(k) data: retrospective, 7,732 encounters, 3 US sites; sensitivity 79.4%, specificity 89.5%, 11.7% of flags were true sepsis.

Independent

Schinkel, van der Poll, Wiersinga (Amsterdam UMC), Am J Respir Crit Care Med 2023, commentary: observational design and record-based sepsis labels make the benefit hard to judge; results seem preliminary.
No independent study of TREWS accuracy or patient outcomes outside the developers’ team was found.

Real-world deployment

Deployed at five Johns Hopkins hospitals, with among the strongest quantified clinical-impact figures found in this scan.

Interoperability

European status

Editorial verdict

Limits and uncertainties

– Mortality benefit comes from observational studies by the developers; no randomised trial was found.
– In FDA testing, about 88% of flags were false alarms (11.7% of flags were true sepsis).
– FDA test sensitivity (79.4%) missed the pre-set target of a lower bound above 80%.
– Not validated in children, maternity, recent post-surgery or unstabilised trauma patients.
– No CE marking or European use found; all published data come from US hospitals.

Sources

https://aiweekly.co/ai-use-cases/industry/healthcare Regulator — FDA, 510(k) database entry K250680, Bayesian Health Sepsis Flagging Device, accessdata.fda.gov, April 2026 https://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfpmn/pmn.cfm?ID=K250680 Regulator — FDA, 510(k) clearance letter and summary K250680 (PDF), accessdata.fda.gov, April 2026 https://www.accessdata.fda.gov/cdrh_docs/pdf25/K250680.pdf Database — Innolitics, text of FDA 510(k) summary K250680 (indications, predicate, performance), fda.innolitics.com, 2026 https://fda.innolitics.com/device/K250680 Study (vendor co-authored) — Adams R, Henry KE, … Saria S, Prospective multi-site study of patient outcomes after TREWS, Nature Medicine, July 2022 (Europe PMC) https://europepmc.org/article/MED/35864252 Commentary — Schinkel M, van der Poll T, Wiersinga WJ, AI for early sepsis detection: a word of caution, Am J Respir Crit Care Med, April 2023 (PMC) https://pmc.ncbi.nlm.nih.gov/articles/PMC10111986/

This profile is not medical advice or an endorsement. Hospitals should perform their own due diligence (regulatory compliance, GDPR, cybersecurity, integration, local clinical validation). How we evaluate

At a glance
Vendor
Bayesian Health
Headquarters
United States
AI type
Predictive
Specialty
Clinical role
Adjunct (decision support)
CE marking
Notified body
EU AI Act
Data hosting
Last verified
Reviewed by
How we evaluate

Every profile documents the precise indication, regulatory status, vendor-supported and independent evidence, real-world deployment, limits and dated sources.

A independent systematic review or prospective multicentre study · B at least one independent peer-reviewed study · C peer-reviewed, vendor-supported · D vendor claims only.

Read the methodology →

Spotted an error?

Vendors, authors and clinicians can request a correction. Corrections are dated on the profile.