Artificial Intelligence for sepsis prediction: A review of real-world clinical implementation.

Authors

Aadith Ashok, Nenad Macesic, Anton Y Peleg

Published in

Clinical microbiology and infection : the official publication of the European Society of Clinical Microbiology and Infectious Diseases. Nov 10, 2026. Epub Nov 10, 2026.

Abstract

Artificial intelligence (AI) tools have the potential to significantly improve sepsis care by analysing numerous variables in a timely and cost-effective manner. With appropriate training and fine-tuning, AI systems can support dynamic sepsis risk prediction and may facilitate more individualised clinical assessment and management. However, their generalisability across settings and effectiveness in improving patient outcomes remain to be determined.
This narrative review draws on contemporary literature to describe the clinical implementation of AI sepsis prediction tools, their impact on patient outcomes and sepsis process measures, and the key socio-technical barriers and solutions for translating these tools into clinically reliable and scalable systems to improve sepsis care.
PubMed, Web of Science, Scopus and IEEE were searched for studies on the clinical implementation of AI sepsis prediction tools. We included studies published 2019-2026 that describe or evaluate model impact on patient outcomes or process measures.
Implementation studies show mixed results at improving patient outcomes such as mortality and length of stay, but adoption of AI tools has shown improvements in process measures such as time to antibiotic administration. In addition to the indeterminate clinical impact, the generalisability and scalability of AI sepsis tools have been limited due to several other factors, including heterogeneity of data, alert fatigue, insufficient clinician trust, ethical concerns, algorithmic bias, and uncertainty for clinicians around how to interpret and respond to model generated predictions. We also report a minimum outcome set that future implementation studies should pre-specify and report on.
Successful integration into clinical practice requires a comprehensive implementation strategy that shifts the research and design focus from improving model accuracy to one that prioritizes real-world evaluation, human-centred design and addresses workflow, training and equity concerns.

PMID:
43008935
Bibliographic data and abstract were imported from PubMed on 11 Nov 2026.

Read full publication at:
Please sign in to see all details.

Stats

1-terrible, 9-excellent. How would you rate this publication? Sign in in to submit your rating.

Post a comment

You need to be signed in to post comments. You can sign in here.

Comments

There are no comments yet.