Accurate Detection of Sepsis During ED Triage Using Machine Learning with Clinical Natural Language Processing

Aug 3, 2026

Upload by Jiven

Sepsis is a life-threatening condition marked by organ dysfunction and a leading driver of death and critical illness worldwide. The study evaluates whether emergency department triage can enable accurate sepsis detection through extraction and synthesis of EHR data using advanced machine learning (KATE Sepsis) and clinical natural language processing. The model’s performance is benchmarked against common screening approaches, including SIRS and qSOFA, across multiple retrospective and prospective cohort settings.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages35

Reading Time1 h 28 min

AccessFree to download

Abstract

Background

Method

Result

Conclusion

Background

Clinical challenge of early sepsis identification

Limitations of existing screening methods

Role of machine learning in sepsis prediction

Accurate detection of sepsis at ED triage using machine learning with clinical natural language processing Oleksandr Ivanov, PhD – Mednition Inc., Ukraine Karin Molander, MD, FACEP – Sutter Mills-Peninsula Medical Center, USA Robert Dunne, MD, FACEP – Ascension Health, USA Stephen Liu, MD, FACEP – Adventist Health White Memorial, USA Kevin Masek, MD – San Mateo Medical Center, USA Erica Lewis, MD – El Camino Hospital, USA Lisa Wolf, PhD, RN, CEN, FAEN – Emergency Nurses Association, USA Debbie Travers, PhD, RN, FAEN – Duke University, USA Deena Brecher, MSN, RN, ACNS-BC, CEN, CPEN, FAEN – Mednition Inc., USA Deb Delaney, DNP, MHA, RN, CEN – Mednition Inc., USA Kyla Montgomery – Mednition Inc., USA Christian Reilly – Mednition Inc., USA Corresponding author: Christian Reilly, [creilly@mednition.com](creilly@mednition.com) Abstract Background Sepsis is a life-threatening condition with organ dysfunction and is a leading cause of death and critical illness worldwide. Accurate detection of sepsis during emergency department triage would allow early initiation of lab analysis, antibiotic administration, and other sepsis treatment protocols. The purpose of this study was to determine whether EHR data can be extracted and synthesized with the latest machine learning algorithms ( KATE Sepsis) and clinical natural language processing to produce highly accurate sepsis predictive models, and compare KATE Sepsis performance with existing sepsis screening protocols, such as SIRS and qSOFA. Method A machine learning model ( KATE Sepsis) was developed using patient encounters with triage data from 16 participating hospitals. KATE Sepsis, SIRS, standard screening (SIRS with source of infection) and qSOFA were tested in three settings. Cohort-A was a retrospective analysis on medical records from a single Site 1. Cohort-B was a prospective analysis of Site 1. Cohort-C was a retrospective analysis on Site 1 with 15 additional sites. Result Across all cohorts, KATE Sepsis demonstrates an AUC of 0.94-0.963 with 73-74.87% TPR and 3.76-7. 17% FPR. Standard screening demonstrates an AUC of 0.682-0.726 with 39.39-51. 19% TPR and 2.9-6.02% FPR. The qSOFA protocol demonstrates an AUC of 0.544-0.56, with 10.52-13. 18% TPR and 1.22-1.68% FPR. For severe sepsis, across all cohorts, KATE Sepsis demonstrates an AUC of 0.935-0.972 with 70-82. 26% TPR and 4.64-8.62% FPR. For septic shock, across all cohorts, KATE Sepsis demonstrates an AUC of 0.96-0.981 with 85.71-89.66% TPR and 4.85-8.8% FPR. SIRS, standard screening, and qSOFA demonstrate low AUC and TPR for severe sepsis and septic shock detection. Conclusion KATE Sepsis provided substantially better sepsis detection performance in triage than commonly used screening protocols. Future research should focus on the impact of KATE Sepsis on administration of antibiotics, readmission rate, morbidity and mortality. Keywords: Sepsis; Severe Sepsis; Septic Shock; Emergency Department; Emergency Nursing; Machine Learning; ED Triage; SIRS; qSOFA; Natural Language Processing 1. Background Sepsis is a life-threatening condition with organ dysfunction caused by a dysregulated host response to infection (Singer et al. 2016) and is a leading cause of death and critical illness worldwide (Angus et al. 2001, Vincent et al. 2014, Fleischmann et al. 2015, Liu et al. 2014, Fleischmann-Struzek et al. 2020, Vincent et al. 2020) . The global incidence of sepsis is estimated at 32 million with 5.3 million deaths per year (Fleischmann, 2016) . In the United States alone more than 1.7 million patients are diagnosed with sepsis annually and nearly 270,000 die from sepsis (CDC, 2021) . In 2017, the United States spent over $38 billion on sepsis treatment, making it the most expensive condition to treat ( Liang, 2020) . Patients who survive sepsis often have long-term health and social consequences (Iwashyna et al. 2010) . Despite advances in medical treatment, the reported incidence of s

FAQs

What problem does the study address in sepsis care?

The study targets the difficulty of early sepsis identification during emergency department triage, where symptoms can be subtle and progression can be rapid.

How is KATE Sepsis evaluated in the study?

KATE Sepsis is developed using triage-related patient encounters from 16 hospitals and tested in multiple cohort settings, including retrospective and prospective analyses, and compared with SIRS and qSOFA.

What are the key performance results compared with standard screening and qSOFA?

Across cohorts, KATE Sepsis achieves substantially higher AUC and TPR with lower FPR, while standard screening and qSOFA show lower overall discrimination, particularly for severe sepsis and septic shock.

Aug 3, 2026

Upload by Jiven