Jove
Visualize
Contact Us
JoVE
x logofacebook logolinkedin logoyoutube logo
ABOUT JoVE
OverviewLeadershipBlogJoVE Help Center
AUTHORS
Publishing ProcessEditorial BoardScope & PoliciesPeer ReviewFAQSubmit
LIBRARIANS
TestimonialsSubscriptionsAccessResourcesLibrary Advisory BoardFAQ
RESEARCH
JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchive
EDUCATION
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualFaculty Resource CenterFaculty Site
Terms & Conditions of Use
Privacy Policy
Policies

Related Concept Videos

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by
Same author

Leveraging the Electronic Medical Record for Functional Neurological Disorder: A Scoping Review.

The Canadian journal of neurological sciences. Le journal canadien des sciences neurologiques·2026
Same author

Experiences of racially, ethnically, sexual- and gender-minoritized adolescents and young adults with cancer in the Canadian health care system.

Supportive care in cancer : official journal of the Multinational Association of Supportive Care in Cancer·2026
Same author

Survey of Intensivist Coverage and Interprofessional Team Staffing in Canadian ICUs.

Critical care medicine·2026
Same author

Interprofessional Staffing in Pediatric Critical Care: Secondary Analysis of a National Cross-Sectional Survey From 2022 to 2023.

Pediatric critical care medicine : a journal of the Society of Critical Care Medicine and the World Federation of Pediatric Intensive and Critical Care Societies·2026
Same author

High school graduation among adolescents born preterm: A Canadian population-based cohort study.

Early human development·2025
Same author

Plasma transfusion practice: A five-year audit of plasma transfusion at 23 hospitals.

Transfusion·2025

Related Experiment Video

Updated: May 21, 2025

A Data-Driven Approach to Quantifying Immune States in Sepsis
07:42

A Data-Driven Approach to Quantifying Immune States in Sepsis

Published on: February 7, 2025

130

Unsupervised clustering for sepsis identification in large-scale patient data: a model development and validation

Na Li1,2,3, Kiarash Riazi4,5, Jie Pan4,5

  • 1Department of Community Health Sciences, Cumming School of Medicine, University of Calgary, CWPH 5E34, 3280 Hospital Dr. NW, Calgary, AB, T2N 4Z6, Canada. Na.Li@ucalgary.ca.

Intensive Care Medicine Experimental
|March 20, 2025
PubMed
Summary

This study introduces a novel unsupervised clustering method for identifying sepsis cases, improving epidemiologic surveillance. The approach effectively identifies sepsis phenotypes using electronic health records and administrative data.

Keywords:
Adult sepsis eventClusteringElectronic health recordsSepsis identificationUnsupervised learning

More Related Videos

Evaluation of a Reliable Biomarker in a Cecal Ligation and Puncture-Induced Mouse Model of Sepsis
05:28

Evaluation of a Reliable Biomarker in a Cecal Ligation and Puncture-Induced Mouse Model of Sepsis

Published on: December 9, 2022

3.3K
Design of Cecal Ligation and Puncture and Intranasal Infection Dual Model of Sepsis-Induced Immunosuppression
07:30

Design of Cecal Ligation and Puncture and Intranasal Infection Dual Model of Sepsis-Induced Immunosuppression

Published on: June 15, 2019

9.8K

Related Experiment Videos

Last Updated: May 21, 2025

A Data-Driven Approach to Quantifying Immune States in Sepsis
07:42

A Data-Driven Approach to Quantifying Immune States in Sepsis

Published on: February 7, 2025

130
Evaluation of a Reliable Biomarker in a Cecal Ligation and Puncture-Induced Mouse Model of Sepsis
05:28

Evaluation of a Reliable Biomarker in a Cecal Ligation and Puncture-Induced Mouse Model of Sepsis

Published on: December 9, 2022

3.3K
Design of Cecal Ligation and Puncture and Intranasal Infection Dual Model of Sepsis-Induced Immunosuppression
07:30

Design of Cecal Ligation and Puncture and Intranasal Infection Dual Model of Sepsis-Induced Immunosuppression

Published on: June 15, 2019

9.8K

Area of Science:

  • Medical informatics
  • Epidemiology
  • Machine learning

Background:

  • Sepsis poses a significant global health challenge, complicated by the lack of a definitive reference standard for case identification.
  • Existing methods for sepsis identification, including administrative codes and electronic health record (EHR)-based algorithms like the Adult Sepsis Event (ASE), have limitations and contribute to variable incidence rates.
  • Clinician identification of sepsis at the bedside remains challenging, underscoring the need for improved identification strategies.

Purpose of the Study:

  • To develop and evaluate a novel approach for identifying sepsis phenotypes using unsupervised clustering methods.
  • To analyze the characteristics of identified sepsis clusters and compare them with existing criteria (ASE).
  • To assess the potential of this new method for enhancing sepsis epidemiologic surveillance.

Main Methods:

  • A retrospective cohort study utilized hospital administrative and EHR data from adult intensive care unit (ICU) admissions across five Canadian medical centers (2015-2017).
  • Data reduction techniques were applied to 592 variables, followed by the application of eight clustering algorithms to 55 principal components.
  • The Robust and Sparse K-means Clustering (RSKC) method was selected based on optimal clustering metrics, and cluster membership was validated using an XGBoost model.

Main Results:

  • The RSKC method identified 48 distinct hospitalization clusters in the development cohort (3660 patients).
  • Eleven 'ASE-majority' clusters, comprising 22.4% of patients, contained 77.8% of all patients meeting ASE criteria for sepsis.
  • A significant proportion (34.9%) of patients not meeting ASE criteria within these majority clusters met more liberal sepsis criteria, indicating potential under-ascertainment by current methods.

Conclusions:

  • Unsupervised clustering of large-scale, diverse medical data presents a promising avenue for identifying sepsis phenotypes.
  • This approach offers a potential improvement for accurate and consistent epidemiologic surveillance of sepsis.
  • The identified clusters provide a nuanced view of sepsis presentations, aiding in a better understanding of the condition.