Machine Learning in the ICU: Predicting Mortality in Bloodstream Infections (ICU:Intensive Care Unit) (ICU)

Machine Learning in the ICU: Predicting Mortality in Patients With Carbapenem-Resistant Gram-Negative Bacilli Bloodstream Infections

Using our own patient data, our study aimed to predict mortality that can develop in Carbapenem-resistant Gram-negative bacilli bloodstream infections with a machine learning-based model.

In the intensive care unit, patients with bloodstream infections, both with and without mortality, will be examined retrospectively in two subgroups for comparison.

Study Overview

• Status: Completed
• Conditions: Carbapenem Resistant Bacterial Infection
• Intervention / Treatment: Diagnostic test: Machine Learning to Estimate Mortality

Detailed Description

Carbapenems are one of the last-resort antibiotics used to treat severe infections caused by multi-drug resistant Gram-negative pathogens. Infections with Carbapenem-resistant Gram-negative bacilli (CR-GNB) have become widespread in the past decade, posing serious threats to public health. Carbapenem-resistant Enterobacteriaceae (CRE), Carbapenem-resistant Acinetobacter baumannii (CRAB), and Carbapenem-resistant Pseudomonas aeruginosa (CRPA) top the priority list of antibiotic-resistant bacteria worldwide. CR-GNB causes a broad spectrum of infections, including bacteremia, urinary tract infections, pneumonia, and intra-abdominal infections. Carbapenem-resistant bloodstream infections are a significant cause of morbidity and mortality, and therapeutic options in treatment are extremely limited. By evaluating risk factors in patients monitored in the intensive care unit, scoring systems that can predict prognosis reduce mortality risk by ensuring the early application of effective antibiotics and timely hemodynamic support that are currently in use.

With the accumulation of big data and advancements in data storage techniques, innovative and pragmatic machine learning methods that have entered our lives demonstrate good prediction performance in the medical field. Machine learning-based models developed to predict mortality in patients monitored in the intensive care unit are available in the literature and provide an opportunity for earlier intervention in patients.

Using our own patient data, In the intensive care unit, patients with bloodstream infections, both with and without mortality, will be examined retrospectively in two subgroups for comparison. The investigators aim to predict mortality that can develop in Carbapenem-resistant Gram-negative bacilli bloodstream infections with a machine learning-based model.

• Study Type: Observational
• Enrollment (Actual): 197

Contacts and Locations

Study Locations

• Turkey (Türkiye)
  • Kocaeli, Turkey (Türkiye)
    • Kocaeli University

Participation Criteria

Eligibility Criteria

• Ages Eligible for Study: Adult; Older Adult
• Accepts Healthy Volunteers: N/A

• Sampling Method: Non-Probability Sample

Study Population

All patients who were monitored in our tertiary intensive care unit for six years retrospectively and developed bloodstream infections with Carbapenem-resistant Enterobacteriaceae, Acinetobacter baumannii and Pseudomonas aeruginosa have been included in the study with their personal data anonymized

Description

Inclusion Criteria:

• In our study, patients who were monitored in our hospital's tertiary Intensive Care Unit between June 2017 and June 2023 and developed bloodstream infections with Carbapenem-resistant Enterobacteriaceae, Carbapenem-resistant Acinetobacter baumannii and Carbapenem-resistant Pseudomonas aeruginosa will be retrospectively included.

Exclusion Criteria:

• Patients under the age of 18 and those with infections other than bloodstream infections will not be included.

Study Plan

How is the study designed?

Number of groups / cohorts

2

Cohorts and Interventions

Group / Cohort	Intervention / Treatment
Deceased Patients; Carbapenem-resistant Gram-negative bacilli Blood Stream Infection With mortality	Diagnostic test: Machine Learning to Estimate Mortality; Using deep learning we try to develop an algorithm and anticipate mortality
Surviving Patients; Carbapenem-resistant Gram-negative bacilli Blood Stream Infection Without mortality	Diagnostic test: Machine Learning to Estimate Mortality; Using deep learning we try to develop an algorithm and anticipate mortality

What is the study measuring?

Primary Outcome Measures

Outcome Measure	Measure Description	Time Frame
Risk of Mortality	The sensitivity and specificity will be defined with AUC-ROC curve (Area Under the Receiver Operating Characteristic curve) using machine learning algorithm	3 months

Collaborators and Investigators

• Sponsor: Kocaeli University
• Investigators: Principal Investigator: özlem güler, Kocaeli University

Publications and helpful links

Helpful Links

• Using machine learning methods to predict in-hospital mortality of sepsis patients in the ICU

Study record dates

Study Major Dates

• Study Start (Actual): April 12, 2024
• Primary Completion (Actual): June 28, 2025
• Study Completion (Actual): June 28, 2025

Study Registration Dates

• First Submitted: December 4, 2023
• First Submitted That Met QC Criteria: December 4, 2023
• First Posted (Actual): December 12, 2023

Study Record Updates

• Last Update Posted (Actual): March 11, 2026
• Last Update Submitted That Met QC Criteria: March 8, 2026
• Last Verified: March 1, 2026

More Information

Terms related to this study

• Keywords: Intensive Care Unit
• Additional Relevant MeSH Terms: Pathologic Processes; Disease Attributes; Systemic Inflammatory Response Syndrome; Inflammation; Bacterial Infections and Mycoses; Sepsis; Infections; Communicable Diseases; Bacterial Infections
• Other Study ID Numbers: GOKAEK-2023/12.31

Plan for Individual participant data (IPD)

• Plan to Share Individual Participant Data (IPD)?: UNDECIDED

Drug and device information, study documents

• Studies a U.S. FDA-regulated drug product: No
• Studies a U.S. FDA-regulated device product: No