Related Experiment Video
Updated: May 20, 2025
06:33
Author Spotlight: Analyzing Bone Marrow Microenvironment in Murine Hematological Malignancies
Published on: November 10, 2023
1.1K
Machine learning-based prognostic model for bloodstream infections in hematological malignancies using Th1/Th2
Qin Li1,2, Nan Lin3, Zuheng Wang4
1Department of Hematology, The First Affiliated Hospital of Guangxi Medical University, Nanning, Guangxi, 530021, China.
BMC Infectious Diseases
|March 27, 2025
Summary
A new machine learning model predicts 30-day mortality in hematologic malignancy patients with bloodstream infections. This model integrates Th1/Th2 cytokines and clinical factors for improved accuracy in predicting outcomes for these vulnerable patients.
Area of Science:
- Hematology
- Infectious Diseases
- Computational Biology
Background:
- Bloodstream infections (BSIs) are a major cause of mortality in hematologic malignancy (HM) patients.
- Rising antibiotic resistance complicates BSI management in this population.
Purpose of the Study:
- To analyze pathogen distribution and drug-resistance patterns in HM patients with BSIs.
- To develop a novel predictive model for 30-day mortality in HM patients experiencing BSIs.
Main Methods:
- Retrospective analysis of 231 HM patients with positive blood cultures.
- Logistic regression and machine learning algorithms (including XGBoost, Logistic Regression, LightGBM) were employed.
- A predictive model integrating Th1/Th2 cytokines (IL-4, IL-6) and clinical features was developed and validated.
Main Results:
- Gram-negative bacteria were the predominant pathogens (64%).
- Significant predictors of 30-day mortality included age, polymicrobial BSI, IL-4, IL-6, and AST levels.
- The developed Logistic Regression model demonstrated strong predictive performance (AUCs 0.802-0.822) with excellent calibration.
Conclusions:
- A novel machine learning-based model effectively predicts 30-day mortality in HM patients with BSIs.
- The model integrates key Th1/Th2 cytokines and clinical features, showing strong performance and clinical applicability.
- This approach offers a promising tool for risk stratification and management of BSIs in HM patients.
