Machine learning-based mortality prediction models for Crimean-Congo hemorrhagic fever patients

Area of Science:

• Infectious Diseases
• Medical Informatics
• Biostatistics

Background:

• Crimean-Congo hemorrhagic fever (CCHF) poses a significant public health threat.
• Identifying mortality risk factors in CCHF patients is crucial for timely intervention.
• Machine learning offers advanced tools for analyzing complex disease data.

Purpose of the Study:

• To identify risk factors for mortality in CCHF patients.
• To evaluate the predictive performance of machine learning models for CCHF mortality.
• To determine the most effective parameters for predicting CCHF-related deaths.

Main Methods:

• Utilized machine learning algorithms including XGboost, Logistic Regression, Random Forest, LightGBM, and Gradient Boosting Classifier.
• Developed predictive models using data from 891 confirmed CCHF cases.
• Evaluated model performance using Receiver Operating Characteristic (ROC) analysis and Area Under the Curve (AUC).

Main Results:

• Achieved high predictive performance with AUC values of 0.849 (XGboost), 0.919 (Logistic Regression), and 0.853 (Gradient Boosting Classifier).
• Identified platelet count, neutrophil-to-lymphocyte ratio (NLR), and neutrophil count as top predictors of mortality.
• The overall fatality rate among confirmed CCHF cases was 3.3%.

Conclusions:

• Machine learning models, particularly Logistic Regression, XGboost, and Gradient Boosting Classifier, can effectively predict CCHF mortality.
• Platelet count, NLR, and neutrophil count are significant indicators for assessing mortality risk in CCHF patients.
• These findings can aid in clinical decision-making and resource allocation for CCHF management.