Clinical Timing-Sequence Warning Models for Serious Bacterial Infections in Adults Based on Machine Learning:...

Area of Science:

• Medical Informatics
• Clinical Decision Support
• Machine Learning in Healthcare

Background:

• Serious bacterial infections (SBIs) are a significant cause of unplanned hospital admissions and mortality.
• Early and accurate identification of SBIs in patients presenting with fever is critical for effective clinical management.

Purpose of the Study:

• To establish and validate clinically applicable models for identifying SBIs in patients with infective fever.
• To leverage machine learning techniques for feature selection and model development.

Main Methods:

• Retrospective collection of clinical and laboratory data from 945 patients with infective fever.
• Application of machine learning algorithms (Boruta, Lasso, RFE) for feature selection.
• Development and validation of predictive models using logistic regression, random forest, and XGBoost, including timing-sequence analysis and nomogram plotting.

Main Results:

• 69.9% of patients had SBIs; key predictors included age, hemoglobin, neutrophil-to-lymphocyte ratio, fibrinogen, and C-reactive protein.
• Two timing-sequence models (early admission and within 24 hours) were developed using logistic regression.
• The logistic regression model for prediction within 24 hours achieved an AUC of 0.780, accuracy of 0.754, and sensitivity of 0.776.

Conclusions:

• The developed clinical timing-sequence warning models are effective in predicting SBIs in patients with infective fever.
• These models demonstrate potential for enhancing clinical decision-making and warrant further validation in prospective, multicenter studies.