Artificial Intelligence to Get Insights of Multi-Drug Resistance Risk Factors during the First 48 Hours from ICU...

Area of Science:

• Medical Informatics
• Infectious Diseases
• Critical Care Medicine

Background:

• Multi-drug resistance (MDR) poses a significant global health threat, particularly in Intensive Care Units (ICUs).
• Current MDR confirmation methods require 48 hours, delaying critical treatment decisions.
• Predicting MDR risk early is crucial for timely intervention in vulnerable ICU patients.

Purpose of the Study:

• To explore artificial intelligence (AI) approaches for predicting MDR risk within the first 48 hours of ICU admission.
• To identify key clinical, demographic, and treatment-related risk factors associated with MDR development in ICUs.
• To evaluate the potential of machine learning (ML) models in forecasting MDR germ emergence.

Main Methods:

• Utilized clinical and demographic data, mechanical ventilation status, and antibiotic exposure within the initial 48 hours of ICU stay.
• Applied three feature selection strategies to identify significant differences between MDR and non-MDR patient groups.
• Developed and analyzed ML models using selected features to predict MDR risk.

Main Results:

• Identified 24 significant features, including SAPS III and Apache II scores, patient age, and department of origin.
• Demonstrated the potential of AI and ML approaches in predicting MDR development.
• Highlighted the value of integrating electronic health record data for enhanced prediction.

Conclusions:

• AI offers a promising avenue for early MDR risk assessment in ICUs, bridging the gap before definitive microbiological results.
• Key patient characteristics and clinical scores can serve as valuable predictors of MDR.
• Further development of AI-driven tools can significantly impact patient management and outcomes in critical care settings.