Identifying Preventable Emergency Admissions in Hospitals Using Machine Learning

Area of Science:

• Health Informatics
• Artificial Intelligence in Medicine
• Public Health

Background:

• Emergency department (ED) overcrowding is a global health challenge.
• Increased ED visits for non-urgent issues exacerbate overcrowding, leading to longer wait times, higher mortality rates, and delayed care for acute conditions.

Purpose of the Study:

• To develop and evaluate machine learning (ML) models for distinguishing between emergent and non-urgent ED visits.
• To optimize resource allocation and improve patient outcomes in emergency care settings.

Main Methods:

• Implementation of four ML models: Decision Tree, Random Forest, AdaBoost, and XGBoost.
• Evaluation of model performance using real-world ED data.

Main Results:

• XGBoost model demonstrated superior performance compared to other models.
• XGBoost achieved the highest accuracy and F1-score in differentiating urgent from non-urgent ED visits.

Conclusions:

• Machine learning, particularly XGBoost, shows significant potential in effectively managing ED overcrowding.
• Accurate differentiation of patient needs can lead to improved efficiency and patient care in emergency departments.