Application of Multistrategy Improvement Gray Wolf Algorithm to Optimize Extreme Gradient Boosting in Emergency...

Area of Science:

• Artificial Intelligence in Healthcare
• Machine Learning for Clinical Decision Support

Background:

• Effective emergency department (ED) triage is crucial for resource allocation and patient outcomes.
• Conventional triage methods struggle with increasing patient volumes and complex cases.

Purpose of the Study:

• To develop and evaluate a novel machine learning triage model for the ED.
• To enhance the accuracy and efficiency of emergency triage decisions.

Main Methods:

• Developed MIGWO-XGBOOST, a machine learning model using Multi-strategy Improved Gray Wolf Optimization (MIGWO) for parameter tuning.
• Processed missing data and split the dataset for training (80%) and testing (20%).
• Benchmarked performance against XGBOOST, GWO XGBOOST, AdaBoost, LSTM, and CNN-BiGRU.

Main Results:

• MIGWO-XGBOOST achieved an 8.5% accuracy improvement over standard XGBOOST.
• Reduced optimization time by 9,285 seconds compared to GWO-XGBOOST.
• Demonstrated superior accuracy over AdaBoost (12.5%), LSTM (3.3%), and CNN-BiGRU (1.9%).

Conclusions:

• MIGWO-XGBOOST offers a robust framework for rapid and accurate ED triage.
• The model enhances predictive strength and computational efficiency in complex data environments.
• Advanced machine learning can significantly support emergency decision-making and optimize patient care.