XG Boost-Based approach for Machine Learning-Based Healthcare System Response Time Prediction

Abstract

The quality of service (i.e. effective reaction time) directly affects the patient's safety and operational metrics in health care systems and is also essential. However, fluctuating patient load, staffing level, road conditions and time variation are all dynamic factors that complicate response time prediction. This work leverages the eXtreme Gradient Boosting (XGBoost) algorithm for predicting health system response times through a computational learning approach. The input data for model training was a dataset of operational and contextual variables (request-time, department-type, staff-number, patient’s number, distance). Several data preprocessing techniques, such as encoding, normalizing and feature selection were applied to enhance the model performance. R2 and MAE are used as parameter to estimate the XGBoost model. According to experimental data, the suggested method outperformed conventional regression models with a high prediction accuracy (R² = 0.85). The most important determinants of response time, according to feature importance analysis, were staff availability, time of day and patient load.The study highlights how machine learning may improve patient care efficiency by reducing system delays and simplifying healthcare resource management.