ABSTRACT Overcrowding within emergency departments

ABSTRACT
Overcrowding within emergency departments (EDs) can have huge negative impact on patients. Therefore there is a need for innovative methods to improve patient flow and prevent overcrowding. One of the possible methods is to use data mining using machine learning techniques to predict EDs admissions. The study uses routinely collected administrative data (120,600 records) from two major acute hospitals in Northern Ireland to compare contrasting machine learning algorithms in predicting the risk of admission from EDs. We use three algorithms to build the predictive models: logistic regression, decision tree and gradient boosting machines (GBM). The GBM performs better with accuracy 80.31%, AUC-ROC 0.859 than the decision tree accuracy 80.06%, AUC-ROC 0.824 and the logistic regression model accuracy 79.94%, AUC-ROC 0.849.When going through logistic regression, we identify several factors related to hospital admissions, including hospital site, age, arrival mode, triage category, care group, previous admission in the past month, and previous admission in the past year. This study focuses on three common machine learning algorithms in predicting patient admissions. Actual implementation of models developed in this study in the decision support tools would provide a snapshot of predicted admissions from the ED at a given time, allowing for advance resource planning and the avoidance of bottlenecks in patient flow, as well as comparison of predicted and actual admission rates. When interpretability is a key consideration, EDs should consider logistic regression models, although GBM’s will be useful where accuracy is paramount.