Machine Learning-Based Techniques for Forecasting Emergency Department Peak Load Due to Respiratory Diseases Patients




Journal Title

Journal ISSN

Volume Title



Chronic Respiratory diseases, mainly asthma and Chronic Obstructive Pulmonary Disease (COPD), affect the lives of people by limiting their activities in various aspects. Overcrowding of hospital emergency departments (EDs) due to respiratory diseases in certain weather and environmental pollution conditions results in the degradation of quality of medical care, and even limits its availability. A useful tool for ED managers would be to forecast peak demand days so that they can take steps to improve the availability of medical care. We trained three classifier models to predict peak demand days of EDs due to respiratory patients. The first classifier is an Artificial Neural Network (ANN)-based classifier using Multi-Layer Perceptron (MLP) with back propagation algorithm that predicts Peak Event (Peak demand days) of patients with respiratory diseases, mainly asthma and COPD visiting EDs in Dallas County of Texas in the United States. We used the number of patients initially diagnosed with ICD-9-CM codes 490 to 496 (Chronic Pulmonary Obstructive Disease and Allied Conditions) in the first experiment. The precision and recall for Peak Event class were 77.1% and 78.0% respectively and those for Non-Peak Events was 83.9% and 83.2% respectively. The overall accuracy of the system is 81.0%. The second and third classifiers are Random Forests (RF)-based classifiers to predict peak days of ED visits by asthma patients. Feature-selection has been employed to develop the simplified or reduced model. Random Forests (RF) method of decision tree technique has been used for predicting daily hospital ED visits/admissions due to asthma exacerbations based on daily weather and environmental pollution measurements. For these classifiers, we used ICD-9-CM code 493 (Asthma). The base or full model uses all eight (8) predictors (historical weather and environment pollution measurements), while the other model, known hereafter as simplified model, uses only four (4) out of these eight, reduced through Correlation Feature Selection (CFS) subset evaluator. The overall accuracy of base model is 81.30%, while that of simple model was 76.59%.



Asthma, Lungs—Diseases, Obstructive, Health planning, Neural networks (Computer science), Decision trees, Weather—Physiological effect, Pollution—Measurement, Hospitals—Emergency services, Machine learning


©2017 The Author. Digital access to this material is made possible by the Eugene McDermott Library. Further transmission, reproduction or presentation (such as public display or performance) of protected items is prohibited except with permission of the author.