Emergency Department Wait Time Prediction based on Cyclical Features by Deep Neural Networks.
Liang Z, Huang JX
AMIA 2022 Clinical Informatics Conference. 2022, May 24-26, Houston, USA, Poster.
Abstract:
We use Python with the Google Tensorflow package (version 2.6) to implement the deep neural network (DNN) models forthe emergency department (ED) wait time prediction. The application is a cross-platform console App that can be run inboth Windows environment and Unix / Linux environment. The system server initially launches the training with a datasetextracted from the EHR database with all the ED visits from January 2019 to December 2020, which contains 92,671 rows intotal. The dataset is chronologically splatted by 85% for training and 15% for validation. The trained models are saved inthe server (Windows server for our deployment) and will be updated periodically every 24 days. The prediction is run every30 minutes triggered by a server script (PowerShell script) based on the latest 5 consecutive ED wait-time records fetched from the EHRsystem.
Liang Z, Huang JX. Emergency Department Wait Time Prediction based on Cyclical Features by Deep Neural Networks.
AMIA 2022 Clinical Informatics Conference. 2022, May 24-26, Houston, USA, Poster.