Modified genetic algorithm-based feature selection combined with pre-trained deep neural network for demand forecasting in outpatient department
作者：
Highlights：
• A deep learning approach to model the patients’ demand for different key resources.
• Normal GA is improved for feature selection tasks by redesigning its key operators.
• An SAE-based pre-training process is designed to improve the initialization of DNN.
• The case study is implemented with real data collected from a hospital.
• Proposed approach can be used as a universal model for demand forecasting in hospital.
摘要
•A deep learning approach to model the patients’ demand for different key resources.•Normal GA is improved for feature selection tasks by redesigning its key operators.•An SAE-based pre-training process is designed to improve the initialization of DNN.•The case study is implemented with real data collected from a hospital.•Proposed approach can be used as a universal model for demand forecasting in hospital.
论文关键词：Deep learning,Feature selection,Modified genetic algorithm,Stacked autoencoder,Demand forecasting in hospital
论文评审过程：Received 14 November 2016, Revised 29 March 2017, Accepted 6 April 2017, Available online 7 April 2017, Version of Record 19 April 2017.
论文官网地址：https://doi.org/10.1016/j.eswa.2017.04.017
