Shengyang Qin1,2, Xinxing Duan1, Paul Kimm3. 1. School of Public Policy & Management, China University of Mining and Technology, Xuzhou, China. 2. Student Affairs Office, Yancheng Teachers University, Yancheng, China. 3. School of Science, Engineering & Design, Teesside University, UK.
Abstract
BACKGROUND: Environmental pollution caused by economic development poses a serious threat to human health. How to prevent and control environmental health risks has received extensive attention. OBJECTIVE: It is to explore the application of deep learning methods in assessment and prediction of environmental health risks. METHODS: A time series prediction model is constructed based on the generalized additive model (GAM) and a long short-term memory neural network (LSTM) prediction model is established based on deep learning, and they are combined into a hybrid model. The three models are used to predict and analyse the number of hospitalizations of the three diseases under environmental pollutants. RESULTS: Compared with the GAM and LSTM models, the mean absolute percentage error (MAPE) value of the hybrid model to predict the number of hospitalized patients with respiratory diseases decreases by 2.3%and 1.9%, respectively. CONCLUSION: The hybrid prediction model proposed can better predict the number of hospitalized patients with systemic diseases under the influence of environmental pollutants, and provide an important reference for the application of deep learning neural networks in risk assessment of environmental health.
BACKGROUND: Environmental pollution caused by economic development poses a serious threat to human health. How to prevent and control environmental health risks has received extensive attention. OBJECTIVE: It is to explore the application of deep learning methods in assessment and prediction of environmental health risks. METHODS: A time series prediction model is constructed based on the generalized additive model (GAM) and a long short-term memory neural network (LSTM) prediction model is established based on deep learning, and they are combined into a hybrid model. The three models are used to predict and analyse the number of hospitalizations of the three diseases under environmental pollutants. RESULTS: Compared with the GAM and LSTM models, the mean absolute percentage error (MAPE) value of the hybrid model to predict the number of hospitalized patients with respiratory diseases decreases by 2.3%and 1.9%, respectively. CONCLUSION: The hybrid prediction model proposed can better predict the number of hospitalized patients with systemic diseases under the influence of environmental pollutants, and provide an important reference for the application of deep learning neural networks in risk assessment of environmental health.
Entities:
Keywords:
Environmental health; deep learning; risk assessment; the long and short-term memory neural network; time series prediction