Prediction of COVID-19 Data Using Improved ARIMA-LSTM Hybrid Forecast Models
COVID-19 has developed into a global public health emergency and has led to restrictions in numerous nations. Thousands of deaths have resulted from the infection of millions of individuals globally. Additionally, COVID-19 has had a significant impact on social and economic activity around the world...
| Main Authors: | , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2023-01-01
|
| Series: | IEEE Access |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/10172007/ |
| _version_ | 1827903739128709120 |
|---|---|
| author | Yong-Chao Jin Qian Cao Ke-Nan Wang Yuan Zhou Yan-Peng Cao Xi-Yin Wang |
| author_facet | Yong-Chao Jin Qian Cao Ke-Nan Wang Yuan Zhou Yan-Peng Cao Xi-Yin Wang |
| author_sort | Yong-Chao Jin |
| collection | DOAJ |
| description | COVID-19 has developed into a global public health emergency and has led to restrictions in numerous nations. Thousands of deaths have resulted from the infection of millions of individuals globally. Additionally, COVID-19 has had a significant impact on social and economic activity around the world. The elderly and those with existing medical issues, however, are particularly vulnerable to the effects of COVID-19. Pneumonia, acute respiratory distress syndrome, organ failure, death, etc. are all possible outcomes in severe cases…Traditional prediction approaches like the ARIMA model and multiple linear regression model to handle the linear prediction problem because the new crown virus is in the process of continual mutation. Deep learning models that can take into account nonlinear elements include BP neural network prediction and LSTM neural network prediction. To combine the benefits of traditional and deep learning predictive models and create superior predictive models, we can blend traditional and deep learning predictive models. When the MSE, RMSE, and MAE of these three combined models, PSO-LSTM-ARIMA, MLR-LSTM-ARIMA, and BPNN-LSTM-ARIMA, are compared. We discovered that the third model, which included MSE, RMSE, and MAE, had the best prediction accuracy. The LSTM model and the ARIMA model were selected for this investigation. To begin, it employed a single model to forecast pandemic data in Germany. The BP neural network, particle swarm method, and multiple linear regression were then utilized to merge it. To corroborate this finding, we re-predicted the epidemic data from Japan and retrieved the MSE, RMSE, and MAE values of the BPNN-LSTM-ARIMA model, which were 6141895.956, 2478.285 and 1249.832. The most accurate model is still this integrated model. The BP neural network coupled LSTM model and ARIMA model offers the highest accurate prediction effect, according to our research. Combinatorial models anticipate outbreak data through our study, which can aid governments and public health authorities in improving their responses and educating the public about pandemic trends and potential future directions. As a result, industries and enterprises may make better risk-management decisions to protect the health and safety of their operations and personnel. It also helps healthcare facilities better prepare and deploy medical resources to better meet the demands from the pandemic. |
| first_indexed | 2024-03-13T00:17:52Z |
| format | Article |
| id | doaj.art-e62f64317ffb4ed4b7df5302adeb9db2 |
| institution | Directory Open Access Journal |
| issn | 2169-3536 |
| language | English |
| last_indexed | 2024-03-13T00:17:52Z |
| publishDate | 2023-01-01 |
| publisher | IEEE |
| record_format | Article |
| series | IEEE Access |
| spelling | doaj.art-e62f64317ffb4ed4b7df5302adeb9db22023-07-11T23:00:20ZengIEEEIEEE Access2169-35362023-01-0111679566796710.1109/ACCESS.2023.329199910172007Prediction of COVID-19 Data Using Improved ARIMA-LSTM Hybrid Forecast ModelsYong-Chao Jin0Qian Cao1Ke-Nan Wang2https://orcid.org/0009-0007-3064-1756Yuan Zhou3Yan-Peng Cao4Xi-Yin Wang5College of Science, North China University of Science and Technology, Tangshan, ChinaCollege of Science, North China University of Science and Technology, Tangshan, ChinaCollege of Science, North China University of Science and Technology, Tangshan, ChinaCollege of Artificial Intelligence, North China University of Science and Technology, Tangshan, ChinaCollege of Science, North China University of Science and Technology, Tangshan, ChinaCollege of Science, North China University of Science and Technology, Tangshan, ChinaCOVID-19 has developed into a global public health emergency and has led to restrictions in numerous nations. Thousands of deaths have resulted from the infection of millions of individuals globally. Additionally, COVID-19 has had a significant impact on social and economic activity around the world. The elderly and those with existing medical issues, however, are particularly vulnerable to the effects of COVID-19. Pneumonia, acute respiratory distress syndrome, organ failure, death, etc. are all possible outcomes in severe cases…Traditional prediction approaches like the ARIMA model and multiple linear regression model to handle the linear prediction problem because the new crown virus is in the process of continual mutation. Deep learning models that can take into account nonlinear elements include BP neural network prediction and LSTM neural network prediction. To combine the benefits of traditional and deep learning predictive models and create superior predictive models, we can blend traditional and deep learning predictive models. When the MSE, RMSE, and MAE of these three combined models, PSO-LSTM-ARIMA, MLR-LSTM-ARIMA, and BPNN-LSTM-ARIMA, are compared. We discovered that the third model, which included MSE, RMSE, and MAE, had the best prediction accuracy. The LSTM model and the ARIMA model were selected for this investigation. To begin, it employed a single model to forecast pandemic data in Germany. The BP neural network, particle swarm method, and multiple linear regression were then utilized to merge it. To corroborate this finding, we re-predicted the epidemic data from Japan and retrieved the MSE, RMSE, and MAE values of the BPNN-LSTM-ARIMA model, which were 6141895.956, 2478.285 and 1249.832. The most accurate model is still this integrated model. The BP neural network coupled LSTM model and ARIMA model offers the highest accurate prediction effect, according to our research. Combinatorial models anticipate outbreak data through our study, which can aid governments and public health authorities in improving their responses and educating the public about pandemic trends and potential future directions. As a result, industries and enterprises may make better risk-management decisions to protect the health and safety of their operations and personnel. It also helps healthcare facilities better prepare and deploy medical resources to better meet the demands from the pandemic.https://ieeexplore.ieee.org/document/10172007/COVID-19 predictionARIMALSTMBPNNMLRPSO |
| spellingShingle | Yong-Chao Jin Qian Cao Ke-Nan Wang Yuan Zhou Yan-Peng Cao Xi-Yin Wang Prediction of COVID-19 Data Using Improved ARIMA-LSTM Hybrid Forecast Models IEEE Access COVID-19 prediction ARIMA LSTM BPNN MLR PSO |
| title | Prediction of COVID-19 Data Using Improved ARIMA-LSTM Hybrid Forecast Models |
| title_full | Prediction of COVID-19 Data Using Improved ARIMA-LSTM Hybrid Forecast Models |
| title_fullStr | Prediction of COVID-19 Data Using Improved ARIMA-LSTM Hybrid Forecast Models |
| title_full_unstemmed | Prediction of COVID-19 Data Using Improved ARIMA-LSTM Hybrid Forecast Models |
| title_short | Prediction of COVID-19 Data Using Improved ARIMA-LSTM Hybrid Forecast Models |
| title_sort | prediction of covid 19 data using improved arima lstm hybrid forecast models |
| topic | COVID-19 prediction ARIMA LSTM BPNN MLR PSO |
| url | https://ieeexplore.ieee.org/document/10172007/ |
| work_keys_str_mv | AT yongchaojin predictionofcovid19datausingimprovedarimalstmhybridforecastmodels AT qiancao predictionofcovid19datausingimprovedarimalstmhybridforecastmodels AT kenanwang predictionofcovid19datausingimprovedarimalstmhybridforecastmodels AT yuanzhou predictionofcovid19datausingimprovedarimalstmhybridforecastmodels AT yanpengcao predictionofcovid19datausingimprovedarimalstmhybridforecastmodels AT xiyinwang predictionofcovid19datausingimprovedarimalstmhybridforecastmodels |