A Universal Model for Prediction of COVID-19 Pandemic Based on Machine Learning
Background: With the current worldwide spreading of the coronary virus (COVID-19) pandemic, accurately predicting the rate of spread of the virus has become an urgent need.
Methods: In this article we propose a universal COVID-19 prediction model that is independent of country-specific factors in this paper. By analyzing the pandemic data in China, we combined the advantages of Gaussian function with that of chi-square distribution function, to render an innovative mathematical model named the H-Gaussian with five parameters to be learned, and solved the parameters by a gradient descent algorithm.
Results: We trained the model with partial historical pandemic data to predict subsequent pandemic trends in several regions, and validated the predictions with real data. The H-Gaussian model was experimentally shown to correctly predict the pandemic trends, and the parameters had good interpretability.
Conclusions: On this basis, the global trends of the pandemic are given based on the data currently available, as well as suggestions for subsequent prevention strategies.
Figure 1
Figure 2
Figure 3
Figure 4
Figure 5
Figure 6
Due to technical limitations, full-text HTML conversion of this manuscript could not be completed. However, the manuscript can be downloaded and accessed as a PDF.
Posted 26 May, 2020
A Universal Model for Prediction of COVID-19 Pandemic Based on Machine Learning
Posted 26 May, 2020
Background: With the current worldwide spreading of the coronary virus (COVID-19) pandemic, accurately predicting the rate of spread of the virus has become an urgent need.
Methods: In this article we propose a universal COVID-19 prediction model that is independent of country-specific factors in this paper. By analyzing the pandemic data in China, we combined the advantages of Gaussian function with that of chi-square distribution function, to render an innovative mathematical model named the H-Gaussian with five parameters to be learned, and solved the parameters by a gradient descent algorithm.
Results: We trained the model with partial historical pandemic data to predict subsequent pandemic trends in several regions, and validated the predictions with real data. The H-Gaussian model was experimentally shown to correctly predict the pandemic trends, and the parameters had good interpretability.
Conclusions: On this basis, the global trends of the pandemic are given based on the data currently available, as well as suggestions for subsequent prevention strategies.
Figure 1
Figure 2
Figure 3
Figure 4
Figure 5
Figure 6
Due to technical limitations, full-text HTML conversion of this manuscript could not be completed. However, the manuscript can be downloaded and accessed as a PDF.