A Linear Regression Prediction Model of Infectious Disease Spread Based on Baidu Migration and Effective Distance.

ABSTRACT

Objective: To analyze the relationship between effective distance and epidemic spread trajectory and between arrival time and scale based on the COVID-19 data outbreak in Wuhan and thus to improve the prediction ability of the spread of infectious disease. Methods: Up to January 28, 2020, the reporting date, the onset date, and the cumulative number of confirmed cases of COVID-19 in each province and city were collected. Baidu migration data was used to calculate the effective distance from Wuhan city to other regions. The reporting date and onset date of the first diagnosed patient were taken as the arrival time, respectively, to establish a linear regression model of effective distance and arrival time. In different provinces and cities, the logarithm of the cumulative number of confirmed cases with a base of 5 was taken as the criteria to determine the level of the cumulative confirmed cases. Based on this, the linear regression model of effective distance and the level of cumulative confirmed cases in the provincial and municipal units was established. Results: The linear correlation between the reporting date of the first confirmed patient and the effective distance was not strong. The coefficients of determination (R 2) for cities with and without the cities of Hubei Province were 0.36 and 0.44, respectively. And the linear correlation between the onset date of the first confirmed patient and the effective distance was strong. And the coefficients of determination (R 2) for cities with and without the cities of Hubei Province were 0.67 and 0.83, respectively. And the linear correlation between the level of cumulative confirmed cases in the provincial and municipal units and the effective distance was strong, with an R 2 of 0.87 and 0.84, respectively. The regression coefficients of each linear model were statistically significant (P < 0.001). Conclusion: The effective distance has a good fit with the model of the onset date of the first confirmed patient and the level of cumulative confirmed cases, which can predict the trajectory, time, and transmission range of the epidemic. It can be taken as the reference for the early warning, prevention, and control of sudden acute infectious diseases from a macro perspective.