A novel spatio-temporal interpolation algorithm and its application to the COVID-19 pandemic

Abstract

This paper describes several interpolation methods for predicting the number of cases of the COVID-19 pandemic. The interpolation methods include some well-known temporal interpolation algorithms including Lagrange interpolation, cubic spline interpolation, and exponential decay interpolation. These temporal interpolation algorithms enable the interpolation of the COVID-19 cases at locations where measures on prior days are available. However, pandemics are not purely temporal but spatio-temporal phenomena. Therefore, the neighboring locations need to be considered too in order to derive accurate interpolation values for future days. This paper introduces a novel spatio-temporal interpolation algorithm that is shown to be better than any purely temporal interpolation algorithm in predicting the COVID-19 cases in the continental United States. In particular, the novel spatio-temporal interpolation method achieves a mean absolute error of 8.44 cases over a million people when predicting two days ahead the number of cases of the COVID-19 pandemic.