Mobility and COVID-19 Transmission Dynamics

In this study, for simplicity, the mobility at the transit stations was considered as a surrogate to represent COVID-19 transmission to avoid nonlinear regression computations. This is based on our previous study results that revealed the mobility at the transit stations (Google mobility) as the most important factor characterizing the new DPC [29 (link)]. When using machine learning, the accuracy of the 2-week new DPC forecasting is > 82.6%, whereas the remaining factors included the weather and condition of state-of-emergency [20 ]. Thus, a time window averaging of the mobility was investigated, which is approximately characterized by the incubation time and latency from sample collection to reporting in healthcare facilities to relate with the ERN [20 ]. The mobility at transit stations was averaged over time windows (days) considering the latency (days) (e.g., setting the duration to 6 days and latency to 4 days means averaging the mobility of 4–9 days before the relevant date).
The correlation between the ERN and public mobility with latency was analyzed using the Pearson and Spearman rank correlation. The JMP software package (SAS Institute, Cary, NC, USA) was used for statistical analysis. A p-value of < 0.05 was considered statistically significant to specify the dominant factors that influence the rates.