2022/02/19 by P. Congdon, Congdon, P.
Economics, Econometrics and Finance · Mathematics · Medicine · #Applications (stat.AP) #COVID-19 epidemiological studies #Data-Driven Disease Surveillance #FOS: Computer and information sciences #Spatial and Panel Data Analysis
paper · pdf · doi:10.48550/arxiv.2202.09547
openalex publication_date 2022/02/19 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28
Spatio-temporal models for infection counts generally follow themes of the broader disease mapping literature, but may need to address specific features of spatio-temporal infection data including considerable time fluctuations (with epidemic phases) and spatial diffusion. Low order autoregression is a feature of several recent spatio-temporal studies of infection data, possibly with lags on both within area infections and on infections in adjacent areas. Many epidemic time series show a period of relatively stable infection levels (possibly characterized as endemicity), followed by a sudden sharp phase of increasing infection levels. After the epidemic peak there is a period of descending rates and return to stability. Hence one may seek to adapt the autoregressive scheme to these pronounced fluctuations, with temporary departures from stationarity, but returning to stationarity as rates descend and infections resume endemic levels. We consider a mixture link model for infection counts that allows adaptivity to both explosive phases and static endemicity. Two case study applications involve COVID area-time data, one for 32 London boroughs since the start of the COVID epidemic, the other focusing on the epidemic phase in 144 area of South East England associated with the Delta variant.