RESUMEN
The basic homogeneous SEIR (susceptible-exposed-infected-removed) model is a commonly used compartmental model for analysing infectious diseases such as influenza and COVID-19. However, in the homogeneous SEIR model, it is assumed that the population of study is homogeneous and, one cannot incorporate individual-level information (e.g., location of infected people, distance between susceptible and infected individuals, vaccination status) which may be important in predicting new disease cases. Recently, a geographically-dependent individual-level model (GD-ILM) within an SEIR framework was developed for when both regional and individual-level spatial data are available. In this paper, we propose to use an SEIR GD-ILM for each health region of Manitoba (central Canadian province) population to analyse the COVID-19 data. As different health regions of the population under study may act differently, we assume that each health region has its own corresponding parameters determined by a homogeneous SEIR model (such as contact rate, latent period, infectious period). A Monte Carlo Expectation Conditional Maximization (MCECM) algorithm is used for inference. Using estimated parameters we predict the infection rate at each health region of Manitoba over time to identify highly risk local geographical areas. Performance of the proposed approach is also evaluated through simulation studies.
RESUMEN
The purposes of our study are to map high-risk areas in Canada as well as quantifying the effects of vaccination intervention and socio-demographic factors on the transmission rates of infection, recovery, and death related to COVID-19. The data of this research included weekly number of COVID19 cases, recovered, and dead individuals from 2020 through 2021 in Canada at health region and provincial levels. These data were associated with cumulative rates of partial and full vaccination and socio-demographic factors. We applied the spatio-temporal Susceptible-Exposed-Infected-Removed (SEIR), and Susceptible-Exposed-Infected-Removed-Vaccinated (SEIRV) models. The results indicated the partial vaccination rate has a greater effect compared with full vaccination rate on decreasing the rate of infectious cases (risk ratio (RR) = 0.18; 95%CrI: 0.16-0.2; RR = 0.60; 95%CrI: 0.55-0.65, respectively) and increasing the rate of recovered cases (RR = 1.39; 95%CrI: 1.28-1.51; RR = 1.21; 95%CrI: 1.23-1.29, respectively). However, for mortality risk reduction, only increasing full vaccination rate was significantly associated (RR = 0.09; 95%CrI: 0.05-0.14). In addition, our results showed that regions with higher rates of elderly and aboriginal individuals, higher population density, and lower socioeconomic status (SES) contribute more to the risk of infection transmission. Rates of elderly and aboriginal individuals and SES of regions were significantly associated with recovery rate. However, elderly individuals rate of regions was only a significant predictor of mortality risk. Based on the results, protection against mild and severe COVID-19 infection after the primary vaccination series decreased.