Bayesian spatio-temporal modeling of COVID-19 incidence in Algerian provinces using integrated nested Laplace approximations
Résumé
The COVID-19 pandemic in Algeria dissplayed significant spatial and temporal heterogeneity, especially during the severe summer 2021 wave driven by the Delta variant. Standard national-level statistics often obscure this critical local variation, creating a need for advanced modeling to inform precise public health interventions. This study aimed to perform a high-resolution spatio-temporal analysis of COVID-19 incidence across Algeria's 48 provinces (Wilayas) to identify persistent high-risk areas and track the dynamics of viral spread. A spatio-temporal analysis was conducted on COVID-19 case data from all 48 Wilayas during epidemiological weeks 26-37 of 2021. We employed a Bayesian hierarchical model fitted using the Integrated Nested Laplace Approximation (INLA). The model incorporated structured spatial (Leroux prior), temporal (random walk of order 1), and spatio-temporal (Type IV interaction) random effects. Model selection was performed using the Watanabe-Akaike Information Criterion (WAIC) and Deviance Information Criterion (DIC) The spatio-temporally structured interaction model provided the best fit. Spatial heterogeneity was the dominant driver of transmission risk, accounting for 83.4% of the explained variance. Northeastern Wilayas, including Constantine and Tebessa, exhibited persistently high relative risks. The national temporal trend showed a sharp peak in early August 2021. The spatio-temporal interaction term (16.5% of variance) captured the progressive westward spread of the virus along the northern coast throughout the study period. This analysis demonstrates the critical utility of Bayesian spatio-temporal models in moving beyond national averages to identify specific high-risk areas and understand the evolving dynamics of an epidemic. The findings provide a valuable evidence base for designing targeted public health strategies. While this foundational study establishes the spatio-temporal risk patterns, future work incorporating socio-economic and environmental covariates will be essential to elucidate the underlying drivers of transmission.
Citer ce document
Accès au document
Texte intégral en lecture en ligne, réservé aux abonnés SPHAERO et aux membres de l'institution. Se connecter
Voir l'article sur le site de la revueAuteur(s)
Statistiques
Consultations : 1
Téléchargements : 0