Bayesian Nowcasting with Laplacian-P-Splines
提出一种结合P样条和拉普拉斯近似的贝叶斯方法，用于在疫情中快速估计未报告的每日病例数，并量化预测不确定性，通过模拟和比利时COVID-19数据验证。
During an epidemic, the daily number of reported infected cases, deaths or hospitalizations is often lower than the actual number due to reporting delays. Nowcasting aims to estimate the cases that have not yet been reported and combine it with the already reported cases to obtain an estimate of the daily cases. In this article, we present a fast and flexible Bayesian approach for nowcasting by combining P-splines and Laplace approximations. Laplacian-P-splines provide a flexible framework for nowcasting that is computationally less demanding as compared to traditional Markov chain Monte Carlo techniques. The proposed approach also permits to naturally quantify the prediction uncertainty. Model performance is assessed through simulations and the nowcasting method is applied to COVID-19 mortality and incidence cases in Belgium. Supplementary materials for this article are available online.