A1181
Title: A framework to estimate the effective reproduction number considering reporting delays
Authors: Xuanan Lin - Keio University (Japan) [presenting]
Hiroshi Shiraishi - Keio University (Japan)
Abstract: Accurate estimation of the effective reproduction number is essential for epidemic monitoring. However, reporting delays can lead to underestimating recent infections and compromise real-time insights. Recent research has adapted the chain ladder method from actuarial science to address this challenge in epidemiology. Building on this, a structural state-space model is applied that explicitly accounts for reporting delays. Originally proposed by a prior study and further extended in a recent study, the model begins with data in a run-off triangle format, where each cell represents the number of cases by onset date and reporting delay. To enable recursive estimation, the triangle is reindexed into a pseudo-time series, converting the two-dimensional structure into a univariate sequence with missing values corresponding to unreported cases. A linear Gaussian state-space model with Kalman filtering is then applied to estimate these values. The latent process captures temporal patterns in reporting delays through a seasonal random walk. Applied to COVID-19 data from Tokyo, this approach enables robust estimation of the effective reproduction number. The method is flexible, interpretable, and computationally efficient, making it well-suited for real-time epidemic assessment under delayed reporting.
