Nowcasting by Bayesian Smoothing: A flexible, generalizable model for real-time epidemic tracking

Sarah F. McGough; Michael A. Johansson; Marc Lipsitch; Nicolas A. Menzies

To assess the degree of autocorrelation and related smoothness in the NobBS predictions, we calculated the 1-week lagged autocorrelation of predictions (ρ a ) and compared this to the 1week lagged autocorrelation of cases (ρc). In addition, we computed metrics reflecting the accuracy of the approaches in capturing the change in cases from week-to-week: the mean absolute error of the change (MAEΔ) and the RMSE of the change (RMSEΔ) ( Table 2 ). The magnitude of change was much larger for the ILI data than dengue data, with average absolute value change of 1,312.6 cases/week versus 9.8 cases/week, yet both showed high autocorrelation (ρc = 0.958 for dengue and ρc = 0.972 for ILI). Comparing the full time series, the nowcasts produced by NobBS exhibited high autocorrelation for both diseases (ρ a = 0.876 for dengue, 0.973 for ILI) while the benchmark approach yielded lower autocorrelation for dengue nowcasts, comparatively (ρa = 0.631 for dengue, 0.970 for ILI). For dengue, over the weeks in which at least 1 case was initially reported, the NobBS approach achieved both lower mean absolute difference between predicted and observed changes in cases (NobBS MAEΔ = 23 vs. benchmark MAEΔ = 50) and lower RMSE of the change (NobBS RMSEΔ = 35.8 vs. benchmark RMSEΔ = 64.6). In addition, NobBS outperformed the benchmark approach over the full time series of dengue cases ( Table 2 ). For ILI, however, the metrics for the weekly change were similar for the two approaches (Table 2) .