Exact Forecasting and Event-Based Prediction in Annual EARS-Net Antimicrobial Resistance Series.

This study investigated how cumulative antimicrobial resistance time series from EARS-Net can be exploited to support anticipation of future resistance dynamics. A pooled forecasting framework was first applied to predict future resistance percentages across multiple time horizons. Exact forecasting provided only limited practical advantage for surveillance purposes, as most future changes were relatively small and larger increases were difficult to capture reliably. The research focus was therefore redirected toward prediction of events of substantial future resistance increase, and a corresponding pooled event-based methodology was developed. Overall, the findings suggest that EARS-Net time series may be more usefully applied for early warning of important resistance increases than for exact forecasting of future resistance levels.