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

Area of Science:

• Microbiology and Epidemiology
• Public Health Surveillance

Background:

• Antimicrobial resistance (AMR) poses a significant global health threat.
• Effective surveillance systems are crucial for monitoring and managing AMR trends.
• Existing time series data from the European Antimicrobial Resistance Surveillance Network (EARS-Net) offer potential for predictive analysis.

Purpose of the Study:

• To investigate the utility of EARS-Net antimicrobial resistance time series for anticipating future resistance dynamics.
• To evaluate the effectiveness of pooled forecasting and event-based methodologies for AMR surveillance.

Main Methods:

• Application of a pooled forecasting framework to predict future resistance percentages across various time horizons.
• Development of a pooled event-based methodology focused on predicting substantial increases in antimicrobial resistance.
• Analysis of EARS-Net time series data to assess predictive capabilities.

Main Results:

• Exact forecasting using EARS-Net data provided limited practical advantage for surveillance due to small predicted changes and unreliability in capturing large increases.
• The event-based methodology showed promise for identifying potential significant rises in antimicrobial resistance.
• Antimicrobial resistance time series data are more valuable for early warning of critical resistance events than for precise future level forecasting.

Conclusions:

• EARS-Net time series data are more effectively utilized for early warning of significant antimicrobial resistance increases.
• An event-based approach is more suitable than exact forecasting for actionable AMR surveillance using EARS-Net data.
• This study highlights a shift in focus towards predictive event detection for enhanced public health response to antimicrobial resistance.