FluHMM: A simple and flexible Bayesian algorithm for sentinel influenza surveillance and outbreak detection

Area of Science:

• Epidemiology
• Public Health
• Biostatistics

Background:

• Timely detection of seasonal influenza epidemics is crucial for effective public health interventions.
• Current methods may require historical data or lack flexibility in phase identification.

Purpose of the Study:

• To introduce FluHMM, a Bayesian algorithm for detecting and monitoring seasonal influenza epidemics.
• To enable early warning systems for public health action based on sentinel surveillance data.

Main Methods:

• Development of FluHMM, a flexible Bayesian algorithm utilizing sentinel surveillance data.
• Segmentation of influenza seasons into five distinct phases: pre-epidemic, epidemic growth, plateau, decline, and post-epidemic.
• Calculation of weekly posterior probabilities for each phase, enabling alert generation.

Main Results:

• FluHMM demonstrated high sensitivity, timeliness, and perfect specificity in analyzing 12 seasons of Greek sentinel surveillance data.
• The algorithm successfully segments seasons and provides interpretable phase probabilities.
• An R package is available to facilitate practical application in public health.

Conclusions:

• FluHMM offers a robust and flexible tool for real-time influenza epidemic monitoring and detection.
• The method's performance suggests significant utility for public health surveillance and response.
• Future extensions could involve integrating multiple data streams for enhanced epidemic surveillance.