A hybrid hierarchical Bayesian model for spatiotemporal surveillance data

Area of Science:

• Epidemiology
• Biostatistics
• Data Science

Background:

• Spatiotemporal data analysis is challenging due to low signal-to-noise and high dimensionality.
• Traditional control charts often violate assumptions for syndromic surveillance data, hindering accurate outbreak detection.

Purpose of the Study:

• To develop a novel hybrid hierarchical Bayesian model for robust spatiotemporal outbreak detection.
• To address limitations of existing methods in handling non-independent, non-normal, and non-stationary surveillance data.

Main Methods:

• Combined Dirichlet process and particle filters within a Markovian state-space model.
• Utilized a modified adjacency matrix as the observation matrix for dimension reduction.
• Implemented online updating for data streaming applications.

Main Results:

• The hybrid model demonstrated superior detection performance compared to existing methods.
• Successfully identified both the 2009 H1N1 pandemic and seasonal influenza outbreaks.
• Achieved dimension reduction and computational efficiency.

Conclusions:

• The Dirichlet process/particle filter model offers a significant advancement in public health surveillance for outbreak detection.
• This methodology is effective for real-time analysis of streaming surveillance data.
• The model successfully handles complex data characteristics often found in syndromic surveillance systems.