How many suffice? A computational framework for sizing sentinel surveillance networks

Area of Science:

• Epidemiology
• Public Health
• Network Science

Background:

• Surveillance networks are crucial for disease detection, outbreak investigation, and epidemic management.
• Effective site selection in surveillance networks balances resource limitations with performance needs.
• Networks can be optimized for specific tasks like determining outbreak intensity, timing, and location.

Purpose of the Study:

• To evaluate the efficacy of different site placement algorithms for disease surveillance networks.
• To compare novel algorithms against existing methods for optimizing network performance and resource allocation.
• To determine if reduced network size impacts surveillance quality.

Main Methods:

• Investigated maximal coverage and K-median site placement models.
• Utilized a large dataset of de-identified Medicaid billing records spanning eight influenza seasons.
• Simulated influenza spread in Iowa to compare network performance using the Huff spatial interaction model.

Main Results:

• A network designed with two fewer sites achieved identical outbreak intensity performance compared to the existing network.
• Outbreak timing detection performance was matched using 59% fewer sites.
• Network size appears to play a minimal role in outbreak timing detection.

Conclusions:

• Novel algorithms can design more efficient surveillance networks outperforming the status quo in intensity and timing.
• Surveillance system size can potentially be reduced without compromising data quality.
• Optimized site selection offers a cost-effective approach to enhance public health surveillance.