Disease Surveillance on Complex Social Networks.

Area of Science:

• Epidemiology
• Network Science
• Public Health Surveillance

Background:

• Expanding infectious disease surveillance incorporates digital, crowd-sourced, and social network data.
• Contact networks are crucial for understanding disease transmission dynamics and identifying effective surveillance sensors.
• Public health agencies can leverage high-resolution data for selective monitoring of informative individuals.

Purpose of the Study:

• To evaluate three sensor selection strategies: most connected, random, and friends-of-random individuals.
• To assess these strategies across diverse social networks and multiple public health surveillance goals.
• To determine the optimal sensor choice based on public health objectives, network structure, and disease characteristics (R0).

Main Methods:

• Simulation of sensor selection strategies in three distinct social networks: scale-free, Venezuelan college student, and Montreal wireless hotspot networks.
• Evaluation across five surveillance goals: early detection of epidemic emergence and peak, and general situational awareness.
• Analysis of the impact of the basic reproduction number (R0) on sensor effectiveness.

Main Results:

• For low R0 diseases, highly connected individuals ('hubs') provide the earliest and most accurate outbreak detection.
• Hubs can be impractical to identify and misleading for general situational awareness, especially in networks with community structure or high R0.
• The friends-of-random strategy is a practical, robust alternative, identifying individuals with higher-than-average epidemiological risk for reasonably early and accurate information without prior network knowledge.

Conclusions:

• Optimal sensor selection for infectious disease surveillance is context-dependent, varying with public health goals, network topology, and disease R0.
• While theoretical optimal strategies exist, they are often impractical for real-world implementation.
• The friends-of-random approach presents a feasible and effective method for enhancing early outbreak detection in diverse network settings.