Scalable and distributed strategies for socially distanced human mobility

Area of Science:

• Computational epidemiology
• Network science
• Public health policy

Background:

• Nations worldwide continue to implement human mobility restrictions to curb COVID-19 transmission.
• Existing social distancing optimization strategies are computationally intensive and require comprehensive social network data.
• Emerging viral strains and vaccine efficacy questions necessitate advanced, intelligent mobility policies.

Purpose of the Study:

• To develop scalable and distributed versions of social distancing optimization approaches.
• To address the computational limitations and data requirements of previous methods.
• To enhance the effectiveness of mobility policies in controlling infectious disease spread.

Main Methods:

• Utilized Markov Chain Monte Carlo Gibbs sampling for distributed optimization.
• Implemented grid-based spatial parallelization to improve computational efficiency.
• Conducted extensive simulation experiments to evaluate strategy performance.

Main Results:

• Proposed strategies demonstrated performance comparable to optimal methods.
• Achieved significant speed-up in computations compared to prior approaches.
• Validated scalability and effectiveness in realistic scenarios, including non-compliance.

Conclusions:

• Scalable and distributed optimization methods offer a computationally feasible solution for mobility-based contagion control.
• These advanced strategies can inform more effective public health policies for infectious disease management.
• The developed approaches provide a robust framework for optimizing interventions in dynamic epidemiological landscapes.