Capturing spatial dependence of COVID-19 case counts with cellphone mobility data.

Area of Science:

• Epidemiology
• Spatial Statistics
• Public Health

Background:

• Traditional spatial models for disease analysis rely on physical proximity, assuming closer regions share more disease transmission.
• This assumption may not fully capture disease dynamics influenced by human movement patterns.

Purpose of the Study:

• To evaluate the effectiveness of telecommunications-derived mobility data in modeling spatial dependence for COVID-19 case counts.
• To compare the explanatory power of mobility data versus physical proximity in spatial epidemiological models.

Main Methods:

• Extended Besag York Mollié (BYM) models to incorporate both physical adjacency and mobility effects.
• Applied Gaussian Markov random field priors to the mobility effect, using inter-regional travel counts as edge weights.
• Utilized modern parameterizations for BYM models to analyze COVID-19 case data from two Spanish communities.

Main Results:

• Telecommunications-derived mobility data, specifically the number of trips between regions, significantly improved the explanation of COVID-19 case count variations.
• Mobility effects demonstrated superior explanatory power compared to traditional physical proximity measures.
• The inclusion of mobility data alongside adjacency effects provided a more nuanced understanding of spatial disease transmission.

Conclusions:

• Human movement patterns, as captured by mobility data, are critical for accurately modeling infectious disease spread.
• Mobility data should be integrated with physical proximity data in spatial models for enhanced COVID-19 surveillance and prediction.
• This approach offers a more robust framework for understanding and managing infectious disease outbreaks in interconnected populations.