Comparing and integrating human mobility data sources for measles transmission modeling in Zambia

Area of Science:

• Epidemiology
• Computational modeling
• Data science

Background:

• Accurate quantification of population mobility is essential for infectious disease dynamics modeling.
• Multiple data sources exist for mobility, but systematic integration methods are lacking.
• Integrating data can mitigate biases inherent in individual datasets.

Purpose of the Study:

• To systematically integrate multiple data sources for quantifying subnational population mobility.
• To develop and evaluate a departure-diffusion model for combining mobility data.
• To assess the impact of different data sources and models on infectious disease spread simulations.

Main Methods:

• Explored summary mobility metrics from mobile phone records, travel surveys, Demographic and Health Surveys, and Facebook location data in Zambia.
• Developed a departure-diffusion model to integrate diverse mobility datasets.
• Utilized a metapopulation model to simulate a measles outbreak using different mobility data inputs.

Main Results:

• Mobility estimates (trip probability, travel locations) varied significantly across data sources.
• Departure-diffusion models incorporating mobile phone records were dominated by this data due to its spatial coverage.
• Travel survey data, when used for mobility parameterization, led to measles introduction in 98% of districts, compared to <50% for mobile phone or Facebook data.

Conclusions:

• Methods for integrating multiple mobility datasets are needed to improve the validity of mobility estimates.
• The choice of data source significantly impacts infectious disease transmission dynamics modeling.
• Integrated mobility data enhances the accuracy of predicting the spatial spread of infectious diseases.