Practical geospatial and sociodemographic predictors of human mobility

Area of Science:

• * Human mobility research
• * Geospatial analysis
• * Public health preparedness

Background:

• * Assessing seasonal human mobility at subnational scales is crucial for urban planning and disease modeling.
• * Spatially and temporally resolved datasets for human mobility are often scarce, sensitive, or proprietary.
• * Understanding pre-pandemic mobility provides a baseline for assessing pandemic impacts and long-term changes.

Purpose of the Study:

• * To explore how broadly available covariates can describe typical seasonal subnational human mobility in Kenya before the COVID-19 pandemic.
• * To enable better modeling of seasonal mobility in low- and middle-income countries (LMICs) in non-pandemic settings.
• * To identify key correlates of human mobility using accessible data, overcoming limitations of sensitive mobile phone datasets.

Main Methods:

• * Utilized the Google Aggregated Mobility Research Dataset (anonymized user mobility flows).
• * Combined mobility data with socioeconomic and geospatial covariates from 2018-2019.
• * Employed spatiotemporal analysis within a Bayesian framework, accounting for spatial and temporal autocorrelations.

Main Results:

• * Pre-pandemic mobility in Kenya primarily involved shorter within-county trips, followed by inter-county and international travel.
• * Mobility peaked in August and December, correlating with school holidays, a significant predictor.
• * Urbanicity, poverty, female education, accessibility to population centers, and temperature were key explanatory covariates.

Conclusions:

• * Broadly available covariates can effectively describe seasonal human mobility patterns in LMICs.
• * Findings provide a crucial baseline for understanding pandemic-induced mobility shifts and potential long-term alterations.
• * Results offer insights for monitoring mobility proxies in disease surveillance and control efforts in LMICs.