Intra-county modeling of COVID-19 infection with human mobility: assessing spatial heterogeneity with business traffic, age and race

Abstract

The novel coronavirus disease (COVID-19) pandemic is a global threat presenting health, economic and social challenges that continue to escalate. Meta-population epidemic modeling studies in the susceptible-exposed-infectious-removed (SEIR) style have played important roles in informing public health and shaping policy making to mitigate the spread of COVID-19. These models typically rely on a key assumption on the homogeneity of the population. This assumption certainly cannot be expected to hold true in real situations; various geographic, socioeconomic and cultural environments affect the behaviors that drive the spread of COVID-19 in different communities. What's more, variation of intra-county environments creates spatial heterogeneity of transmission in different sub-regions. To address this issue, we develop a new human mobility flow-augmented stochastic SEIR-style epidemic modeling framework with the ability to distinguish different regions and their corresponding behavior. This new modeling framework is then combined with data assimilation and machine learning techniques to reconstruct the historical growth trajectories of COVID-19 confirmed cases in two counties in Wisconsin. The associations between the spread of COVID-19 and human mobility, business foot-traffic, race & ethnicity, and age-group are then investigated. The results reveal that in a college town (Dane County) the most important heterogeneity is spatial, while in a large city area (Milwaukee County) racial-ethnic heterogeneity becomes more apparent. Scenario studies further indicate a strong response of the spread rate on various reopening policies, which suggests that policymakers may need to take these heterogeneities into account very carefully when designing policies for mitigating the spread of COVID-19 and reopening.

Machine Learning on the COVID-19 Pandemic, Human Mobility and Air Quality: A Review

“…[58] Detroit, US Investigating the impacts of COVID-19 and social distancing measures on traffic volume and safety. [59] Dane and Milwaukee County, City of Madison, US Modeling of COVID-19 spread and investigating the associations between COVID-19 transmission and mobility, business foot-traffic, and socioeconomic features. [60] Wuhan, China Predicting the number of COVID-19 infection cases related to patient recovery and death.…”

A Novel Model for Simulating COVID-19 Dynamics Through Layered Infection States that Integrate Concepts from Epidemiology, Biophysics and Medicine: SEI3R2S-Nrec

“…More sophisticated models use statistics and covariates to obtain λ or β (e.g., 11) or include multi-SEIR models for partitioning age, spatial location, etc. (9,12,17,18). Principles that have emerged from such modeling efforts include: formalizing the Reproduction Ratio (Ro, secondary cases/case) as a function of rates for various model formulations that can include additions such as birth rates (5)(6)(7)(8), illustrating advantageous properties of successful pathogen mutation (5)(6)(7), identifying conditions for stability and steadystate behavior (5)(6)(7)(8), identifying conditions for a "herd immunity threshold" (HIT) (14)(15)(16), and accommodating seasonal fluctuations (11,13).…”

“…A rich line of epidemiological research is based on developing collective social interaction contact networks that can facilitate viral transmission, often using "big data" analytics. Examples include use of mobility geolocation datasets (e.g., for metropolitan Portland area (23)(24), larger regions (17,18,25,26)). These can be helpful, but also controversial (e.g., huge variation in estimates of degree of spread through events such as the Sturgus Motorcycle Rally, up to hundreds of thousands).…”

“…Still another body of work is based on combining complex connection networks and/or geolocation "spacetime" network-type models to SEIRS-type compartmental models, which can be challenging since the two UoD's tend not to mesh well. Such approaches involve making assumptions on networks, using SEIR networks with partitioned subpopulations such as by age (12,17,18) or make heterogeneous connectivity distribution networks usually assumed to persistent over time (13,15), and have developed a strong theoretical foundation (12)(13)(14)(15). A key feature is capturing population heterogeneity assuming a unimodal distribution with a single mean and then a spread, for instance, with examples of assumed variational distribution including normal (Gaussian), binomal, Poisson, a power law, or centered exponential decay.…”

“…The cleanest approach is for fluxes between compartments to serve as "dynamic constraints" for the internal dynamic model. If there is a desire to pass "risk personas" between compartments, there can be subfluxes within a connectivity matrix (12,17,18), but this can lead to parameter explosion. This is best pursued by multidisciplinary teams having the targeted expertise needed to create and evaluate such submodels.…”