Spatial epidemiological models are deterministic or stochastic frameworks designed to investigate how spatial heterogeneities and host movement dynamics influence local and regional disease patterns. These models are essential for moving beyond homogeneous mixing assumptions and for capturing realistic geographic contexts in which infectious diseases emerge, spread, and respond to localized interventions.
ββββββββββββββββββββββββββββββββββββββββββββ
𧱠Compartmental Structure and Flow: The GPM Taxonomy
Spatial models based on compartmental thinking are commonly organized through the GeographyβPopulationβMovement (GPM) framework. This taxonomy decomposes a spatial epidemiological system into three interacting components that jointly determine disease dynamics.
Intra-Population Model (IPM)
The IPM describes transmission and disease progression within a single spatial unit or patch, typically using classical compartmental structures such as SIR or SEIR. It governs local infection, recovery, and removal processes.
Movement Model (MM)
The MM specifies how individuals move between spatial units. These movement rules couple local epidemics and are responsible for spatial diffusion, synchronization or desynchronization of outbreaks, and delays in epidemic peaks across regions.
Geographic Model (GEO)
The GEO component defines the physical and demographic structure of space. It specifies how regions are divided into patches or points of interest, their population sizes, and relevant environmental or demographic heterogeneities.
Together, these components provide a unifying structure for spatial compartmental models, partial differential equation models with continuous space, and agent-based models that explicitly track individual movement and interactions.
ββββββββββββββββββββββββββββββββββββββββββββ
π Mathematical Formulation: Metapopulation Ordinary Differential Equation System
In a metapopulation formulation, each spatial unit (or patch) is associated with its own compartmental dynamics. The evolution of a compartment in one patch depends on both local transmission processes and population exchange with other patches.
For susceptible individuals in patch i, the general form is
dSα΅’/dt = β Ξ² Sα΅’ Iα΅’ / Nα΅’
+ Ξ£β±Όβ α΅’ ( Mβ±Όα΅’ Sβ±Ό β Mα΅’β±Ό Sα΅’ )
where
Nα΅’ is the total population in patch i,
Ξ² Sα΅’ Iα΅’ / Nα΅’ represents local infection pressure governed by the IPM,
Mα΅’β±Ό denotes the movement rate from patch i to patch j, and
the summation term captures the net gain or loss of susceptibles due to movement.
Analogous equations govern exposed, infectious, and recovered compartments, resulting in a system of coupled ordinary differential equations across all spatial units.
ββββββββββββββββββββββββββββββββββββββββββββ
π’ Table 1. Parameter Definitions
| Parameter | Definition |
|---|---|
| Ξ² | Local transmission rate within a patch (IPM) |
| Ξ³ | Local recovery or removal rate within a patch |
| Nα΅’ | Total population size of patch i |
| Sα΅’ | Number of susceptible individuals in patch i |
| Iα΅’ | Number of infectious individuals in patch i |
| Mα΅’β±Ό | Movement rate from patch i to patch j |
| Mβ±Όα΅’ | Movement rate from patch j to patch i |
ββββββββββββββββββββββββββββββββββββββββββββ
π Table 2. Typical Parameter Ranges
| Parameter | Typical Range (per day) | Interpretation |
|---|---|---|
| Ξ² | 0.10 β 0.30 | Local transmission intensity |
| Ξ³ | 0.05 β 0.10 | Average infectious period of 10β20 days |
| Mα΅’β±Ό | 0.001 β 0.10 | Daily probability of movement between patches |
ββββββββββββββββββββββββββββββββββββββββββββ
π Applicability and Limitations
Applicability
Spatial GPM-based models are essential for identifying geographic hotspots, sourceβsink dynamics, and the spatial timing of epidemic waves. They are widely used to evaluate mobility restrictions, regional vaccination strategies, and targeted non-pharmaceutical interventions.
Policy Relevance
By explicitly modeling movement and spatial structure, these frameworks support evidence-based decisions on travel regulations, regional lockdowns, and optimal allocation of limited public health resources.
Data Dependence
Accurate implementation depends on reliable spatial and mobility data, such as commuting patterns, transportation networks, or mobile device traces. Parameter uncertainty in movement rates can substantially affect predictions.
Modeling Trade-offs
Agent-based spatial models provide fine-grained realism but are computationally intensive. Metapopulation models are more tractable at large scales but typically assume homogeneous mixing within each spatial unit, which may oversimplify local contact structures.
ββββββββββββββββββββββββββββββββββββββββββββ
π Selected References
Oshinubi, K., Chen, Y., Doerry, E., et al. A systematic review of spatial epidemiological modeling approaches applied during pandemics.
Riley, S. Large-scale spatial-transmission models of infectious disease.
Balcan, D., Gonçalves, B., Hu, H., et al. Modeling the spatial spread of infectious diseases through global mobility networks.
Chen, K., Jiang, X., Li, Y., and Zhou, R. Stochastic agent-based modeling of disease transmission influenced by human mobility.