When individual behaviour matters: homogeneous and network models in epidemiology

Area of Science:

• Epidemiology
• Network Science
• Mathematical Biology

Background:

• Population-level disease dynamics are heavily influenced by host contact patterns.
• Many epidemiological models simplify these patterns, often assuming homogeneous mixing.
• Network-based approaches have emerged to model contact heterogeneity explicitly.

Purpose of the Study:

• To quantify the departure of real populations from homogeneous mixing assumptions using a network perspective.
• To evaluate methodologies for incorporating contact heterogeneity in epidemiological models.
• To compare the accuracy of network models versus modified compartmental models.

Main Methods:

• Utilized a network perspective to analyze host contact patterns.
• Quantified heterogeneity in underlying network structure and epidemiological dynamics.
• Evaluated various methods for incorporating contact heterogeneity, including network and SIR models.

Main Results:

• Human contact patterns exhibit more heterogeneity than homogeneous-mixing models assume.
• Contact pattern variability is less extreme than some previous speculations suggested.
• Network models demonstrated greater accuracy and intuitiveness for heterogeneous populations.

Conclusions:

• Homogeneous-mixing compartmental models are suitable for nearly homogeneous populations.
• Modified compartmental models can be effective for specific non-homogeneous networks.
• Network models are generally superior for predicting disease spread in heterogeneous host populations.