🧬 biology

Kinetic formulation of compartmental epidemic models

This paper introduces a kinetic model coupling individual movement with pathogen transmission that formally reduces to the classic SIRS epidemic model under specific limits, while also establishing solution existence and uniqueness and exploring connections to evolutionary game theory.

Original authors: Carolina Strecht-Fernandes, Fabio A. C. C. Chalub

Published 2026-01-30

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: Carolina Strecht-Fernandes, Fabio A. C. C. Chalub

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ⚕️ This is an AI-generated explanation of a preprint that has not been peer-reviewed. It is not medical advice. Do not make health decisions based on this content. Read full disclaimer

Imagine a crowded dance floor where people are moving around. Some are healthy (Susceptible), some are sick and spreading a virus (Infectious), and some have recovered but might get sick again later (Recovered).

Usually, scientists model how diseases spread using simple math that treats the whole crowd as a single, mixed-up soup. They ask, "If 10% are sick, how many healthy people will get sick next?" This works well for big pictures, but it misses the details of how people actually move and interact.

This paper introduces a new, more detailed way to model this dance floor. Think of it as upgrading from a blurry photo of the crowd to a high-definition video where you can see every single person's speed and direction.

Here is the breakdown of their new approach:

1. The "Kinetic" Dance Floor

Instead of just counting heads, the authors track every individual's position (where they are) and velocity (how fast and in what direction they are moving).

The Rule of Transmission: In this model, a healthy person only catches the virus if they are standing right next to a sick person and they are moving in the same direction at the same speed.
The Analogy: Imagine two people walking past each other quickly. They barely have time to shake hands, so the virus doesn't spread. But if they are walking side-by-side in the same direction for a long time (like on a slow-moving walkway), they have plenty of time to interact, making transmission much more likely.

2. The Three Forces of Movement

The paper describes how people move using three "forces":

The "Drunkard's Walk" (Regression to the Mean): People naturally tend to wander randomly, but they also have a tendency to slow down and align with the average speed of the crowd around them.
The "Avoidance" Instinct: Healthy people can sense where the sick people are. If they see a cluster of sick people, they might change direction to steer clear of them. (The paper notes this is a "preferred direction" that healthy people try to follow).
The "State Change": This is the actual disease part. If a healthy person and a sick person are close and moving together, the healthy person might turn sick. If a sick person rests, they might recover. If a recovered person loses their immunity, they become healthy again.

3. Zooming Out: From Video to Soup

The authors show that if you take this detailed "video" model and zoom out far enough (mathematically speaking, by looking at the average behavior over time and space), it turns back into the classic, simple "soup" model (called the SIRS model) that epidemiologists have used for decades.

The Metaphor: Think of the kinetic model as a high-speed camera recording every drop of water in a river. If you stop the camera and just look at the river's flow from a helicopter, you see the smooth, simple current. The paper proves that their complex, drop-by-drop model perfectly matches the smooth, simple river model when you look at the big picture.

4. What They Proved

The authors did two main things with their math:

Consistency: They proved that their fancy, detailed model is mathematically compatible with the old, simple model. If you remove the movement details, you get the standard disease equations back.
Existence: They proved that their complex equations actually have valid solutions. In plain English, this means the model doesn't break or produce nonsense numbers; it describes a scenario that is mathematically possible to solve.

5. A Special Example: The "Cowardly" Sick

The paper also looked at a specific scenario where sick people might voluntarily slow down or stop moving (like staying home when they feel ill).

The Result: They found that if sick people move much slower than healthy people, the disease spreads less intensely. It's like if the "sick" dancers on the floor suddenly sat down; the healthy dancers would zip past them without much contact, slowing the spread of the virus.

Summary

This paper builds a bridge between two worlds: the complex, detailed world of how individuals move and interact, and the simple, broad world of standard disease statistics. It shows that if you understand the "dance moves" of individuals (speed, direction, and avoidance), you can derive the standard rules of epidemics, while also gaining the ability to study how things like movement restrictions or avoidance behaviors change the outcome of an outbreak.

Technical Summary: Kinetic Formulation of Compartmental Epidemic Models

Problem Statement
The paper addresses the need to bridge the gap between microscopic individual behaviors (movement and interaction) and macroscopic compartmental epidemic models (such as SIRS). Traditional compartmental models, typically formulated as systems of Ordinary Differential Equations (ODEs), assume spatial homogeneity and do not account for the specific mechanics of transmission, such as the duration of contact or the relative velocity of individuals. While kinetic models have been applied to biological phenomena like chemotaxis and cancer dynamics, their application to epidemiology requires a formulation where disease transmission is explicitly linked to the phase-space density of individuals (position and velocity), rather than just spatial density. The authors aim to construct a kinetic model that couples population movement with pathogen dynamics and rigorously demonstrate its compatibility with the classical SIRS model under specific limits.

Methodology
The authors introduce a kinetic model where the population is divided into three classes: Susceptible ( $S$ ), Infectious ( $I$ ), and Recovered ( $R$ ). The state of the system is described by density functions $f_j(t, x, v)$ for $j \in \{S, I, R\}$ , representing the distribution of individuals in space ( $x \in \Omega$ ) and velocity ( $v \in V$ ).

Kinetic Equation: The evolution of $f_j$ is governed by a Boltzmann-type equation:
$\partial_t f_j + v \cdot \nabla_x f_j = Q_j[f]$
The transition kernel $Q_j[f]$ is decomposed into three components:
- Regression ( $Q^{(R)}$ ): A mean-reversion term driving velocities toward the mean, ensuring conservation of mass.
- Preferred Direction ( $Q^{(P)}$ ): A drift term modeling the tendency of susceptible individuals to move away from regions of high infectious density (avoidance behavior).
- Class Transition ( $Q^{(T)}$ ): A non-conservative term inspired by evolutionary game theory (specifically replicator dynamics) and the SIRS model. Crucially, transmission ( $S \to I$ ) occurs only when an $S$ and an $I$ individual share the same position and velocity (phase-space alignment), reflecting the requirement for prolonged close contact.
Macroscopic Limit (Hydrodynamic Limit): The authors perform a formal asymptotic analysis using a small dimensionless parameter $\varepsilon$ , representing the ratio of microscopic velocity to macroscopic displacement speed. They assume an expansion of the kernel $Q^{(\varepsilon)}_j = Q^{(R)}_j + \varepsilon Q^{(P)}_j + \varepsilon^2 Q^{(T)}_j + o(\varepsilon^2)$ .
- Order $\varepsilon^0$ : Yields a local equilibrium where the velocity distribution is spherically symmetric.
- Order $\varepsilon^1$ : Determines the first-order correction to the distribution, introducing drift terms dependent on spatial gradients of density.
- Order $\varepsilon^2$ : By integrating the equation over velocity space, the authors derive a system of Partial Differential Equations (PDEs) for the spatial densities $\rho_j = \int_V f_j dv$ . This "mesoscopic model" is of the advection-diffusion-reaction type.
Existence and Uniqueness: The authors prove the existence and uniqueness of global solutions for the kinetic system in $L^p$ spaces. This proof is conducted under the technical assumption that the preferred direction kernel is zero ( $\eta = \xi = 0$ ), i.e., without the avoidance drift, to avoid complications arising from kernels dependent on solution gradients.

Key Contributions and Results

Derivation of the Mesoscopic Model: The paper derives a specific PDE system (Eqs. 10–12) that describes the evolution of $S$ , $I$ , and $R$ densities. This system includes diffusion terms ( $D \nabla_x \rho_j$ ) and a drift term for susceptible individuals ( $\Gamma \nabla_x \rho_I \rho_S$ ) that models the avoidance of infectious individuals.
Recovery of the SIRS Model: The authors demonstrate that for spatially homogeneous solutions (where $\nabla_x \rho_j = 0$ ), the derived mesoscopic PDE system reduces exactly to the classical SIRS ODE system. This formally validates the kinetic model as a generalization of the standard compartmental model.
Mathematical Rigor: The paper establishes the existence and uniqueness of solutions for the kinetic model in appropriate function spaces for the case without avoidance drift.
Numerical and Analytical Examples:
- A numerical simulation illustrates the dispersion of particles and the transition dynamics.
- An analysis of constant parameters shows how the diffusion and drift coefficients relate to the underlying velocity distribution.
- A 1D example with distinct velocity sets for different classes models the voluntary confinement of infectious individuals (reducing $v_I$ ), demonstrating that this confinement reduces the diffusion coefficient of the infectious class and thus limits disease spread.

Significance and Claims
The authors claim that this work provides a rigorous kinetic foundation for compartmental epidemic models, explicitly linking individual motility and contact duration to macroscopic transmission rates. By framing the class transitions within the context of evolutionary game theory (replicator dynamics), the model offers a pathway to generalize epidemic modeling beyond standard SIRS structures.

The paper modestly positions itself as an initial exploration of the interplay between kinetic theory, mathematical epidemiology, and evolutionary game theory. It acknowledges limitations, such as the restriction of the existence proof to the case without avoidance drift and the lack of a rigorous proof for the convergence of the kinetic solution to the mesoscopic limit as $\varepsilon \to 0$ . The authors suggest that future work could explore more general transition kernels, the inclusion of vaccination strategies based on local and global information, and the potential for pattern formation or finite-time blow-up in systems with attractive (rather than repulsive) interactions.