🌀 nonlinear sciences

Epidemic Transmission Modelling on the Birth-death Evolving Network with Indirect Contacts

This paper investigates the impact of a birth-death evolving network, which models population migration and heritable node deletion to capture indirect contacts, on SIRS epidemic dynamics, revealing that these network properties facilitate epidemic transmission through newly created indirect contacts.

Original authors: Minyu Feng, Yuhan Li, Jürgen Kurths

Published 2026-04-14

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Minyu Feng, Yuhan Li, Jürgen Kurths

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

The Big Picture: A Moving, Changing City

Imagine a bustling city where people are constantly arriving and leaving. In this city, people get sick, recover, and can get sick again.

Most old studies on how diseases spread treated this city like a frozen statue. They assumed the population was fixed, the roads (connections) never changed, and people stayed in the same spots forever.

This paper says: "That's not how the real world works!"

Instead, the authors built a model of a living, breathing city that changes while the disease is spreading. They call this a "Birth-Death Evolving Network."

Birth: New people move into the city.
Death: People move out of the city.
Evolving: The connections between people change as people come and go.

The Secret Ingredient: The "Inheritance" of Connections

The most unique part of this paper is a concept they call "Indirect Contacts" or "Heritable Disconnections."

The Analogy: The Musical Chairs of Friendship
Imagine a group of friends playing a game.

The Old Way: If one friend (let's call him Bob) leaves the game, he just disappears. His connections to Alice and Charlie are cut. Alice and Charlie no longer know each other through Bob.
The New Way (This Paper): When Bob leaves, he doesn't just vanish. He leaves behind his "social legacy." He introduces Alice and Charlie to a new person, Dave, who takes over Bob's role. Now, Dave is friends with Alice and Charlie.

Why does this matter for a virus?
In the old model, when Bob leaves, the virus loses a bridge. In this new model, the virus gets a new bridge. Even though Bob is gone, the connection he used to have is "inherited" by someone else. This creates indirect contacts—new paths for the virus to jump on, even if the original person is gone.

How They Studied It: The "Queueing System"

To understand the math, the authors used a clever analogy: A Bank with Three Lines.

Imagine a bank with three service counters:

Counter S (Susceptible): People who are healthy but can get sick.
Counter I (Infected): People currently sick.
Counter R (Recovered): People who got better and have temporary immunity.

People moving in: New customers arrive at the bank (migrants entering the city).
People moving out: Customers leave the bank (people moving away).
Moving between counters:
- If a healthy person (S) talks to a sick person (I), they get sick and move to Counter I.
- If a sick person (I) gets better, they move to Counter R.
- If a recovered person (R) loses their immunity, they go back to Counter S.

The authors used a "Markovian Queueing Network" (a fancy math term for tracking these lines) to calculate how many people would be in each line at any given time, considering that the bank itself is growing and shrinking.

What They Discovered

1. The "Inheritance" Makes the Virus Stronger
They found that when people leave and their connections are "inherited" by others, the virus spreads easier and faster.

Think of it like this: If a fire breaks out in a house, and the house is demolished, the fire usually stops. But in this model, when the house is demolished, the fire instantly jumps to a new house that inherits the old house's neighbors. The fire (virus) never really loses its footing.

2. The "Threshold" Drops
There is a "tipping point" (threshold) for how contagious a virus needs to be to cause a massive outbreak.

In a static city, the virus needs to be very contagious to take over.
In this "moving city" with indirect contacts, the virus needs to be less contagious to cause a massive outbreak. The constant reshuffling of connections keeps the virus alive even when it should have died out.

3. Migration Matters
The speed at which people move in and out of the city changes everything.

If people move too fast, the virus might not have time to spread.
If people move at a "Goldilocks" speed (not too fast, not too slow), the virus finds a perfect rhythm to keep spreading globally.

The Takeaway

This paper teaches us that we can't just count how many people are sick; we have to watch how the network of people changes.

If you are trying to stop a pandemic, you can't just treat the sick people. You also have to understand that when people move away, they leave behind "ghost connections" that can still spread the disease. By ignoring these indirect contacts, we might think a disease is under control when it's actually just waiting for the next person to pick up the baton.

In short: The virus is like a game of "telephone" where the players keep swapping seats and passing the message to new people. If you don't account for the seat-swapping, you'll never understand how the message (the virus) traveled so far.

1. Problem Statement

Traditional epidemic modeling often assumes that the underlying contact network is static or evolves on a time scale significantly slower than the disease propagation. However, in real-world scenarios, network structures and epidemic dynamics co-evolve on the same time scale. Furthermore, individual behaviors such as population migration (inflow/outflow) and the creation of indirect contacts (where a departing individual's connections are inherited by others) are often overlooked.

This paper addresses the gap in modeling the co-evolution of epidemics and networks, specifically focusing on:

How birth-death processes (random node arrival and departure) affect epidemic dynamics.
The impact of heritable disconnections (indirect contacts) where a leaving node's edges are transferred to a new "inheritor" node, altering the contact structure dynamically.
The quantification of how these network dynamics influence the population sizes of Susceptible ( $S$ ), Infected ( $I$ ), and Recovered ( $R$ ) states.

2. Methodology

A. Network Model: Birth-Death Evolving Network with Heritable Disconnections

The authors propose a dynamic network model where:

Node Dynamics: Nodes enter (birth) and leave (death) the network following Poisson processes with rates $\lambda$ (inflow) and $\mu$ (outflow). Node lifespans follow an exponential distribution.
Connection Mechanism: New nodes attach to existing nodes via preferential attachment (Barabási-Albert style) with $m$ edges.
Heritable Disconnection: When a node leaves, it is not simply removed with its edges. Instead, its connections are inherited by an existing "inheritor" node (selected randomly in this study). This creates indirect contacts, effectively bridging the inheritor to the departed node's former neighbors.

B. Epidemic Model: SIRS on a Markovian Queueing Network

The epidemic process is modeled using the Susceptible-Infected-Recovered-Susceptible (SIRS) framework, integrated with the network dynamics:

Transitions: $S \to I$ (infection rate $\beta$ ), $I \to R$ (recovery rate $\gamma$ ), $R \to S$ (loss of immunity rate $\alpha$ ).
Queueing Analogy: The system is modeled as a Markovian queueing network.
- Epidemic states ( $S, I, R$ ) act as "service centers."
- Individuals act as "customers."
- Transitions between states and inflows/outflows are treated as service completions and arrivals/departures.
Mathematical Formulation: The authors derive transition probabilities for the population size of each state ( $S(t), I(t), R(t)$ ) over a time interval $\Delta t$ . They formulate deterministic differential equations for the system and apply the Next Generation Matrix method to derive the basic reproductive number ( $\mathcal{R}_0$ ).

3. Key Contributions

Unified Co-evolution Framework: The paper establishes a unified model where network topology evolution (birth/death/inheritance) and epidemic spreading occur simultaneously on the same time scale, moving beyond the static or slow-evolving network assumptions.
Indirect Contact Mechanism: It formally introduces and models "heritable disconnections" as a mechanism for indirect contacts. This captures real-world phenomena where a person leaving a region (e.g., migration) leaves behind a social structure that is inherited by others, potentially creating new transmission pathways.
Stochastic Analysis via Queueing Theory: By mapping the epidemic process to a Markovian queueing network, the authors derive analytical expressions for the probability of population size changes and the stationary reproductive number ( $\mathcal{R}_0$ ) in a dynamic environment.
Theoretical Threshold Derivation: The paper provides a closed-form expression for $\mathcal{R}_0$ that incorporates network evolution parameters ( $\lambda, \mu$ ) and inheritance mechanisms, offering a theoretical basis for predicting outbreak thresholds in evolving populations.

4. Key Results

A. Impact of Migration and Network Parameters

Stationary States: The system reaches a stationary state where population sizes fluctuate around specific values if the ratio of inflow to outflow ( $\lambda/\mu$ ) is constant.
Parameter Sensitivity:
- Infection Rate ( $\beta$ ): Higher $\beta$ accelerates the time to reach stationarity and increases the final infected population. A critical threshold exists; below it, the epidemic dies out locally; above it, it spreads globally.
- Recovery Rate ( $\gamma$ ) & Reviving Rate ( $\alpha$ ): These parameters have opposing effects on the $I$ and $R$ populations. Higher recovery rates reduce the infected population but increase the recovered population.

B. Effect of Indirect Contacts (Heritable Disconnections)

Facilitation of Transmission: Simulations comparing models with and without indirect contacts reveal that indirect contacts significantly facilitate epidemic transmission.
Lowered Threshold: The presence of indirect contacts lowers the epidemic threshold for the infection rate ( $\beta$ ).
Increased Prevalence: Under stationary conditions, models with indirect contacts maintain a larger infected population size compared to models where departing nodes simply sever all connections.
Global vs. Local Spread: Indirect contacts allow the epidemic to persist and spread globally even at infection rates where a standard model (without inheritance) would result in the disease dying out locally.

C. Validation of $\mathcal{R}_0$

Numerical simulations confirmed the theoretical $\mathcal{R}_0$ derived in Theorem 2.
When the infection rate $\beta$ was below the theoretical critical value ( $\beta_c \approx 0.00088$ ), the final infected population was zero.
When $\beta > \beta_c$ , a positive, stable number of infected individuals emerged, validating the theoretical threshold.

5. Significance and Conclusion

This research provides a critical advancement in epidemiological modeling by integrating network dynamics and individual behavioral changes (migration and indirect contact inheritance) into a single stochastic framework.

Practical Implications: The findings suggest that in real-world scenarios involving high population mobility (e.g., tourism, migration, or temporary workforces), the "inheritance" of social ties by remaining individuals can act as a hidden vector for disease transmission. Public health strategies must account for these indirect contacts, as they can lower the threshold for outbreaks and sustain epidemics longer than predicted by static models.
Theoretical Value: The use of Markovian queueing networks to model SIRS dynamics on evolving graphs offers a robust mathematical tool for analyzing complex, time-varying biological systems.
Future Work: The authors acknowledge limitations, such as the random selection of inheritor nodes (which could be refined based on similarity metrics) and the challenge of deriving explicit stationary distributions for non-homogeneous Markov processes. Future work will aim to address these analytical challenges.