Optimal Curing Strategy for Competing Epidemics Spreading Over Complex Networks

Juntao Chen, Yunhan Huang, Rui Zhang, Quanyan Zhu

TL;DR

This paper develops a mathematical framework for finding the optimal "curing" strategy to suppress two competing epidemics spreading across a complex network. The authors prove a "non-coexistence" phenomenon—where only one strain can survive at a steady state—and provide a gradient-descent algorithm to balance the economic cost of treatment against the social cost of disease severity.

What Problem Does This Paper Address?

Most epidemic modeling focuses on a single pathogen. However, in the real world, multiple "strains" often compete for the same susceptible population. This applies to biology (different influenza strains), cybersecurity (competing malware), and even economics (competing products in viral marketing).

The challenge is that these processes are interdependent: an individual infected by Strain A cannot be simultaneously infected by Strain B. Controlling such a system is complex because a treatment that suppresses one strain might inadvertently create an opening for the other to dominate. The authors seek to determine the most cost-effective way to allocate recovery resources (like vaccines or anti-virus software) to drive the network toward a disease-free state or a manageable equilibrium.

Methods

The researchers utilize a Degree-Based Mean-Field (DBMF) approximation of the Susceptible-Infected-Susceptible (SIS) model. Unlike models that track every individual link, DBMF groups nodes by their "degree" (number of connections), making it computationally feasible for massive, large-scale networks.

Key technical components include:

  • Stability Analysis: Using Lyapunov functions and eigenvalue analysis of linearized systems to determine when the network will settle into a disease-free state versus an endemic state.
  • Optimization Framework: They formulate an objective function that minimizes the sum of the control cost (the effort to cure nodes) and the epidemic severity (the total infected population).
  • Fixed-Point Iterative Scheme: Because the steady-state equations are non-linear and coupled, the authors developed a gradient-descent algorithm that uses a fixed-point iteration to find the optimal curing rates ($u_1, u_2$).

Datasets

As a theoretical control-theory paper, the authors do not use a "training" dataset in the machine-learning sense. Instead, they validate their results using:

  • Synthetic Scale-Free (SF) Networks: Generated using the Barabási-Albert model with 500 nodes. These networks mimic real-world social and computer networks where a few "hubs" have many connections while most nodes have few.
  • Statistical Parameters: They use an average degree ($\langle k \rangle$) of 1.996 and a second moment ($\langle k^2 \rangle$) of 13.75 to simulate the network's connectivity properties.

Key Insights and Ideas

  • The Non-Coexistence Phenomenon: Perhaps the most significant theoretical finding is that at a steady state, two competing strains cannot coexist. The network will either become disease-free or settle into an "exclusive equilibrium" where only the more "fit" or less-controlled strain survives.
  • Distribution Independence in the Disease-Free Regime: The authors found that if the goal is to completely eradicate the diseases, the optimal strategy depends only on the average degree and the second moment of the network, not the specific shape of the entire degree distribution.
  • Equilibria Switching: As the unit cost of curing changes (e.g., as a vaccine becomes cheaper), the optimal strategy can cause the network to "jump" or switch between equilibria—for example, moving from a state where Strain 1 is dominant to one where Strain 2 is dominant, before finally reaching a disease-free state.
  • The Fulfilling Threshold: They formally define a "fulfilling threshold"—the specific level of control effort beyond which the network is guaranteed to reach a stable, disease-free equilibrium.

Key Results

  • Global Optimality: The proposed algorithm successfully identifies the curing strategy that minimizes total costs across different epidemic regimes.
  • Robustness: The authors proved that the epidemic's severity is continuous with respect to the curing effort. This means small errors in applying the control strategy won't lead to catastrophic, sudden outbreaks.
  • Model Mismatch Resilience: Numerical experiments showed that even if the controller has a slightly inaccurate model of the network (e.g., wrong connectivity statistics), a conservative control design can still successfully drive the system to a disease-free state.

Limitations and Future Work

  • Steady-State Focus: The current work focuses on the "long run" (steady state). It does not explicitly optimize for the transient phase—the period immediately following an outbreak where rapid response is most critical.
  • Binary Competition: The model assumes "perfect" competition (one infection blocks the other). It does not account for "co-infection," where a host could carry both strains simultaneously.
  • Two-Strain Limit: The framework is currently built for two competing processes. Extending this to $N$ strains or more complex interdependencies (e.g., Strain A helps Strain B spread) remains an open challenge.
  • Perfect Knowledge: The authors assume the network operator knows the degree distribution. Future work could explore "blind" or adaptive curing strategies where the network structure is partially unknown.

Summary updated

Abstract

Optimal curing strategy of suppressing competing epidemics spreading over complex networks is a critical issue. In this paper, we first establish a framework to capture the coupling between two epidemics, and then analyze the system's equilibrium states by categorizing them into three classes, and deriving their stability conditions. The designed curing strategy globally optimizes the trade-off between the curing cost and the severity of epidemics in the network. In addition, we provide structural results on the predictability of epidemic spreading by showing the existence and uniqueness of the solution. We also demonstrate the robustness of curing strategy by showing the continuity of epidemic severity with respect to the applied curing effort. A gradient descent algorithm based on a fixed-point iterative scheme is proposed to find the optimal curing strategy. Depending on the system parameters, the curing strategy can lead to switching between equilibria of the epidemic network as the control cost varies. Finally, we use case studies to corroborate and illustrate the obtained theoretical results.
BibTeX
@article{chen2021optimal,
  author = {Juntao Chen and Yunhan Huang and Rui Zhang and Quanyan Zhu},
  title = {{Optimal Curing Strategy for Competing Epidemics Spreading Over Complex Networks}},
  journal = {IEEE Trans. Signal Inf. Process. Over Networks},
  year = {2021},
  url = {https://openreview.net/forum?id=VRPCyDolfr},
}

Loading related content...

Figure