Optimal Control of an SIR Model with Noncompliance as a Social Contagion

Chloe Ngo, Christian Parkinson, and Weinan Wang · 2025

Read Optimal Control of an SIR Model with Noncompliance as a Social Contagion by Chloe Ngo, Christian Parkinson, and Weinan Wang (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2509-09075.txt.

Introduction

Chloe Ngo and 2 co-authors study optimal control of an SIR disease model where noncompliance spreads socially in a 2025 paper. SIR splits a population into susceptible, infected, and recovered groups. Here people also split into those who follow prevention measures and those who do not.

Noncompliance itself spreads like a contagion alongside the disease. The authors derive the reproductive ratio and analyze disease free equilibria. A policymaker can use mandates, treatment, and education campaigns as controls to minimize a cost function. Existence of optimal controls is proved for general costs. Numerical tests show how optimal strategy depends on policymaker preferences.

Catalog record

CREATORChloe Ngo, Christian Parkinson, and Weinan Wang
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE68 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2509-09075

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Optimal Control of an SIR Model with Noncompliance as a Social Contagion Chloe Ngo, Christian Parkinson, and Weinan Wang Abstract . We propose and study a compartmental model for epidemiology with human behavioral effects. Specifically, our model incorporates governmental prevention measures aimed at lowering the disease infection rate, but we split the population into those who comply with the measures and those who do not comply and therefore do not receive the reduction in infectivity. We then allow the attitude of noncompliance to spread as a social contagion parallel to the disease. We derive the reproductive ratio for our model and provide stability analysis for the disease-free equilibria. We then propose a control scenario wherein a policy-maker with access to control variables representing disease prevention mandates, treatment efforts, and educational campaigns aimed at encouraging compliance minimizes a cost functional incorporating several cost concerns. We characterize optimal controls via the Pontryagin optimality principle and present simulations which demonstrate the behavior of the control maps in several different parameter regimes. 1. Introduction In this paper, we are interested in incorporating human behavioral effects into a Susceptible-Infected-Recovered (SIR) type compartmental model for epidemic spread of a generic infectious disease. At the outset of an epidemic, governments will enact public health protocols meant to slow disease spread. However, adherence with non-pharmaceutical intervention (NPI) measures such as mask-wearing, shelter-at-home mandates, and social distancing is certainly imperfect. For example, Dolan provides an interesting account of opposition to prolonged public health protocols during the 1918 Flu epidemic and the actions of the so-called “anti-mask league” [ 20 ] , and in a study by the Pew Research Center from June 4-10, 2020 during the height of the COVID 19 pandemic, 16% of 9654 Americans surveyed reported that they “hardly ever” or “never” wear a mask [ 23 ] . Likewise, the Angus Reid Institute reported on July 15, 2020 that 45% of 1503 Canadians surveyed responded that they “rarely” or “never” wear masks, and among the 1202 participants who reported that they do not always wear masks, 15% reported that their main reason for not wearing a mask is that they believe masks are ineffective and 11% reported their main reason for not wearing a mask as “no one else is wearing them” [ 1 ] . Other studies reported similar levels of nonadherence with several types of NPIs [ 5 , 10 , 33 ] . From reports like this, it appears that there will always be nontrivial subpopulations who do not comply with NPI measures, and given evidence that NPI measures have a positive effect on epidemic outcomes [ 30 ] , accounting for this noncompliance is key to increasing fidelity in mathematical models of epidemiology. Accordingly, we propose a compartmental model for epidemiology wherein the population is split into those who comply with public health protocols (dubbed the compliant population), and those who do not comply (dubbed the noncompliant population). The disease is assumed to spread slower among the compliant population. Social contagion theory [ 12 ] posits that attitudes, opinions and behaviors can spread among a population in a manner similar to that of an infectious disease, and has been used to model prescription and illicit drug use [ 11 , 16 , 22 ] , alcohol consumption [ 48 ] , governmental corruption [ 32 ] , adverse mental health conditions such as depression [ 40 ] , and public sentiment [ 31 ] in a variety of contexts including opinions regarding anti-smoking legislation [ 35 ] and, importantly, vaccine and treatment uptake during epidemics [ 9 , 24 ] . Given this, and the aforementioned study wherein people reported not wearing masks because “no one else is wearing them” [ 1 ] , we suppose that the attitude of noncompliance with NPIs spreads throughout the population as a social contagion. We further imagine a policy-maker with several different options for epidemic control who aims to minimize a cost functional which trades off costs for increased infections and increased noncompliance with economic costs for increased use of controls. Incorporation and assessment of the effects of heterogeneous behavior into epidemiological models is becoming increasingly popular (see, for example, [ 4 , 21 , 27 , 34 , 36 ] , or [ 43 ] where the authors propose a rather general framework similar to the modeling we use below, though with linear transfer between behavior classes). Likewise, there are several recent studies of optimal control for epidemiological models in different modeling scenarios, such as optimal treatment or vaccination [ 6 , 13 , 26 , 50 ] , optimal quarantine under a variety of limitations [ 3 , 15 ] , and assessment of the effects of contact heterogeneity [ 14 ] or the use of different types of cost functionals [ 44 ] . However, very few studies include both optimal control for epidemiology and mechanistic modeling of human behavior. One exception is [ 8 ] which takes a more theoretical look at a system of controlled partial differential equations describing an epidemic. Other studies incorporate control and human behavior more implicitly, by, for example, allowing infection rates to depend on ambient dynamics under the assumption that people will change their behavior based on local disease effects [ 37 ] or media coverage [ 18 ] . In this manuscript, we develop a compartmental epidemiological model which incorporates human behavior in an explicit, mechanistic manner. We then consider optimal control of the model and observe the differences in optimal control strategies in different parameter regimes. The remainder of the manuscript is organized as follows. In section 2 , we introduce our model, derive the reproductive ratio, and discuss stability of disease-free equilibria (though a full proof of our stability theorem is relegated to appendix A ). In section 3 , we introduce the optimal control problem that we will solve and give a characterization of the optimal control maps via the Pontryagin optimality principle. In section 4 , we discuss the results of simulation of our model in different parameter regimes. Finally, section 5 includes a brief conclusion and discussion of avenues for future work. 2. An SIR Model with Behavioral Effects: Disease-Free Equilibria and Reproductive Ratios As with all ordinary and partial differential equation (O/PDE) based epidemiological models, we build off of the classical SIR model of Kermack and McKendrick [ 25 ] . We assume a constant birth rate b>0 into the susceptible population, and that deaths occur proportional to population sizes with natural rate \delta>0 . The infection spreads via mass action with rate \beta>0 and infectious individuals recover with rate \gamma>0 . This leads to the SIR model (2.1) \begin{split}\frac{dS}{dt}&=b-\beta SI-\delta S,\\ \frac{dI}{dt}&=\beta SI-(\gamma+\delta)I,\\ \frac{dR}{dt}&=\gamma I-\delta R.\end{split} As mentioned above, we incorporate human behavior in a manner inspired by social contagion theory. Each population is divided into a compliant portion (denoted by S,I,R ) and a noncompliant portion (denoted by S^{*},I^{*},R^{*} ), and we assume that, parallel to the disease, noncompliance spreads via a second mass action term. With this assumption, we propose the following model: (2.2) \begin{split}\frac{dS}{dt}&=(1-\xi)b-\beta(1-\alpha(t))S(I+I^{*})-(\overline{\mu}-\mu(t))SN^{*}+\nu(t)S^{*}-\delta S,\\ \frac{dI}{dt}&=\beta(1-\alpha(t))S(I+I^{*})-(\gamma+\eta(t))I-(\overline{\mu}-\mu(t))IN^{*}+\nu(t)I^{*}-\delta I,\\ \frac{dR}{dt}&=(\gamma+\eta(t))I-(\overline{\mu}-\mu(t))RN^{*}+\nu(t)R^{*}-\delta R,\\ \frac{dS^{*}}{dt}&=\xi b-\beta S^{*}(I+I^{*})+(\overline{\mu}-\mu(t))SN^{*}-\nu(t)S^{*}-\delta S^{*},\\ \frac{dI^{*}}{dt}&=\beta S^{*}(I+I^{*})-\gamma I^{*}+(\overline{\mu}-\mu(t))IN^{*}-\nu(t)I^{*}-\delta I^{*},\\ \frac{dR^{*}}{dt}&=\gamma I^{*}+(\overline{\mu}-\mu(t))RN^{*}-\nu(t)R^{*}-\delta R^{*}.\end{split} Here the birth rate

Read the complete text