Learning Fair Policies for Infectious Diseases Mitigation using Path Integral Control
Zhuangzhuang Jia Hyuk Park Gökçe Dayanıklı Grani A. Hanasusanto · 2025
Read Learning Fair Policies for Infectious Diseases Mitigation using Path Integral Control by Zhuangzhuang Jia Hyuk Park Gökçe Dayanıklı Grani A. Hanasusanto (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-2502-09831.txt.
Introduction
Zhuangzhuang Jia, Hyuk Park, Gökçe Dayanıklı, and Grani Hanasusanto present a 2025 paper on fair policies for fighting infectious disease. Vaccination and lockdown plans can protect some groups more than others. The authors build a decision framework that treats fairness as a goal alongside saving lives and money.
Their test bed is a stochastic disease model with several population groups. The learning method, based on path integral control, searches for action plans under uncertainty. Different measures of unfairness can be plugged into the goal. In a case study, the method produced vaccination and lockdown plans that were more even-handed than conventional ones. The authors offer the framework as a source of insight for policy makers.
Catalog record
| CREATOR | Zhuangzhuang Jia Hyuk Park Gökçe Dayanıklı Grani A. Hanasusanto |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 44 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2502-09831
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Learning Fair Policies for Infectious Diseases Mitigation using Path Integral Control Zhuangzhuang Jia Hyuk Park Gökçe Dayanıklı Grani A. Hanasusanto Abstract Infectious diseases pose major public health challenges to society, highlighting the importance of designing effective policies to reduce economic loss and mortality. In this paper, we propose a framework for sequential decision-making under uncertainty to design fairness-aware disease mitigation policies that incorporate various measures of unfairness. Specifically, our approach learns equitable vaccination and lockdown strategies based on a stochastic multi-group SIR model. To address the challenges of solving the resulting sequential decision-making problem, we adopt the path integral control algorithm as an efficient solution scheme. Through a case study, we demonstrate that our approach effectively improves fairness compared to conventional methods and provides valuable insights for policymakers. Keywords: stochastic optimal control; path integral control; fairness; SIR; public health 1 Introduction The COVID-19 pandemic has caused an unprecedented global impact, resulting in over 7 million deaths worldwide ( Mathieu et al. 2020 ) and causing extensive socioeconomic disruption—including $3.8 trillion in global consumption losses, the loss of 147 million full-time jobs, and a reduction in income amounting to $2.1 trillion ( Lenzen et al. 2020 ) . Moreover, numerous reports indicate that lower-income communities have been disproportionately affected by the pandemic. Rönkkö et al. 2022 report that the overall income of these communities during the pandemic fell further below pre-pandemic levels compared to the declines experienced by wealthier and average-income communities. In addition, the infection and mortality rates from COVID-19 have been significantly higher in these communities due to socioeconomic disparities, such as higher population density, limited access to private transportation, and reduced availability of timely medical treatments. This growing health inequality presents a critical societal challenge that policymakers must address ( Bambra et al. 2020 ) . To address this issue, our work aims to incorporate fairness consideration into the design of disease mitigation policies to ensure the equitable allocation of limited government resources across communities with different socioeconomic statuses . The proposed approach and our main contributions are summarized as follows: 1. We propose a novel framework for sequential decision-making under uncertainty that integrates a stochastic multi-group SIR model with various unfairness penalties, enabling policymakers to design fair vaccination and lockdown strategies. To the best of our knowledge, our work is the first to incorporate the consideration of fairness in the SIR model. 2. Solving the proposed sequential decision-making problem is computationally challenging with conventional methods, as the inclusion of unfairness penalties and the dynamics of the SIR model results in nonlinear partial differential equations (PDEs). To address the computational challenge, we utilize the path integral control algorithm, an optimization-free scheme that efficiently solves the problem by Monte Carlo sampling. 3. We conduct a numerical case study on COVID-19 to derive managerial insights. The results demonstrate that our region-specific policies significantly improve fairness across different socioeconomic regions compared to conventional homogeneous policies (across regions). Notably, our fairness-aware policies suggest prioritizing vaccination efforts in lower-income regions. 2 Related Work Modeling Infectious Disease Spread. The classical SIR model, introduced in Kermack and McKendrick 1927 , is a widely used dynamical system for modeling disease evolution, assuming deterministic and homogeneous dynamics across the entire population. To account for uncertainty in disease spread, various stochastic extensions of the SIR model have been developed ( Bartlett 1956 ; Bailey 1975 ; Beretta et al. 1998 ; Andersson and Britton 2012 ; Britton 2010 ; Kiss et al. 2017 ; Cai et al. 2015 ; Cai et al. 2017 ; Karako et al. 2020 ; Martcheva 2015 ; Allen 2017 ; Britton et al. 2019 ; Zhou et al. 2021 ; Laaribi et al. 2023 ; Greenwood and Gordillo 2009 ) . Aurell et al. 2022 explore a graphon game of epidemic control, with one special case involving multi-population modeling that considers different age and city groups. Their model examines the effects of different policies when individuals choose their own socialization level (control) in the game setup. However, their approach is based on an individualized version of multi-group SIR whereas our framework focuses on the group level. In Acemoglu et al. 2021 , a non-homogeneous (i.e., multi-group) version of the deterministic SIR model is considered, incorporating different age groups—young, middle-aged, and old—and deriving age-specific policies to minimize overall costs. In this paper, we extend their framework to the stochastic counterpart. In addition, while the motivation behind their multi-group approach is to derive group-specific policies, our method also considers fairness, which, to the best of our knowledge, has not been explored in existing work. Fairness in AI for Social Decision Making. With the advancement of AI, automated decision-making and policymaking have become increasingly prevalent in critical social domains such as college admissions, loan approvals, and criminal justice. However, these AI-driven systems have raised many concerns, as they may not be entirely objective and can even exacerbate existing human biases ( Dastin 2022 ; Angwin et al. 2022 ; Samorani et al. 2022 ) . Many studies have explored fairness considerations in AI-driven decision-making across various applications. For example, Azizi et al. 2018 study the problem of housing allocations for homeless youth and propose fair, efficient, and interpretable policies. In the context of micro-lending, Liu et al. 2019 develop a fairness-aware re-ranking algorithm that balances recommendation accuracy with borrower-side fairness while also accounting for lenders’ preferences for diversity. Similarly, Berk et al. 2021 provide an integrated examination of fairness and accuracy trade-offs in criminal justice risk assessments, demonstrating the inherent challenges in satisfying multiple fairness criteria simultaneously. Furthermore, Kallus et al. 2022 assess disparities in lending and healthcare applications when the protected class membership is not observed in the data. They provide exact characterizations of the tightest possible set of all true disparities that are consistent with the available data. Beyond these studies, many other works examine the broader challenges and trade-offs involved in designing fair AI systems across various domains ( Mouzannar et al. 2019 ; Corbett-Davies et al. 2023 ; Kleinberg et al. 2017 ; Dai et al. 2025 ; Aghaei et al. 2019 ; Raghavan et al. 2020 ; Chen et al. 2023 ; Jia et al. 2024 ; Wang et al. 2024 ; Liu et al. 2018 ; Baker and Hawn 2022 ; Taskesen et al. 2020 ; Bertsimas et al. 2013 ; Nguyen et al. 2021 ; Corbett-Davies et al. 2017 ; Rahmattalabi et al. 2022 ; Mashiat et al. 2022 ; Rahmattalabi et al. 2021 ; Freeman et al. 2020 ; Athanassoglou and Sethuraman 2011 ) . Path Integral Control. Path integral control has emerged as a promising solution scheme for solving a certain class of nonlinear stochastic optimal control problems ( Kappen 2005 ) . It has recently been applied in many reinforcement learning domains, including autonomous driving ( Mohamed et al. 2022 ; Williams et al. 2016 ; Gandhi et al. 2021 ; Ha et al. 2019 ) , robotics ( Theodorou et al. 2010 ; Chebotar et al. 2017 ; Williams et al. 2017b ; Yin et al. 2023 ; Park et al. 2024 ; Patil et al. 2022 ) , visual serving techniques ( Mohamed et al. 2021 ; Costanzo et al. 2023 ; Mohamed 2021