Explore Research Products in the PAR
It may take a few hours for recently added research products to appear in PAR search results.


Title: Understanding LLMs' Capabilities to Support Spatially-disaggregated Epidemiological Simulations
Evaluating public health interventions during disease outbreaks requires an understanding of the spatial patterns underlying epidemiological processes. In this study, we explore how Large Language Models (LLMs) can leverage spatial understanding and contextual reasoning to support spatially-disaggregated epidemiological simulations. We present an approach in which a system dynamics model queries an LLM at key decision points to determine appropriate mitigation strategies, informed by local profiles and the current outbreak status, and incorporates these strategies into the simulations. Through a series of experiments with COVID-19 data from San Diego County, we show how different LLMs perform in tasks requiring spatial adaptation of mitigation strategies, and how incorporating connectivity information through Retrieval-Augmented Generation (RAG) enhances the performance of these customizations. The results reveal significant differences among LLMs in their ability to account for spatial structure and optimize mitigation strategies accordingly. This highlights the importance of selecting the right model and enhancing it with relevant contextual information for effective public health interventions.  more » « less
Award ID(s):
2200197
PAR ID:
10561167
Author(s) / Creator(s):
; ; ; ;
Publisher / Repository:
Zenodo
Date Published:
Subject(s) / Keyword(s):
Large Language Models (LLMs) Epidemiological Modeling Spatial Reasoning System Dynamics San Diego
Format(s):
Medium: X
Right(s):
Creative Commons Attribution 4.0 International
Sponsoring Org:
National Science Foundation
More Like this
  1. ; ; ; ; (, Zenodo)
    Evaluating public health interventions during disease outbreaks requires an understanding of the spatial patterns underlying epidemiological processes. In this study, we explore how Large Language Models (LLMs) can leverage spatial understanding and contextual reasoning to support spatially-disaggregated epidemiological simulations. We present an approach in which a system dynamics model queries an LLM at key decision points to determine appropriate mitigation strategies, informed by local profiles and the current outbreak status, and incorporates these strategies into the simulations. Through a series of experiments with COVID-19 data from San Diego County, we show how different LLMs perform in tasks requiring spatial adaptation of mitigation strategies, and how incorporating connectivity information through Retrieval-Augmented Generation (RAG) enhances the performance of these customizations. The results reveal significant differences among LLMs in their ability to account for spatial structure and optimize mitigation strategies accordingly. This highlights the importance of selecting the right model and enhancing it with relevant contextual information for effective public health interventions. 
    more » « less
  2. ; ; ; ; ; ; (, SSRN)
    Incorporating human behavior into epidemiology models is increasingly realized as a useful way to assess the effectiveness of public health policies. Due to limited agreement on how individuals perceive, process, and respond to risk, standard disease models often implicitly assume individuals rationally respond to disease risk and accurately process information about the risks they face, neglecting important psychological processes. We present a cognitive epidemiological model framework supplemented with state variables representing human risk perception and behavior to explore how risk perception influences epidemiological and behavioral dynamics, and how these processes interact with public health policy strategies. We consider public policies which target top-down effects (directly mandating epidemic-controlling actions) or bottom-up effects (encouraging change by influencing people's behavior and risk perception), or consist of multiple policy effects simultaneously. Top-down policies may have unintentional indirect bottom-up effects, which we show can lead to surprising interactions in how top-down and bottom-up policy actions combine to shape epidemics. Our model shows that different policy types produce distinct epidemiological and behavioral dynamics, with policies targeting bottom-up change better able to delay epidemic peaks but at the cost of greater vulnerability to the detrimental effects of mistimed interventions and people free-riding off others' protective behaviors. These results emphasize the benefits of collaboration between epidemiologists and social scientists to parameterize the behavioral and cognitive aspects of models like ours, which can enhance predictions and provide more nuanced policy recommendations. 
    more » « less
  3. ; ; ; (, BMC Public Health)
    Abstract Background Individual behavioural decisions are responses to a person’s perceived social norms that could be shaped by both their physical and social environment. In the context of the COVID-19 pandemic, these environments correspond to epidemiological risk from contacts and the social construction of risk by communication within networks of friends. Understanding the circumstances under which the influence of these different social networks can promote the acceptance of non-pharmaceutical interventions and consequently the adoption of protective behaviours is critical for guiding useful, practical public health messaging. Methods We explore how information from both physical contact and social communication layers of a multiplex network can contribute to flattening the epidemic curve in a community. Connections in the physical contact layer represent opportunities for transmission, while connections in the communication layer represent social interactions through which individuals may gain information, e.g. messaging friends. Results We show that maintaining focus on awareness of risk among each individual’s physical contacts promotes the greatest reduction in disease spread, but only when an individual is aware of the symptoms of a non-trivial proportion of their physical contacts (~ ≥ 20%). Information from the social communication layer without was less useful when these connections matched less well with physical contacts and contributed little in combination with accurate information from physical contacts. Conclusions We conclude that maintaining social focus on local outbreak status will allow individuals to structure their perceived social norms appropriately and respond more rapidly when risk increases. Finding ways to relay accurate local information from trusted community leaders could improve mitigation even where more intrusive/costly strategies, such as contact-tracing, are not possible. 
    more » « less
  4. ; ; ; (, ACM)
    Recent advances in large-scale human mobility datasets have opened new opportunities to improve public health through data-driven strategies, advanced computational methods, and interdisciplinary approaches. A key focus in epidemiological research is the estimation and analysis of social contact patterns, representing the frequency and nature of interactions among different demographic groups. These patterns are vital for modeling disease transmission, evaluating public health interventions, and guiding resource allocation. However, obtaining accurate and representative contact data remains a major challenge. In this paper, we propose a novel, scalable framework for generating and analyzing large-scale social contact datasets derived from foot-traffic data. Our approach integrates statistical modeling to estimate demographic distributions, such as age groups, at millions of points of interest (POIs) across the United States and globally, including restaurants, stores, hospitals, and schools. This framework enables actionable insights to inform public health strategies and improve population health outcomes. Moreover, the resulting datasets have broad cross-sector utility, supporting applications in strategic business planning, resource distribution, and personalized marketing and advertising. 
    more » « less
  5. ; ; ; ; (, BMC Public Health)
    Abstract BackgroundCOVID-19 mitigation strategies such as masking, social distancing, avoiding group gatherings, and vaccination uptake are crucial interventions to preventing the spread of COVID-19. At present, COVID-19 data are aggregated and fail to identify subgroup variation in Asian American communities such as Hmong Americans. To understand the acceptance, adoption, and adherence to COVID-19 mitigation behaviors, an investigation of Hmong Americans’ contextual and personal characteristics was conducted. MethodsThis study aims to describe COVID-19 mitigation behaviors among Hmong Americans and the contextual and personal characteristics that influence these behaviors. A cross-sectional online survey was conducted from April 8 till June 1, 2021, with Hmong Americans aged 18 and over. Descriptive statistics were used to summarize the overall characteristics and COVID-19 related behaviors of Hmong Americans. Chi-square and Fisher’s Exact Test were computed to describe COVID-19 mitigation behaviors by gender and generational status (a marker of acculturation). ResultsThe sample included 507 participants who completed the survey. A majority of the Hmong American participants in our study reported masking (449/505, 88.9%), social distancing (270/496, 55.3%), avoiding group gatherings (345/505, 68.3%), avoiding public spaces (366/506, 72.3%), and obtaining the COVID-19 vaccination (350/506, 69.2%) to stay safe from COVID-19. Women were more likely to socially distance (P = .005), and avoid family (P = .005), and social gatherings (P = .009) compared to men. Social influence patterns related to mitigation behaviors varied by sex. Men were more likely compared to women to be influenced by Hmong community leaders to participate in family and group gatherings (P = .026), masking (P = .029), social distancing (P = .022), and vaccination uptake (P = .037), whereas healthcare providers and government officials were social influencers for social distancing and masking for women. Patterns of social distancing and group gatherings were also influenced by generational status. ConclusionContextual and personal characteristics influence COVID-19 mitigation behaviors among English speaking Hmong Americans. These findings have implications for identifying and implementing culturally appropriate health messages, future public health interventions, policy development, and ongoing research with this population. 
    more » « less
  • Citation Formats
  • MLA
  • APA
  • Chicago
  • BibTeX
  • Text Encoded Conversion
  • XML
  • Save / Share this Record
  • Send to Email