Challenges, tasks, and opportunities in modeling agent-based complex systems

Developing an Agent-Based Model that Predicts Listeria spp. Transmission to Assess Listeria Control Strategies in Retail Stores

Contamination of fresh produce with Listeria monocytogenes can occur throughout the supply chain, including at retail, where Listeria spp., including L. monocytogenes, may be introduced and spread via various routes. However, limited tools are available for retailers to assess practices that can enhance control of Listeria transmission to fresh produce. Therefore, we developed an agent-based model that can simulate Listeria transmission in retail produce sections to optimize environmental sampling programs and evaluate control strategies. A single retail store was used as a model environment, in which various routes of Listeria introduction into and transmission between environmental surfaces were modeled. Model prediction (i.e., Listeria prevalence) was validated using a published longitudinal study for all surfaces that were included in both the model and the validation data. Sensitivity analysis using the Partial Rank Correlation Coefficient showed that (i) initial Listeria concentration from incoming produce, (ii) transfer coefficient from produce to employee's hands, and (iii) transfer coefficient from consumer to produce were the top three parameters that were significantly (p < 0.0018) associated with the mean Listeria prevalence across all agents, suggesting that the accuracy of these parameters are important for prediction of overall Listeria prevalence at retail. Cluster analysis grouped agents with similar contamination patterns into six unique clusters; this information can be used to optimize the sampling plans for retail environments. Scenario analysis suggested that (i) more stringent supplier control as well as (ii) practices reducing Listeria transmission via consumer hands may have the largest impact on reducing finished product contamination. Overall, we show that an agent-based model can serve as a foundational tool to help with decision making on Listeria control strategies at retail.

Voronoi diagrams and Delaunay triangulation for modelling animal territorial behaviour

We explore the use of movable automata in numerical modelling of male competition for territory. We used territorial dragonflies as our biological inspiration for the model, assuming two types of competing males: (a) faster and larger males that adopt a face-off strategy and repulse other males; (b) slower and smaller males that adopt a non-aggressive strategy. The faster and larger males have higher noise intensity, leading to faster motion and longer conservation of motion direction. The velocity distributions resemble the Maxwell distributions of velocity, expected in Brownian dynamics, with two probable velocities and distribution widths for the two animal subpopulations. The fast animals' trajectories move between visually fixed density folds of the slower animal subpopulation. A correlation is found between individual velocity and individual area distribution, with smaller animals concentrated in a region of small velocities and areas. Attraction between animals results in a modification of the system behaviour, with larger animals spending more time being surrounded by smaller animals and being slowed down by their interaction with the surroundings. Overall, the study provides insights into the dynamics of animal competition for territory and the impact of attraction between animals.

Coupling an agent-based model and an ensemble Kalman filter for real-time crowd modelling

Agent-based modelling has emerged as a powerful tool for modelling systems that are driven by discrete, heterogeneous individuals and has proven particularly popular in the realm of pedestrian simulation. However, real-time agent-based simulations face the challenge that they will diverge from the real system over time. This paper addresses this challenge by integrating the ensemble Kalman filter (EnKF) with an agent-based crowd model to enhance its accuracy in real time. Using the example of Grand Central Station in New York, we demonstrate how our approach can update the state of an agent-based model in real time, aligning it with the evolution of the actual system. The findings reveal that the EnKF can substantially improve the accuracy of agent-based pedestrian simulations by assimilating data as they evolve. This approach not only offers efficiency advantages over existing methods but also presents a more realistic representation of a complex environment than most previous attempts. The potential applications of this method span the management of public spaces under 'normality' to exceptional circumstances such as disaster response, marking a significant advancement for real-time agent-based modelling applications.

Exploring Winegrowers' Behaviours and Ecological Impacts Under Climate Change and Policy Scenarios-Examples from Three European Winegrowing Regions

Viticulture is an example of a socio-ecological system that poses serious challenges for sustainable soil management and pesticide use, with various interactions between winegrowers' decision-making and ecological consequences. This study introduces an agent-based model (ABM) on winegrowers' decision on inter-row management and pesticide use. The ABM builds upon an empirical study of winegrowers' decision-making in European viticultural landscapes and has been built for three case study regions: Leithaberg (Austria), Palatinate (Germany) and Târnave (Romania). The ABM allows for analysing potential effects of policy instruments including mandatory vegetation cover in the inter-rows, the reduction of fungicide use and ban of insecticides against Lobesia botrana. The effects of policies differ between the case study regions, indicating how important the local context is for effective policies. For example, policies aiming at higher inter-row vegetation cover had the strongest effects on vegetation cover, landscape aesthetics and soil loss in Târnave since many vineyards are currently intensively tilled and there exist no policies supporting inter-row vegetation cover in Romania.

The Epidemiological and Economic Impact of COVID-19 in Kazakhstan: An Agent-Based Modeling

BACKGROUND

Our study aimed to assess how effective the preventative measures taken by the state authorities during the pandemic were in terms of public health protection and the rational use of material and human resources.

MATERIALS AND METHODS

We utilized a stochastic agent-based model for COVID-19's spread combined with the WHO-recommended COVID-ESFT version 2.0 tool for material and labor cost estimation.

RESULTS

Our long-term forecasts (up to 50 days) showed satisfactory results with a steady trend in the total cases. However, the short-term forecasts (up to 10 days) were more accurate during periods of relative stability interrupted by sudden outbreaks. The simulations indicated that the infection's spread was highest within families, with most COVID-19 cases occurring in the 26-59 age group. Government interventions resulted in 3.2 times fewer cases in Karaganda than predicted under a "no intervention" scenario, yielding an estimated economic benefit of 40%.

CONCLUSION

The combined tool we propose can accurately forecast the progression of the infection, enabling health organizations to allocate specialists and material resources in a timely manner.

The impacts of climate change, energy policy and traditional ecological practices on future firewood availability for Diné (Navajo) People

Local-scale human-environment relationships are fundamental to energy sovereignty, and in many contexts, Indigenous ecological knowledge (IEK) is integral to such relationships. For example, Tribal leaders in southwestern USA identify firewood harvested from local woodlands as vital. For Diné people, firewood is central to cultural and physical survival and offers a reliable fuel for energy embedded in local ecological systems. However, there are two acute problems: first, climate change-induced drought will diminish local sources of firewood; second, policies aimed at reducing reliance on greenhouse-gas-emitting energy sources may limit alternatives like coal for home use, thereby increasing firewood demand to unsustainable levels. We develop an agent-based model trained with ecological and community-generated ethnographic data to assess the future of firewood availability under varying climate, demand and IEK scenarios. We find that the long-term sustainability of Indigenous firewood harvesting is maximized under low-emissions and low-to-moderate demand scenarios when harvesters adhere to IEK guidance. Results show how Indigenous ecological practices and resulting ecological legacies maintain resilient socio-environmental systems. Insights offered focus on creating energy equity for Indigenous people and broad lessons about how Indigenous knowledge is integral for adapting to climate change. This article is part of the theme issue 'Climate change adaptation needs a science of culture'.

Identifying key socioecological factors influencing the expression of egalitarianism and inequality among foragers

Understanding how resource characteristics influence variability in social and material inequality among foraging populations is a prominent area of research. However, obtaining cross-comparative data from which to evaluate theoretically informed resource characteristic factors has proved difficult, particularly for investigating interactions of characteristics. Therefore, we develop an agent-based model to evaluate how five key characteristics of primary resources (predictability, heterogeneity, abundance, economy of scale and monopolizability) structure pay-offs and explore how they interact to favour both egalitarianism and inequality. Using iterated simulations from 243 unique combinations of resource characteristics analysed with an ensemble machine-learning approach, we find the predictability and heterogeneity of key resources have the greatest influence on selection for egalitarian and nonegalitarian outcomes. These results help explain the prevalence of egalitarianism among foraging populations, as many groups probably relied on resources that were both relatively less predictable and more homogeneously distributed. The results also help explain rare forager inequality, as comparison with ethnographic and archaeological examples suggests the instances of inequality track strongly with reliance on resources that were predictable and heterogeneously distributed. Future work quantifying comparable measures of these two variables, in particular, may be able to identify additional instances of forager inequality. This article is part of the theme issue 'Evolutionary ecology of inequality'.

Dynamic calibration with approximate Bayesian computation for a microsimulation of disease spread

The global COVID-19 pandemic brought considerable public and policy attention to the field of infectious disease modelling. A major hurdle that modellers must overcome, particularly when models are used to develop policy, is quantifying the uncertainty in a model's predictions. By including the most recent available data in a model, the quality of its predictions can be improved and uncertainties reduced. This paper adapts an existing, large-scale, individual-based COVID-19 model to explore the benefits of updating the model in pseudo-real time. We use Approximate Bayesian Computation (ABC) to dynamically recalibrate the model's parameter values as new data emerge. ABC offers advantages over alternative calibration methods by providing information about the uncertainty associated with particular parameter values and the resulting COVID-19 predictions through posterior distributions. Analysing such distributions is crucial in fully understanding a model and its outputs. We find that forecasts of future disease infection rates are improved substantially by incorporating up-to-date observations and that the uncertainty in forecasts drops considerably in later simulation windows (as the model is provided with additional data). This is an important outcome because the uncertainty in model predictions is often overlooked when models are used in policy.

Responses of well-aquifer system to four earthquakes and implication for contaminant transport in Changan landfill, China

Evaluation of aquifer response to earthquakes is important to understanding the evolution of aquifer properties and contaminant transport. In Changan Landfill, a multi parameter monitoring system was installed to collect data every 12 h. Principal Component Analysis (PCA), energy density and wavelet analysis, and tidal analysis were used to reveal the response mechanisms to four earthquakes, sensitivity of various parameters to seismic stimuli, and permeability evolution, respectively. The results showed that the chemical characteristics of two aquifers were distinct. The shallow aquifer was polluted by leachate, and the deep aquifer was less polluted. Coseismic responses were dominated by the deep aquifer and the relative contribution of each aquifer remained stable. Coseismic chemical changes were dominated by the opening and closing of fractures, and the long-term evolution was controlled by seasonality. The deep aquifer had a higher sensitivity to seismic stimuli than the shallow aquifer. Tidal analysis showed that a general permeability decrease during this monitoring period reduced contaminant transport, but some contaminants crossed the aquitard between the two aquifers after the Zizhong earthquake, due to an increase in vertical permeability.

Teaching the Modeling of Human-Environment Systems: Acknowledging Complexity with an Agent-Based Model

Agent-based modeling is a promising tool for familiarizing students with complex systems as well as programming skills. Human-environment systems, for instance, entail complex interdependencies that need to be considered when modeling these systems. This complexity is often neglected in teaching modeling approaches. For a heterogeneous group of master's students at a German university, we pre-built an agent-based model. In class, this was used to teach modeling impacts of land use policies and markets on ecosystem services. As part of the course, the students had to perform small research projects with the model in groups of two. This study aims to evaluate how well students could deal with the complexity involved in the model based on their group work outcomes. Chosen indicators were, e.g., the appropriateness of their research goals, the suitability of the methods applied, and how well they acknowledged the limitations. Our study results revealed that teaching complex systems does not need to be done with too simplistic models. Most students, even with little background in modeling and programming, were able to deal with the complex model setup, conduct small research projects, and have a thoughtful discussion on the limitations involved. With adequate theoretical input during lectures, we recommend using models that do not hide the complexity of the systems but foster a realistic simplification of the interactions.

SUPPLEMENTARY INFORMATION

The online version contains supplementary material available at 10.1007/s10956-022-10022-z.

Likelihood-Free Dynamical Survival Analysis applied to the COVID-19 epidemic in Ohio

The Dynamical Survival Analysis (DSA) is a framework for modeling epidemics based on mean field dynamics applied to individual (agent) level history of infection and recovery. Recently, the Dynamical Survival Analysis (DSA) method has been shown to be an effective tool in analyzing complex non-Markovian epidemic processes that are otherwise difficult to handle using standard methods. One of the advantages of Dynamical Survival Analysis (DSA) is its representation of typical epidemic data in a simple although not explicit form that involves solutions of certain differential equations. In this work we describe how a complex non-Markovian Dynamical Survival Analysis (DSA) model may be applied to a specific data set with the help of appropriate numerical and statistical schemes. The ideas are illustrated with a data example of the COVID-19 epidemic in Ohio.

Multi-agent simulation model updating and forecasting for the evaluation of COVID-19 transmission

Agent-based models have been an emerging approach in epidemiological modelling, specifically in investigating the COVID-19 virus. However, there are challenges to its validation due to the absence of real data on specific socio-economic and cognitive aspects. Therefore, this work aims to present a strategy for updating, verifying and validating these models based on applying the particle swarm optimization algorithm to better model a real case. For such application, this work also presents a new framework based on multi-agents, whose significant contribution consists of forecasting needed hospital resources, population adaptative immunization and reports concerning demographic density, including physical and socio-economic aspects of a real society in the modelling task. Evaluation metrics such as the data's Shape Factor (SF), Mean Square Error (RMSE), and statistical and sensitivity analyses of the responses obtained were applied for comparison with the real data. The Brazilian municipality of Passa Vinte, located in the State of Minas Gerais (MG), was used as a case study. The model was updated in cumulative cases until the 365th day of the pandemic. The statistical and sensitivity analysis results showed similar patterns around the actual data up to the 500th day of the pandemic. Their mean values of SF and RMSE were 0.96 and 7.22, respectively, showing good predictability and consistency, serving as an adequate tool for decision-making in health policies.

Agent-based modelling for Urban Analytics: State of the art and challenges

Agent-based modelling (ABM) is a facet of wider Multi-Agent Systems (MAS) research that explores the collective behaviour of individual ‘agents’, and the implications that their behaviour and interactions have for wider systemic behaviour. The method has been shown to hold considerable value in exploring and understanding human societies, but is still largely confined to use in academia. This is particularly evident in the field of Urban Analytics; one that is characterised by the use of new forms of data in combination with computational approaches to gain insight into urban processes. In Urban Analytics, ABM is gaining popularity as a valuable method for understanding the low-level interactions that ultimately drive cities, but as yet is rarely used by stakeholders (planners, governments, etc.) to address real policy problems. This paper presents the state-of-the-art in the application of ABM at the interface of MAS and Urban Analytics by a group of ABM researchers who are affiliated with the Urban Analytics programme of the Alan Turing Institute in London (UK). It addresses issues around modelling behaviour, the use of new forms of data, the calibration of models under high uncertainty, real-time modelling, the use of AI techniques, large-scale models, and the implications for modelling policy. The discussion also contextualises current research in wider debates around Data Science, Artificial Intelligence, and MAS more broadly.

Quantitative Characterization of Complex Systems—An Information Theoretic Approach

A significant increase in System-of-Systems (SoS) is currently observed in the social and technical domains. As a result of the increasing number of constituent system components, Systems of Systems are becoming larger and more complex. Recent research efforts have highlighted the importance of identifying innovative statistical and theoretical approaches for analyzing complex systems to better understand how they work. This paper portrays the use of an agnostic two-stage examination structure for complex systems aimed towards developing an information theory-based approach to analyze complex technical and socio-technical systems. Towards the goal of characterizing system complexity with information entropy, work was carried out in exploring the potential application of entropy to a simulated case study to illustrate its applicability and to establish the use of information theory within the broad horizon of complex systems. Although previous efforts have been made to use entropy for understanding complexity, this paper provides a basic foundation for identifying a framework to characterize complexity, in order to analyze and assess complex systems in different operational domains.