An agent-based policy laboratory for COVID-19 containment strategies
Ben Vermeulen*, Andreas Pyka**, Matthias Müller
* For inquiries about the conceptual or operational model, contact b.vermeulen@uni-hohenheim.de
** For inquiries into research collaboration as well as promotional and media usage, contact a.pyka@uni-hohenheim.de
Please cite as:
Vermeulen, Ben, Müller, Matthias and Pyka, Andreas (2021) 'Social Network Metric-Based Interventions? Experiments with an Agent-Based Model of the COVID-19 Pandemic in a Metropolitan Region' Journal of Artificial Societies and Social Simulation 24 (3) 6 http://jasss.soc.surrey.ac.uk/24/3/6.html. doi: 10.18564/jasss.4571
Following the outbreak of COVID-19, governments are looking for a balance between containment on the one hand and retaining liberties of citizens and keeping the economy afloat on the other. Policy makers turn to academic modelers for policy guidance. Not surprisingly, academics are quick to present models also on the internet. For instance, few days ago, a simulation model of the COVID-19 diffusion process of The Washington Post was launched to explain the impact of 'social distancing'. In this model, the contamination process is driven by homogeneous agents roaming an undefined space and is thus more physics-based than social scientific. Moreover, the Neherlab, of the Biozentrum, University of Basel, recently launched a metapopulation model using a system of differential equations that essentially structure spatial features and social interactions.
In society, however, people generally infect each other in private and public spaces such as their own house, their office or school, in the supermarket, at mass gatherings such as sports events or concerts, etc. Moreover, certain people have central or gatekeeper positions in social networks and may thus be pivotal in spreading or containing COVID-19, while other people have peripheral positions and not even contract it. As such, people’s agenda of activities, their social relationships, as well as the social setting of the locations matter a great deal. In contrast to the two models mentioned, agent-based models are particularly well-suited to study emerging socio-temporal patterns of real-world epidemics driven by the heterogeneity and autonomy of agents.
On this page, you find a first version of such an agent-based simulation model (with policy intervention features) in which agents are members of households, have a certain age and associated (basic) agenda determining where they spend time, and hence when with whom they interact – and potentially transmit the SARS-CoV-2 coronavirus. It is assumed that only when interaction takes place at particular locations, here: houses (orange), offices (purple), supermarket/ shop (red), school (pink), recreation & leisure (green), agents can infect one another. Moreover, there is a simple disease progression model akin to the SEIR metapopulation model specifying the infection transitions going from susceptible/ uninfected (green), exposed but latent/ non-infectious during incubation (light yellow), exposed and mildly presymptomatically infectious (yellow), infected and infectious (orange), severely ill (dark orange), critically ill (red), then either recovering and immune (purple) or deceased (blue).
House [in residential area]
School [in residential area]
Workspace [in industrial/ commercial area]
Recreation area
Supermarket/ shop in commercial area
Hospital [in designated spot]
Cemetery (possibly with graves)
General policies:
School policy:
Hospitalization policy:
Office policy:
Public gatherings policy:
The policy laboratory allows you, as policy maker or concerned citizen, to experiment with a range of policy interventions to contain the spread of the COVID-19 disease, either intermediate (but realize there is always a substantial time lag!), or from the start of a fresh run or a new world with an alternative random seed. The controls at the bottom also allow you to pause, step, or fast-forward through the timeline of the real-time simulation. The stacked graph on the right and some statistics (state incidence, R0 and CFR) are provided for insights and cross-comparison.
You are kindly invited to experiment with the various policy interventions and ascertain that outcomes are not always as trivial as purported by mainstream models. Think for instance of how early expiration or rather immunity of ‘gate-keepers’ would stop further diffusion – so is containment always good? And who should be isolated or quarantined? Are more local or focused options possible?
We are aware that the world we model is yet incomplete and many more policy interventions are possible. At this point, however, we use this policy laboratory to illustrate and raise awareness of complex system properties, interlocked layers of diffusion and the possibilities of dynamic mixes of location-based or network-position based policy interventions. Moreover, the model may form the basis for enhancing our phenomenological understanding of epidemics in real-world complexity as well as for engineering “next-gen” policy interventions exploiting spatio-temporal features of the world as well as social network relationships beyond indiscriminate total lockdowns.
Technical information and data references.
1. Incubation time distribution is assumed to be a (discretized) lognormal, which has been fitted to Li et al. (2020) as referred to by WHO (2020).
2. Household composition distribution is derived from CBS Dataset 37975 for 2019 and demographic statistics derived from Dataset 7461 for 2019 is used to validate Case Fatality Rate
3. Assumed infection state progression rates by age-group as calibrated to Li et al. / WHO by Aksamentov, I., Noll, N., Neher, R. (2020).
4. A SEIR-inspired stage-gate disease progression model is based on the infection progression model presented in An der Heiden & Buchholz (2020) of the Robert Koch Institute.
5. Emerging R0s in early stages have been checked against basic containment strategies reported in Wang et al. (2020).
6. Preliminary checks on the serial interval for cross-validity have been checked against Wang et al. (2020), Du et al. (2020).
7. Infectiousness, notably with regard to physical distance, between/ across age cohorts is unknown, also for disease stage (exposed, infected, severe, critical) and assumed a uniform 10%. Note that, unlike metapopulation models, our model does have physical distance such that this has been used for scaling R0 and CFR.
For inspiration on containment strategies, see e.g. Ferguson et al. (2020), Halloran et al (2008), and Hellewell et al. (2020). For a taxonomy of ABMs for epidemiological studies, see Hunter et al. (2017).
Acknowledgements
We want to thank Juan Francisco Robles Fuentes (UGR), Frank Kugelmeier (St. Ursula Attendorn), Rüdiger Kessel (Metrodata), Tobias Gaisser (TH Nürnberg), Cristina Ponsiglione (UniNa), Aykut Kibritçioğlu (TAU), Eleonora Psenner (EURAC), Yasushi Hara (Hitotsubashi University), Bin-Tzong Chie (TKU), Giuseppe Vizzari (UniMib), Marc Printz (Bluewin), Bogang Jun (Inha), Nicolas Béfort (Neoma), Kurtuluş Barış Öner, Luis Rubalcaba Bermejo (UAH) and friends and family members for translations, testing, GUI design comments, code checks, and developing ideas for future extensions. All remaining issues of course are our responsibility.
Bibliography.
Ajelli, M.; Gonçalves, B.; Balcan, D.; Colizza, V.; Hu, H.; Ramasco, J. J. et al. (2010): Comparing large-scale computational approaches to epidemic modeling: agent-based versus structured metapopulation models. In BMC Infectious Diseases 10 (1).
Aksamentov, I., Noll, N., Neher, R. (2020): COVID-19 Scenario webapp. Available online at https://neherlab.org/covid19, checked on 3/22/2020.
Aleman, Dionne M.; Wibisono, Theodorus G.; Schwartz, Brian (2011): A Nonhomogeneous Agent-Based Simulation Approach to Modeling the Spread of Disease in a Pandemic Outbreak. In Interfaces 41 (3), pp. 301–315. DOI: 10.1287/inte.1100.0550.
An der Heiden, M.; Buchholz, U. (2020): Modellierung von Beispielszenarien der SARS-CoV-2-Epidemie 2020 in Deutschland. Robert Koch-Institut (RKI), checked on 3/22/2020.
Anderson, R. M.; Heesterbeek, H.; Klinkenberg, D.; Hollingsworth, T. D. (2020): How will country-based mitigation measures influence the course of the COVID-19 epidemic? In The Lancet 395, pp. 931–934. DOI: 10.1016/S0140-6736(20)30567-5.
Chinazzi, Matteo; Davis, Jessica T.; Ajelli, Marco; Gioannini, Corrado; Litvinova, Maria; Merler, Stefano et al. (2020): The effect of travel restrictions on the spread of the 2019 novel coronavirus (COVID-19) outbreak. In Science (New York, N.Y.). DOI: 10.1126/science.aba9757.
Das, Tapas K.; Savachkin, Alex A.; Zhu, Yiliang (2008): A large-scale simulation model of pandemic influenza outbreaks for development of dynamic mitigation strategies. In IIE Transactions 40 (9), pp. 893–905. DOI: 10.1080/07408170802165856.
Day, Troy; Park, Andrew; Madras, Neal; Gumel, Abba; Wu, Jianhong (2006): When is quarantine a useful control strategy for emerging infectious diseases? In American journal of epidemiology 163 (5), pp. 479–485. DOI: 10.1093/aje/kwj056.
Dowd, Jennifer Beam; Rotondi, Valentina; Adriano, Liliana; Brazel, David M.; Block, Per; Ding, Xuejie et al. (2020): Demographic science aids in understanding the spread and fatality rates of COVID-19 (medRxiv preprint).
Du, Z.; Xu, X.; Wu, Y.; Wang, L.; Cowling, B. J.; Ancel Meyers, L. (2020): Serial interval of COVID-19 among publicly reported confirmed cases. In Emerging infectious diseases. DOI: 10.3201/eid2606.200357.
Ferguson, Neil M.; Cummings, Derek A. T.; Fraser, Christophe; Cajka, James C.; Cooley, Philip C.; Burke, Donald S. (2006): Strategies for mitigating an influenza pandemic. In Nature 442 (7101), pp. 448–452. DOI: 10.1038/nature04795.
Ferguson, N.; Laydon, D.; Nedjati Gilani, G.; Imai, N.; Ainslie, K.; Baguelin, M. et al. (2020): Report 9: Impact of non-pharmaceutical interventions (NPIs) to reduce COVID19 mortality and healthcare demand. With assistance of Medical Research Council (MRC).
Halloran, E. M.; Ferguson; N. M.; Eubank, S.; Longini, I. M.; Cummings, D.A.T. et al. (2008): Modeling targeted layered containment of an influenza pandemic in the United States. In Proceedings of the National Academy of Sciences of the United States of America 105 (12), pp. 4639–4644.
Hellewell, Joel; Abbott, Sam; Gimma, Amy; Bosse, Nikos I.; Jarvis, Christopher I.; Russell, Timothy W. et al. (2020): Feasibility of controlling COVID-19 outbreaks by isolation of cases and contacts. In The Lancet Global Health 8 (4), e488-e496. DOI: 10.1016/S2214-109X(20)30074-7.
Hunter, Elizabeth; Mac Namee, Brian; Kelleher, John D. (2017): A Taxonomy for Agent-Based Models in Human Infectious Disease Epidemiology. In JASSS 20 (3). DOI: 10.18564/jasss.3414.
Li, Qun; Guan, Xuhua; Wu, Peng; Wang, Xiaoye; Zhou, Lei; Tong, Yeqing et al. (2020): Early Transmission Dynamics in Wuhan, China, of Novel Coronavirus-Infected Pneumonia. In The New England journal of medicine. DOI: 10.1056/NEJMoa2001316.
Roche, B.; Drake, J. M.; Rohani, P. (2011): An Agent-Based Model to study the epidemiological and evolutionary dynamics of Influenza viruses. In BMC Bioinformatics 12 (87), pp. 1–10.
Tang, Biao; Bragazzi, Nicola Luigi; Li, Qian; Tang, Sanyi; Xiao, Yanni; Wu, Jianhong (2020): An updated estimation of the risk of transmission of the novel coronavirus (2019-nCov). In Infectious Disease Modelling 5, pp. 248–255. DOI: 10.1016/j.idm.2020.02.001.
The Novel Coronavirus Pneumonia Emergency Response Epidemiology Team (2020): The Epidemiological Characteristics of an Outbreak of 2019 Novel Coronavirus Diseases (COVID-19) — China, 2020. In China CDC Weekly 2 (8).
Wang, Chaolong; Liu, Li; Hao, Xingjie; Guo, Huan; Wang, Qi; Huang, Jiao et al. (2020): Evolving Epidemiology and Impact of Non-pharmaceutical Interventions on the Outbreak of Coronavirus Disease 2019 in Wuhan, China (medRxiv preprint), checked on 3/21/2020.
World Health Organization (2020): Report of the WHO-China Joint Mission on Coronavirus Disease 2019 (COVID-19). World Health Organization.