Agent-based modeling for realistic reproduction of human mobility and contact behavior to evaluate test and isolation strategies in epidemic infectious disease spread

TL;DR

This study presents a modular, trip-based agent-based model (ABM) implemented in the MEmilio framework to simulate infectious disease spread with realistic human mobility and contact patterns. By integrating time-varying viral load, infection states, vaccination and immunity, and flexible nonpharmaceutical interventions, the model can reproduce city-scale COVID-19 dynamics and evaluate testing and isolation strategies alongside quarantine policies. The Brunswick 2021 calibration demonstrates the approach’s ability to match deaths, ICU occupancy, and detected infections, while extensive analyses reveal that quarantine length often outweighs quarantine efficiency and that symptom-based testing can substantially mitigate spread, potentially replacing lockdowns when testing capacity is high. The work provides a scalable tool—capable of running on consumer hardware with HPC-enabled ensemble analyses—that informs policy decisions on NPIs and pandemic preparedness across regions and potentially other diseases.

Abstract

Agent-based models have proven to be useful tools in supporting decision-making processes in different application domains. The advent of modern computers and supercomputers has enabled these bottom-up approaches to realistically model human mobility and contact behavior. The COVID-19 pandemic showcased the urgent need for detailed and informative models that can answer research questions on transmission dynamics. We present a sophisticated agent-based model to simulate the spread of respiratory diseases. The model is highly modularized and can be used on various scales, from a small collection of buildings up to cities or countries. Although not being the focus of this paper, the model has undergone performance engineering on a single core and provides an efficient intra- and inter-simulation parallelization for time-critical decision-making processes. In order to allow answering research questions on individual level resolution, nonpharmaceutical intervention strategies such as face masks or venue closures can be implemented for particular locations or agents. In particular, we allow for sophisticated testing and isolation strategies to study the effects of minimal-invasive infectious disease mitigation. With realistic human mobility patterns for the region of Brunswick, Germany, we study the effects of different interventions between March 1st and May 30, 2021 in the SARS-CoV-2 pandemic. Our analyses suggest that symptom-independent testing has limited impact on the mitigation of disease dynamics if the dark figure in symptomatic cases is high. Furthermore, we found that quarantine length is more important than quarantine efficiency but that, with sufficient symptomatic control, also short quarantines can have a substantial effect.

Figures (18)

• Figure 1: Two exemplary time steps of the ABM simulation with five agents. From top left to bottom right, persons interact and potentially infect each other at a location. Then, agents potentially move to other location types, e.g., from a supermarket to work, or to the hospital if they have severe symptoms. After a time step, the infection state can progress, e.g., from mildly infected to severely infected.
• Figure 2: Course of viral load (left) and different, corresponding viral sheds (right) over 20 days. The presented viral sheds are given for varying factor $s_{\textrm{f,p}}$ in \ref{['eqn:viral_shed']}, ranging from 0.01 (blue) to 0.28 (red).
• Figure 3: Infection state transition model.One-way arrows represent the possible transitions from one infection state to another. Rounded gray borders indicate noninfectious states while red angular borders indicate symptom states that are used for infectious agents only. Orange to violet colors indicate the symptomatic severity and dashed borders indicate if a person is considered as symptomatic (for testing schemes).
• Figure 4: An exemplary Testing Strategy with three different Testing Schemes and different Testing Criteria (middle part, blue).We define Testing Schemes for specific locations or, more generally, for different location types (top). Each time an agent enters one of these locations, the Testing Criteria are evaluated (middle). If the criteria are met, the agent undergoes testing with a specified probability and test type. The resulting test remains valid for a predetermined duration. Additionally, the testing strategy can be configured to be active only for a specified time period (bottom).
• Figure 5: Example of Testing Schemes 1 and 2 from \ref{['fig:test_strategies']}. In Testing Scheme 1, agents are obliged to test themselves before going to work. It is sufficient to use a Rapid-Antigen Test if they do not show symptoms. All agents who are not symptomatic, i.e., the first three agents in the bottom row, do an antigen test from which the first is false-negative, the second positive and the third negative. Through Testing Scheme 2, a PCR-Test is required if the agent shows symptoms. The last remaining, symptomatic agent in the bottom row performs a PCR test which turns out to be positive. The two positive tests resulted in quarantining at home.