Using Machine Learning to drive social learning in a Covid-19 Agent-Based Model

Aug 4, 2026

Disease transmission and governmental interventions shape the spread of Covid-19, making models valuable for optimizing policy actions. Agent-Based Models (ABMs) can govern agents via rules or data-driven learning, and governments can behave as isolated or social learners. This work builds a data-driven social approach where government decisions use a machine learning model trained on European disease data, linking risk perception to the intervention stringency index and comparing four algorithms for multi- and tri-class classification.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages4

Reading Time10 min

AccessFree to download

Introduction

Background: risk perception and policy decisions

Role of agent-based models and social learning

Methods

Agent-Based Model (ABM)

Data and stringency index construction

Using Machine Learning to drive social learning in a Covid-19 Agent-Based Model Ellen-Wien Augustijn 1, Rosa Aguilar Bolivar1, and Shaheen Abdulkareem2 1 Department of Geoinformation Processing (GIP), University of Twente, Enschede, The Netherlands 2 Department of computer science, University of Duhok, Duhok, Iraq Correspondence: Ellen-Wien Augustijn ([p.w.m.augustijn@utwente.nl](p.w.m.augustijn@utwente.nl)) Abstract. Disease transmission and governmental interventions influence the spread ofCovid-19. Models can be essential tools to optimise these governmental interventions. This requires the exploration of various ways to implement government agent behaviour. In Agent-Based Models (ABMs), government agent behaviour can be rule-based or data-driven, and the agent can be an isolated learner (using only its own data) or a social learner. We explore the creation of a data-driven social approach in which behaviour is based on a Machine Learning (ML) algorithm, and the government considers data from other European countries as input for their decision-making. Governmental actions start with risk perception, based on several parameters, e.g. the number of disease cases, deaths, and hospitalisation rate. The interventions are measured via the stringency index, measuring the simultaneous number of interventions (working from home, wearing a facemask, closing schools, etc.) taken. We test four machine learning algorithms (Bayesian Network (BN), c4.5, Naïve Bayes (NB) and Random Forest (RF)), using a 5-class and a 3-class classification of the stringency level. The algorithms are trained on disease data from many European countries. The bestperforming algorithms were c4.5 and RF. The next step is to implement these algorithms into the ABM and evaluate the outcomes compared to the original model. Keywords. Agent-Based Modelling, Machine Learning, Covid-19 1 Introduction During the Covid pandemic, we learned that governments play an important role in disease interventions. They can enforce lockdowns, make wearing face masks mandatory and implement vaccination campaigns. Governmental decision-making is based on a strategy of risk perception and coping appraisal. Governments decide on the risk level based on disease incidence, the number of available hospital beds etc. To understand the impact of governmental decisions on disease diffusion, we need to integrate disease models with policy models (Hadley et al., 2021) . Agent-Based Models (ABMs) are good tools for modelling bottom-up disease diffusion and personal decision-making. In many cases, governments are not modelled as agents. When included, governments are modelled as isolated entities that apply rule-based behaviour to decide what interventions to use (Augustijnet al., 2022) . However, decision-making might be a more social activity in which governments of various countries collaborate and share experiences. Social agents are interactive; they communicate with their neighbours (in this case, other European governments) to learn effectively within their groups (Abdulkareem et al., 2020) . A complicating factor is that at the pandemic's beginning, nobody had much experience with policies for effective disease control of Covid-19. To simulate this learning process, the intelligence of the government agent should increase during the simulation. This type of learning is best achieved by replacing the rule-based agent decision-making in the ABM with a Machine Learning (ML) algorithm that learns directly from data. When implementing agent learning via ML, many decisions have to be made concerning the type of ML algorithm, the data used to train the ML algorithm, and the architecture of linking the ML and ABM. In this research, we take the first step in replacing an isolated government agent that uses rule-based decision- making with a model where an ML algorithm drives the government agent’s decisions for a situation where the exchange of information with other

FAQs

How does the proposed approach enable social learning for government agents in the ABM?

It replaces rule-based government decision-making with a machine learning model that learns from data, using information from other European countries as inputs for government decisions.

What variables are used for risk perception and intervention selection in the government agent model?

Risk perception is based on the number of positive tests per 100,000 inhabitants per week and the number of hospitalised individuals per day, mapped to five risk levels, while interventions are assessed using the stringency index.

Which machine learning algorithms are tested and what classification task is used?

Four algorithms are evaluated: Bayesian Network, c4.5, Naïve Bayes, and Random Forest, using both 5-class and 3-class classification of the intervention stringency level.

Aug 4, 2026

Upload by Lucas Martin