How individuals’ opinions influence society’s resistance to epidemics: an agent-based model approach

Background Protecting public health from infectious diseases often relies on the cooperation of citizens, especially when self-care interventions are the only viable tools for disease mitigation. Accordingly, social aspects related to public opinion have been studied in the context of the recent COVID-19 pandemic. However, a comprehensive understanding of the effects of opinion-related factors on disease spread still requires further exploration. Methods We propose an agent-based simulation framework incorporating opinion dynamics within an epidemic model based on the assumption that mass media channels play a leading role in opinion dynamics. The model simulates how opinions about preventive interventions change over time and how these changes affect the cumulative number of cases. We calibrated our simulation model using YouGov survey data and WHO COVID-19 new cases data from 15 different countries. Based on the calibrated models, we examine how different opinion-related factors change the consequences of the epidemic. We track the number of total new infections for analysis. Results Our results reveal that the initial level of public opinion on preventive interventions has the greatest impact on the cumulative number of cases. Its normalized permutation importance varies between 69.67% and 96.65% in 15 models. The patterns shown in the partial dependence plots indicate that other factors, such as the usage of the pro-intervention channel and the response time of media channels, can also bring about substantial changes in disease dynamics, but only within specific ranges of the dominant factor. Conclusions Our results reveal the importance of public opinion on intervention during the early stage of the pandemic in protecting public health. The findings suggest that persuading the public to take actions they may be hesitant about in the early stages of epidemics is very costly because taking early action is critical for mitigating infectious diseases. Other opinion-related factors can also lead to significant changes in epidemics, depending on the average level of public opinion in the initial stage. These findings underscore the importance of media channels and authorities in delivering accurate information and persuading community members to cooperate with public health policies. Supplementary Information The online version contains supplementary material available at 10.1186/s12889-024-18310-6.


Table S1
Parameter tuples used to reproduce opinion dynamics (compliance level) observed in 15 countries.The root mean squared error (RMSE) is computed between the average sequence of 2,000 ABM instances with a population of 50,000 and data.Note that Testbeds 1 to 15 are calibrated with datasets corresponding to 15 countries, namely Australia, Canada, Denmark, France, Germany, Italy, Japan, Netherlands, Norway, Singapore, Spain, Sweden, the UK, the USA, and Vietnam, in that order.In this section, we extend our experimental design and test all ten opinion-related factors' (O 1 , • • • , O 10 ) impact on the disease spread.We generate 1,000 distinct tuples from the ranges shown in Table 1 using Latin hypercube sampling with the minimax correlation criterion.Again, each tuple is examined based on the 15 testbeds.All other parameter values except for the 10 opinion-related ones are fixed to theirs in the used testbed.For further analysis, we track the number of total new infections during the simulation time (Y C ) and use the log base 10 of the values as the response variable.As we found in the results with 6 parameters, the mean of the initial opinions on intervention (O 1 ) dominantly affects the disease spread in all 15 models again.

Basis
Table S3 Normalized permutation importance (sum to 1.0) of 10 opinion-related factors in their corresponding random forest regression models and their R 2 values.The 15 explanatory models are generated based on the results of 1,000 simulation instances each with the corresponding testbed.The result shows that, among the 10 factors, the average of the public's initial opinion values (O 1 ) dominates the rest in terms of the permutation importance.

Basis
Fig. S1.Violin plots of the log-scaled epidemic size (Y C ) distributions of 1000 opinion-related parameter tuples in baseline models fitted to the data for 15 countries.Plots are sorted by the mean of the log-scaled epidemic size.Epidemic sizes are measured in the number of total new infections during the simulation time in a virtual social system with a 50,000 population size.The red dots in the figure represent the response values (Y C ) of the models fitted to the corresponding data.Plots show that the consequence could be much better or much worse depending on the given conditions of opinion dynamics.

Fig
Fig. S3.One one-way (O 1 ) and nine two-way partial dependence plots of the regression model for the cases in Group 2: Results based on Testbeds 1, 3, 8, 9, 12, and 15.

Fig. S4 .
Fig. S4.One one-way (O 1 ) and nine two-way partial dependence plots of the regression model for the cases in Group 3: Results based on Testbeds 2, 7, and 13.
Table S2This table provides the parameter tuples used to replicate the observed dynamics of new COVID-19 infections in 15 countries based on opinion trends generated using the corresponding tuples from Table1.We calculate two root mean squared errors (RMSE) to assess the model's performance.RMSE 1 compares the average sequence of 2,000 agent-based model (ABM) instances with a population of 50,000 to the new infection data.RMSE 2, on the other hand, compares the average sequence of 500 ABM instances to the sequence generated by the corresponding approximation model.