A smartphone game is helping researchers understand how people may actually behave during pandemics, potentially informing future public health interventions.
PSI and UMass Chan Medical School researchers, alongside colleagues from across the University of Oxford and other UK, European and international institutions, have presented their experimental epidemic games (“epigames”) approach in a recent Nature Health article.
Epigames simulate an infectious disease outbreak in real-life environments such as a conference or a workplace, observing the spread of a fictional pathogen in real time and generating crucial data for a next generation of realistic and sophisticated policy-relevant models.
The Epigames app offers a template for this approach that leverages smartphones’ proximity sensing capabilities: participants download the app, provide informed consent to have their data used for research, and interact with others as usual. As the epigame progresses, the Epigames app measures the proximity between participants via the Bluetooth signal in their phones; after the status of some initial randomly selected participants changes to “infected”, they become able to spread the disease to others within the game, triggering a simulated outbreak that yields the real-life contact network of the participants in high resolution.
Users are also able to make decisions within the app, getting rewards or penalties for actions such as isolating, getting vaccinated or infecting others; and to respond to survey questions about their knowledge and attitudes on infectious disease prevention. This adds a layer of behavioural data to the contact network automatically collected by the app, something that is unique to the epigames approach.
Several epigames have already been run in conference and campus environments, including the 2024 and 2025 editions of the International Pandemic Sciences Conference. A pre-print analysing results, now submitted for publication, is available at Zenodo, providing a proof of concept for the methodological approach and confirming that the networks derived by epigames reflect context-specific social organisations. Another pre-print analysis on medRxiv, and also submitted for publication, has additionally explored the usefulness of the behavioural data generated during epigames.
First author and co-lead Andrés Colubri, Assistant Professor at UMass Chan Medical School, said: “Epigames are the result of many ideas evolving and cross-pollinizing each other in the past 20 years or more, from participatory simulations and field experiments to proximity sensing and the use of games for research.
“What is different about epigames, compared to previous similar concepts, is their malleability to different settings and contexts, the ability to integrate multiple layers of information, and the high degree of mechanistic realism via simulating disease spreads over proximal interactions just as direct contact pathogens do.”
PSI’s Associate Professor Jasmina Panovska-Griffiths, senior author and co-lead on the study, commented: “In modelling of infectious diseases spread, data on the underlying networks on which pathogens spread, including their temporal and spatial structures and how interventions alter the spread are scarce, inconsistent and seldom incorporate behavioural features.
“Epigames address and fill this knowledge gap between policy-relevant models of pathogen transmission and the data they require: contact networks at high spatial and temporal resolution, behavioural data and environmental factors.”
Not only do epigames offer a more realistic picture of contact networks, measuring number, duration and proximity of interactions in a live scenario; they also integrate behavioural data from decisions, attitudinal data from surveys, and environmental data that may influence participants’ interactions.
By analysing different datasets and bringing together real and simulated environments, epigames provide more accurate insights into how people may respond during actual pandemics, addressing the limitations of self-reporting as well as concerns about the validity of simulated experiments.
Since the epigames approach and resulting models are grounded in empirical data, their outputs are more likely to be trusted by decision-makers compared to purely computational approaches.
“Furthermore, due to their flexible construction, epigames are adaptable and able to test a wide range of pharmaceutical and non-pharmaceutical interventions to reduce disease spread”, added Professor Panovska-Griffiths.