Abstract: Infections produced by non-symptomatic (pre-symptomatic and asymptomatic) individuals have been identified as major drivers of COVID-19 transmission. Non-symptomatic individuals, unaware of the infection risk they pose to others, may perceive themselves—and be perceived by others—as not presenting a risk of infection. Yet, many epidemiological models currently in use do not include a behavioral component and do not address the potential consequences of risk misperception. To study the impact of behavioral adaptations to the perceived infection risk, we use a mathematical model that incorporates the behavioral decisions of individuals based on a projection of the system's future state over a finite planning horizon. We found that individuals' risk misperception in the presence of non-symptomatic individuals may increase or reduce the final epidemic size. Moreover, under behavioral response, the impact of non-symptomatic infections is modulated by symptomatic individuals' behavior. Finally, we found that there is an optimal planning horizon that minimizes the final epidemic size.
Introduction
Traditional epidemiological models often treat human behavior as a fixed parameter. However, during a pandemic, people are active agents: they reduce their social contacts when they perceive high risk and increase them when they feel safe. A critical challenge with COVID-19 is that a significant portion of transmission is "silent"—driven by asymptomatic or pre-symptomatic individuals. Because these individuals do not exhibit symptoms, they (and those around them) often misperceive the risk of infection. This research, published in Scientific Reports, explores how this risk misperception, coupled with adaptive human behavior, changes the total number of people who eventually get sick (the "attack rate").
Methodology
The authors developed a mathematical model that bridges epidemiology and behavioral economics:
- Epidemiological Framework: An S-E-I-A-R compartmental model tracking Susceptible, Exposed, Symptomatic ($I$), Asymptomatic ($A$), and Recovered ($R$) populations.
- Behavioral Framework: A Markov Decision Process where individuals choose their daily number of contacts to maximize "utility" (social/economic benefit) while minimizing the "cost" (risk of infection).
The Novelty: The model incorporates a finite planning horizon ($\tau$). Individuals project the state of the epidemic into the future over a set number of days to decide how much to socially distance today. Crucially, the model assumes "risk misperception": the public only reacts to the prevalence of symptomatic cases, effectively ignoring the hidden risk posed by asymptomatic spreaders.
To calibrate the model, the authors used early COVID-19 literature for parameters such as the basic reproductive number ($R_0 \approx 2.4$), latency periods, and recovery rates, alongside asymptomatic estimates from studies in New York, Austin, and Wuhan.
The Adaptive Behavior Paradox
In standard models with fixed behavior, a higher proportion of asymptomatic cases usually results in a smaller epidemic because asymptomatic individuals are typically less infectious. However, when behavior is adaptive, this relationship flips. Because "silent" spread does not trigger the same behavioral response as visible symptoms, people "let their guard down," leading to a larger final epidemic size.
As shown in the figure above, comparing scenarios with lower ($\sigma = 0.3$) and higher ($\sigma = 0.6$) proportions of asymptomatic cases reveals that higher asymptomatic prevalence can lead to a higher final attack rate ($R(t)$). This occurs because the visible symptomatic prevalence ($I(t)$) remains lower, causing individuals to maintain higher contact rates ($C_t^S$) for a longer duration, inadvertently facilitating spread through the asymptomatic population ($A(t)$).
Key Insights and Results
- The Compliance Feedback Loop: If symptomatic individuals are highly compliant with isolation, it lowers the visible prevalence of the disease. This makes healthy people feel safer, leading them to increase their contact rates and mix more with asymptomatic spreaders.
- Threshold Effect: The study identifies a clear threshold where the trade-off between reduced infectiousness (of asymptomatics) and reduced risk perception (by the public) flips. If asymptomatic individuals are relatively infectious (relative infectiousness $\epsilon > 0.6$), the final epidemic size is higher than in a scenario with zero asymptomatic cases.
- The "Goldilocks" Planning Horizon: The length of time people look ahead matters. If the planning horizon is too short, people ignore future risks; if it is too long, they may overweight the eventual utility of being recovered. The authors found an optimal window (20–25 days in simulations) that minimizes the final epidemic size.
Limitations and Future Work
The model assumes a homogeneous population, whereas real-world behavior varies by age, occupation, and socioeconomic status. Additionally, the model assumes people have perfect (if biased) awareness of current symptomatic prevalence, which may be degraded by testing delays or information fatigue. The authors suggest that future models should incorporate mass testing and vaccination, which fundamentally change how risk is perceived and managed.
Summary updated