Human movement avoidance decisions during Coronavirus disease 2019 in Japan.

Understanding host behavioral change in response to epidemics is important to forecast the disease dynamics. To predict the behavioral change relevant to the epidemic situation (e.g., the number of reported cases), we need to know the epidemic situation at the moment of decision, which is difficult to identify from the records of actually performed human mobility. In this study, the largest travel accommodation reservation data covering half of the existed accommodations in Japan was analyzed to observe decision-making timings and how it responded to the changing epidemic situation during Japan's Coronavirus Disease 2019 until February 2023. To this end, we measured mobility avoidance index proposed in Ito et al., 2022 to indicate people's decision of mobility avoidance and quantified it using the time-series of the accommodation booking/cancellation data. We observed matches of the peak dates of the mobility avoidance and the number of reported cases, and mobility avoidance changed proportional to the logarithmic number of reported cases. We also found that the slope of mobility avoidance against the change of the logarithmic number of reported cases were similar among the epidemic waves, while the intercept of that was much reduced as the first epidemic wave passed by. People measure the intensity of epidemic by logarithm of the number of reported cases. The sensitivity of their response is established during the first wave and the people's response became weakened after the first experience, as if the number of reported cases were multiplied by a constant small factor.

Future behaviours decision-making regarding travel avoidance during COVID-19 outbreaks.

Human behavioural changes are poorly understood, and this limitation has been a serious obstacle to epidemic forecasting. It is generally understood that people change their respective behaviours to reduce the risk of infection in response to the status of an epidemic or government interventions. We must first identify the factors that lead to such decision-making to predict these changes. However, due to an absence of a method to observe decision-making for future behaviour, understanding the behavioural responses to disease is limited. Here, we show that accommodation reservation data could reveal the decision-making process that underpins behavioural changes, travel avoidance, for reducing the risk of COVID-19 infections. We found that the motivation to avoid travel with respect to only short-term future behaviours dynamically varied and was associated with the outbreak status and/or the interventions of the government. Our developed method can quantitatively measure and predict a large-scale population's behaviour to determine the future risk of COVID-19 infections. These findings enable us to better understand behavioural changes in response to disease spread, and thus, contribute to the development of reliable long-term forecasting of disease spread.