Inferring the timing of individual mobility decisions from accommodation reservation data during the COVID-19 outbreak.

Understanding the changes in human mobility in response to outbreaks is important for controlling emerging infectious disease outbreaks. This requires an understanding of the mechanism of human behavioural response as well as the timing of decisions for future mobility. However, most human mobility data only record the executed mobility that results from decision-making, and not the timing of decisions. In this study, we used accommodation reservation data to extract the decision-making process in response to the changing epidemic situation and compared it with data on executed mobility, 'stay time' in workplaces and stay time in places other than home or workplaces to clarify when people decide on their mobility. We confirmed that the decision-making process estimated from accommodation reservation data can accurately predict human mobility. The decision-making process estimated from accommodation reservation data was more strongly associated with stay time in places other than home or workplaces than stay time in workplaces. Furthermore, the comparison between the estimated decision-making process and mobility data quantitatively revealed that mobility was the result of integrating two types of decisions made in recent weeks (within two and five weeks for mobility to workplaces and places other than home or workplaces, respectively) and previous weeks.