BACKGROUND:
Understanding when people make mobility decisions—whether for travel, commuting, or daily activities—during infectious disease outbreaks is critical for predicting epidemic dynamics and guiding public health interventions. Traditional mobility datasets capture realized movement but not the timing of underlying decisions that drive behavior.

METHODS:
We analyzed online accommodation reservation data from four Japanese prefectures (Miyagi, Aichi, Osaka, Fukuoka) from 2015–2023. New bookings and cancellations were interpreted as decisions to increase or reduce contact, respectively. Predicted mobility was compared with anonymized mobile phone location data (“stay time”) from 2018–2022, covering workplaces and other locations. We quantified the influence of past decisions on current mobility and examined effects by travel distance (short vs long).

RESULTS:
Mobility outside workplaces was accurately predicted from reservation-based decision indices (R² = 0.775), while workplace mobility was less predictable (R² = 0.264). Decisions made within the previous five weeks had the strongest impact on current mobility outside workplaces, with smaller contributions from older decisions. Long-distance mobility was more influenced by past decisions than short-distance mobility, indicating integration of multiple weeks of behavioral choices. These findings clarify the timing of mobility decisions relative to epidemic dynamics.

CONCLUSIONS:
Accommodation reservation data provide a real-time window into the timing of human mobility decisions—including travel, commuting, and daily movement—during infectious disease outbreaks. Recognizing that current mobility reflects cumulative past decisions can improve epidemic forecasting and inform public health planning and interventions.