A real-world Fitbit-derived dataset of activity, sleep, and heart rate with matched clinical factors in on-treatment lung cancer patients

Consumer-grade wearable devices are gaining attention in oncology as tools to complement performance status assessment and predict short-term clinical outcomes. However, the limited availability of wearable data for research has constrained progress in this area. To address this gap, we present a wearable-derived dataset comprising activity, sleep, and heart rate data collected via Fitbit devices from 178 lung cancer patients undergoing anti-cancer treatment, with variable follow-up durations. The dataset also includes demographic and clinical variables, such as histological subtype, anti-cancer treatment regimen, performance status scores, and emergency department visits. Data quality was assessed using daily wear adherence derived from 1-minute heart rate records, showing that 81.8% (12,409/15,175) of patient-days met the 80% wear adherence threshold. We further examined validity by confirming previously reported physiological and behavioral patterns, including lower daily steps and distance and higher resting heart rate in patients with poorer performance status. These data may support studies of wearable-derived markers of performance status and the development and external validation of predictive models for short-term clinical outcomes.