Continuous remote monitoring of performance status in patients with NSCLC through fitness tracker–derived life-log data.

1568 Background: Traditional methods for assessing the performance status of cancer patients often rely on clinical evaluations and patient-reported histories. This study utilizes life-log data derived from fitness trackers (FTs) in patients with non-small cell lung cancer (NSCLC) to offer a comprehensive and dynamic perspective on performance status throughout their treatment. Methods: Patients with NSCLC scheduled for a minimum of 12 weeks of treatment were enrolled. Custom software facilitated the remote collection of life-log data from participants' FTs (Fitbit Inspire 3), encompassing step count, calories burned, heart rate, sleep patterns, etc. Quality of life surveys assessing patient-reported outcomes (PROs) were completed during regular visits (every 3 weeks) and unscheduled visits (USV), in addition to collecting clinical data such as laboratory parameters and body composition metrics. An end-of-monitoring survey assessed participants' familiarity with this monitoring approach. The study aimed to correlate FT data with clinical parameters and identify patterns predictive of treatment related USVs. Results: Of the 98 enrolled patients, 69 patients (70.4%) completed the 12-week follow-up. The cohort predominantly consisted of older adults (≥65 years) (55.1%) and patients in palliative care (75.5%), with most undergoing cytotoxic chemotherapy (66.3%) and some receiving oral agents (20%). FT-based remote monitoring revealed that older patients had a comparable 12-week completion rate to younger patients. (69.2% vs. 76.2%, P=0.604). Analysis of life-log data from three days before a visit (D-3) to the visit day (D0) showed significant reductions in average step counts ( P=0.022) and walking distance ( P=0.021) for USVs compared to regular visits, while resting heart rate (HR) showed a significant increase for USVs ( P=0.049). PRO-based PROMIS physical function scores and EQ-5D-3L scores showed a significant correlation with both D-3 to D0 acquired resting HR ( P<0.001, P=0.002) and average step counts ( P=0.011, P=0.013), respectively. End of monitoring survey was done from 72 participants indicating high satisfaction and familiarity with the FT-based remote monitoring system, median score of 90 out of 100. Conclusions: FT-derived life-log data effectively captured dynamic performance status in NSCLC patients, correlating with clinical outcomes and PRO. These findings support further research into FT data's predictive value for clinical interventions, underscoring its potential in personalized remote patient management.