Radio frequency identification to measure the duration of machine-paced assembly tasks: Agreement with self-reported task duration and application in variance components analyses of upper arm postures and movements recorded over multiple days

Technical advances in inertial measurement units (IMUs) with data logging functionality have enabled multi-day collection of fullshift upper arm postures and movements. Such data are useful for characterizing job-level exposures and, when coupled with task-level information, can inform interventions to mitigate high-exposure tasks. Previously reported methods for capturing task-level information, however, were limited primarily to self-report diaries or direct observation. In this study of machine-paced manufacturing workers (n=6), a low-cost radio frequency identification (RFID) system was used to collect information about when, and for how long, specific assembly tasks were performed during up to 14 consecutive work shifts (76 total work shifts across the six participants). The RFID data were compared to information collected with a self-report diary using Bland-Altman analyses. In addition, the RFID data were paired with IMU data to identify task-level exposures from within full-shift recordings of upper arm postures and movements. These data were then used to estimate the relative contributions of between- and within-worker sources of variance to overall variance in posture and movement summary measures using hierarchical random-effects analysis of variance (ANOVA) techniques. Average estimates of daily task duration based on RFID data were comparable to estimates obtained by self-report (mean bias < ±1 minute) but with substantial variability (limits of agreement > ±100 minutes). In addition, the ANOVA models containing task-level information suggested a substantial amount of the overall exposure variance was attributed to repeated observations of the same task within a work day. These findings (i) suggest that while the RFID system used in this study performed adequately, further refinement, validation, and/or alternative strategies may be needed and (ii) underscore the importance of repeated full-shift and task-based measurement approaches in characterizing physical exposures, even in machine-paced environments.