Novel methods to detect impacts within whole-body vibration time series data

We present three candidate mathematical models for detecting impacts within time series accelerometer data in the context of whole-body vibration (WBV). In addition to WBV, data included recordings of erector spinae muscle activity and trunk posture collected during use of agricultural machines in a previous study. For each model, we evaluated associations between several mechanical and biomechanical variables at the time of predicted impact onset and the odds of subsequently observing a bilateral response of the erector spinae muscles. For all models, trunk posture at the time of impact onset was strongly associated with an observed bilateral muscle response; these associations were not observed when impacts were randomly assigned. Results provide a framework for describing the number and magnitudes of impacts that may help overcome ambiguities in current exposure metrics, such as the vibration dose value, and highlight the importance of considering posture in the evaluation of occupational WBV exposures. Practitioner summary: Common metrics of exposure to whole-body vibration do not quantify the number or magnitudes of impacts within time series accelerometer data. Three candidate impact detection methods are presented and evaluated using real-world data collected during use of agricultural machines. Results highlight the importance of considering posture when evaluating vibration exposure.