Determining the minimum sampling rate needed to accurately quantify cumulative spine loading from digitized video

Cumulative low back loads have been linked to the reporting of low back pain. Traditional video-based methods used to estimate these loads are time intensive for data collection and analysis. Sampling less frequently would help to reduce the associated time and cost of this type of approach. The purpose of this study was to determine how the error in estimated cumulative low back loads is affected by reducing video sampling rate. Ten healthy male university students performed three laboratory, sagittal plane lifts of varying mass (2.3, 8.8, and 15.9 kg), speed (0.2, 0.4, 0.8 m/s), and postural demand (lift from floor to table; lower from shelf to table; lift from floor over barrier and lower to floor) while being videotaped (60 frames/s). Digitized body coordinates and anthropometrics were input into a static biomechanical model, resulting in estimates of low back compression and shear forces, and moment. Load-time histories for each condition underwent rectangular integration at 60 (gold standard), 30, 20, 15, 12, 10, 6, 5, 4, 3, 2 and 1 frames/s, resulting in estimates of low back cumulative loads. Mean relative errors with respect to 60 frames/s for all cumulative loads and all conditions were found to be below 8% at 1 frame/s, and less than 3% at 2 frames/s. In addition, analyses at sampling rates above 3 frames/s were not significantly different than the cumulative loads determined at 60 frames/s, for all conditions. The accuracy of cumulative loads exhibited even at low sampling rates can be attributed, in part, to the fact that overestimations and underestimations of the integrated loads tend to cancel out over the length of the tasks considered.