Implementing an accelerometer-based pelvis segment for low back kinetic analyses during dynamic movement tasks.

Abstract

An accelerometer-based pelvis has been employed to study segment and joint kinematics during scenarios involving close human-object interface and/or line-of-sight obstructions. However, its accuracy for examining low back kinetic outcomes is unknown. This study compared reaction moments and contact forces of the L5S1 joint calculated with an accelerometer-based and optically tracked pelvis segment. An approach to correct the global pelvis position as a function of thigh angle was developed. One participant performed four dynamic tasks: forward bend, squat, sit-to-stand-to-sit, and forward lunge. A standard bottom-up inverse dynamics approach was used and the root mean square error (RMSE) and coefficient of determination (R²) were calculated to examine kinetic differences between the optical and accelerometer approaches. The RMSE observed for L5S1 reaction flexion-extension moments ranged from 1.32 Nm to 2.20 Nm (R² ≥ 0.98). The RMSE for net shear and compression reaction forces ranged from 2.13 to 10.45 N and 0.63 - 4.96 N, respectively. Similarly, the RMSE for L5S1 joint contact shear and compression ranged from 13.45 N to 19.51 N (R² ≥ 0.85) and 31.18 N - 55.97 N (R² ≥ 0.97), respectively. In conclusion, the accelerometer-based pelvis together with the approach to correct the global pelvis position is a feasible approach for computing low back kinetics with a single equivalent muscle model. The observed error in joint contact forces represents less than 5 % of the NIOSH recommended action limits and is unlikely to alter the interpretation of low back injury risk.