An IMU-based automatic measurement system for detecting postural changes during sedentary work activities

Prolonged sitting while working on a computer is closely associated with postural changes that significantly increase the risk of musculoskeletal disorders. The effectiveness of a wearable configuration comprising five Inertial Measurement Units (IMUs) in detecting incorrect seated postures is evaluated in this study, while the most informative sensor locations and kinematic features for workplace monitoring are identified. Twenty healthy young adults performed four standardised seated tasks involving trunk flexion and lateral inclination, while Movella DOT sensors were positioned on their foreheads, acromion processes and T5 and L5 vertebrae. Data were collected at 120 Hz in order to quantify variations in segmental orientation during repeated trials. Analysis revealed that central tendency measures (specifically the mean, and the 25th and 75th percentiles) exhibited greater discriminatory power than intra-trial stability metrics and amplitude-based measures. In terms of sensor placement, units located on the shoulders and forehead were found to be the most effective, achieving optimal Effect Index (EI) values. However, although dynamic parameters such as jerk reached statistical significance (p < 0.05) in critical segments such as T5 and L5, their low selectivity suggests they play a marginal role in static postural classification. These results provide a solid foundation for the development of optimised monitoring systems. Focusing on a simplified set of high-performance sensors and features reduces computational complexity without compromising detection accuracy.