IoT Device for Sitting Posture Classification Using Artificial Neural Networks

Nowadays, the percentage of time that population spends sitting has increased substantially due to use computers as main tool for work or leisure and increase in jobs with a high office workload. As consequence, it is common suffer musculoskeletal pain, mainly back, which can lead both temporary chronic damage. This pain related holding posture during prolonged period sitting, usually front computer. presents IoT monitoring system while sitting. The consists device equipped Force Sensitive Resistors (FSR) that, placed on chair seat, detects points where user exerts pressure when complemented Machine Learning model based Artificial Neural Networks, was trained recognize neutral correct well six most frequent postures involve risk damage locomotor system. In this study, data collected from 12 participants each seven positions considered, using developed sensing device. Several neural network models were evaluated order improve classification effectiveness. Hold-Out technique used guide training evaluation process. results achieved mean accuracy 81% by means consisting two hidden layers 128 neurons each. These demonstrate feasible distinguish different few sensors allocated surface implies lower costs less complexity