Calibration of low back load exposure estimation through surface EMG signals with the use of artificial, neural network technology

摘要

A new calibration method is proposed for ambulatory systems for low back load exposure estimation based on surface EMG and kinematic data. The method uses an artificial neural network to learn the relation between compressive force in the intervertebral disc at L/sub 4/-L/sub 5/ (C) and smoothed rectified surface EMG signal (SRE) under full dynamic conditions. In vivo tests show that a accurate calibration is possible selecting a training set of 600 samples out of 2 minutes of calibration data. This offers load exposure estimation sensitive to unknown time-varying external loads, compensated for force-length and force-velocity relationships and compensated for inter-individual load handling differences.