Comparing Armband EMG-based Lifting Load Classification Algorithms using Various Lifting Trials

Abstract

The objective of this study was to evaluate the performance of machine learning (ML) algorithms developed using surface electromyography (EMG) armband sensor data in predicting hand-load levels (5 lb and 15 lb) from diverse lifting trials. Twelve healthy participants (six male and six female) performed repetitive lifting with three different lifting conditions, i.e., symmetric (S), asymmetric (A), and free-dynamic (F) lifts. ML models were developed with four lifting datasets (S, A, S+A, and F) and were cross-validated using F as the test dataset. Mean classification accuracy was significantly lower in models developed with the S dataset (78.8%) compared to A (83.2%) and F (83.4%). Findings indicate that the ML model developed with controlled symmetric lifts was less accurate in predicting the load of more dynamic, unconstrained lifts, which is common in real-world settings.