Assessing ergonomic and postural data for pain and fatigue markers using machine learning techniques

Ergonomic data obtained from trials with human participants at a number of workstations are evaluated in terms of whether different workstations elicit different fatigue and pain responses. Data is analyzed using a pair of simple machine-learning based classifiers in order to identify activities associated with the workstations that lead to or avoid pain and fatigue. Results indicate that information content sufficient to predict pain and fatigue is present in this data, with evidence of information increase consistent with postures held for a period of time. Additional analysis will be performed to isolate postures associated with fatigue and pain in follow-up work.