| Description |
Many manual laborers develop musculoskeletal disorders performing their work duties. The prevalence of these issues is concerning because of the negative effects on workers' health and well-being, the great cost to workers who are not able to work temporarily or permanently, and the great cost to companies in worker's compensation payments. With the aim of combating this problem, this work is focused on how to accurately identify, based on sensor readings, what task a subject is performing. From an identified task, a risk score can be calculated to measure danger to an individual's musculoskeletal system. Various machine learning algorithms were tried to see what classifiers label more tasks correctly. The aim is that a good classifier can be used as part of the Lifting Coach system, which combines embedded sensors with smart processing to combat musculoskeletal disorders. This research found that classifiers that had access to richer information, especially when that information was normalized, performed better. Random Forests performed the best of the classifiers. |
