Feature extraction and modeling of the variability of performance in terms of biomechanical motion patterns during MMH tasks.

In investigating manual material handling (MMH) jobs, such as lifting, the quantification of the various kinematic and kinetic parameters of the lift is an important step towards functional assessment and evaluation. Experimental data collection generates a large quantity of data for the different kinetic, kinematic, and electromyographic parameters over the various lifting cycles. In order to efficiently manage and interpret the data, it is important to use appropriate tools which would reduce the dimension of the original data set without sacrificing any important features. Furthermore, the generated parameters are often expressed as a function of the lifting cycle resulting in complex waveforms as the unit of analysis. Appropriate statistical analysis of these waveforms or motion profiles should reflect their vectorial constitution as a function of the lifting cycle rather than the usual method of using traditional descriptive statistics based on collapsing the data over the cycle.