Evaluation of methods for the development of representative responses and corridors from biomechanical data using mechanical models

Abstract

The development of representative responses and corridors from biomechanical data is essential for the validation of computational models and anthropometric test devices. While various empirical and statistical methods have been proposed for generating representative responses and corridors from data, it is often unclear on what basis the generated representative responses and corridors typify the biomechanical behaviour or its range of variations. The quality of the representative responses and corridors generated by various techniques from the literature was evaluated by comparing two linear mass-spring-damper models with the responses of average (or median) models and simulations using random resampling. One of these models was validated to lower limb axial impact experiments in order to provide a realistic evaluation of actual biomechanical data. This lower limb model has parameters dependent on the anthropometry and bone mineral density of the specimens and was significantly better in predicting the responses of a target specimen than the mass-scaling method. While more intricate scaling methods are possible, this result implies that model-based approaches may be a better tool for predicting the responses of a target population than the traditional mass-scaling method. The evaluation results showed that the consistency of any method's representativeness relative to the ‘true’ corridor was dependent on the type of data collected in the original experiments. Since there were inconsistencies in the quality of the generated corridors with these two models, a model-based evaluation should be considered as an additional step in choosing a corridor development method for a particular problem.