Implementation and evaluation of a system for automatic classification of feet and walking alterations using latex insoles with pressure sensors and accelerometers, based on artificial neural networks

Several muscular-skeletal problems can affect the foot and thus the structural development of a child, causing damage to his or her tone-postural system. Hence, the early detection of any possible postural alteration is essential in avoiding further adult life complications. There are several methods and instruments for feet evaluation; however, they are generally restricted to clinical environments, and are not appropriate for daily evaluations. We propose a system for feet alterations classification, for children with ages between 6 and 10 years, daily use, and based on latex insoles with pressure sensors and accelerometers. Our system uses a microcontrolled circuit connected to the insole, with a module for signals storage. For classification, the system uses an artificial neural network (ANN). In preliminary tests, a user applied specific plantar pressure distributions, for training the ANN. The ANN achieved 100 % of correct classifications for the signals used during training, and 96 % for the remaining signals. We also created a visualization of feet pressures during walking, in order to conduct a dynamic analysis of pressure distributions. The results suggest that system can classify feet and walking types. We will now conduct more systematic tests with more volunteers, to statistically evaluate the results.