Classical and Intelligent ARX Models for Classification of Gait Events

Gait event detection is important for diagnosis and evaluation. This is a challenging endeavor that can be addressed with Computational Intelligence (CI). Four different CI models were developed and compared. Spatio-temporal parameters during normal walking in a treadmill were collected from a healthy volunteer. Gait events were classified by three experts in human motion. All identification systems were trained and tested with the collected data and experts´ mean classification. Fit percentage was obtained to evaluate models performance. Nonlinear Autoregressive Models with Exogenous Variables (NARX) had the best performance for gait events classification with a fit percentage of 88.59%. High frequency components were the main source of error for classical models. NARX was able to integrate criteria from the three experts for gait event detection. NARX models are suitable for gait event identification. Future work will include implementation of supervisory systems and additional data.