Walking pattern classification using a granular linguistic analysis

Graphical abstractDisplay Omitted HighlightsBased on previous research in the field of computing with perceptions.Highly interpretable and efficient linguistic model used to recognize the gait phases.Set of parameters that characterize relevant aspects of the gait.Fuzzy rule-based classifier that discriminates among different walking patterns.Parameters capable of recognizing among five walking patterns with an accuracy of 84%. Classifying walking patterns helps the diagnosis of health status, disease progression and the effect of interventions. In this paper, we develop previous research on human gait to extract a meaningful set of parameters that allow us to design a highly interpretable system capable of identifying different gait styles with linguistic fuzzy if-then rules. The model easily discriminates among five different walking patterns, namely: normal walk, on tiptoes, dragging left limb, dragging right limb, and dragging both limbs. We have carried out a complete experimentation to test the performance of the extracted parameters to correctly classify these five chosen gait styles.