Journal of Biomechanics 38 (2005) 401–408 A machine learning approach for automated recognition of movement patterns using basic, kinetic and kinematic gait data R. Begg a, *, J. Kamruzzaman b a Biomechanics Unit, Centre for Rehabilitation, Exercise & Sport Science, City Flinders Campus, Victoria University, P.O. Box 14428, Melbourne City MC, Vic., 8001, Australia b Gippsland School of Computing & IT, Monash University, Churchill Campus, Vic., 3842, Australia Accepted 5 May 2004 Abstract This paper investigated application of a machine learning approach (Support vector machine, SVM) for the automatic recognition of gait changes due to ageing using three types of gait measures: basic temporal/spatial, kinetic and kinematic. The gaits of 12 young and 12 elderly participants were recorded and analysed using a synchronized PEAK motion analysis system and a force platform during normal walking. Altogether, 24 gait features describing the three types of gait characteristics were extracted for developing gait recognition models and later testing of generalization performance. Test results indicated an overall accuracy of 91.7% by the SVM in its capacity to distinguish the two gait patterns. The classification ability of the SVM was found to be unaffected across six kernel functions (linear, polynomial, radial basis, exponential radial basis, multi-layer perceptron and spline). Gait recognition rate improved when features were selected from different gait data type. A feature selection algorithm demonstrated that as little as three gait features, one selected from each data type, could effectively distinguish the age groups with 100% accuracy. These results demonstrate considerable potential in applying SVMs in gait classification for many applications. r 2004 Elsevier Ltd. All rights reserved. Keywords: Gait; Support vector machine; Gait classification; Elderly 1. Introduction It is well established that ageing influences gait patterns and considerable research has documented changes during unobstructed and obstructed walking that suggest age-related declines in lower limb control (Princea et al., 1997; Begg and Sparrow, 2000). The major aim has been to identify key variables of gait degeneration in elderly individuals that might be predictors of falling behaviour. Research has shown that significant changes in gait can occur with age in temporal and distance measures such as gait velocity, stride length, and stance and swing phase times (Hage- man and Blanke, 1986; Winter, 1991). In addition, foot- ground reaction force data during braking and propul- sive phases (Winter, 1991; Nigg et al., 1994; Begg et al., 1998) and joint angular motion data such as the ankle, knee and hip joint angles (Judge et al., 1996; Kerrigan et al., 1998) have shown effects of aging. To date, however, the relative influence of these measures in differentiating the age groups has not been demon- strated. Automated recognition of gait pattern changes by a machine classifier from their respective measures is expected to offer many potential advantages. For example, Maki (1997) using spatial-temporal measures of gait has shown significant changes in gait character- istics in the elderly fallers when compared to gait characteristics of elderly non-fallers. This research has particularly shown that some foot placement gait measures (e.g., step width and stride variability) displayed greater associations with falls prediction. Therefore, early identification of gait changes due to falling behaviour by a machine classifier might trigger initiation of necessary measures to prevent injurious falls such as an exercise intervention program (Lord ARTICLE IN PRESS *Corresponding author. Tel: +61-3-9248-1116; fax: +61-3-9248- 1110. E-mail address: rezaul.begg@vu.edu.au (R. Begg). 0021-9290/$-see front matter r 2004 Elsevier Ltd. All rights reserved. doi:10.1016/j.jbiomech.2004.05.002