236 Introduction In gait laboratory, ground reaction forces (GRFs) are recorded using floor-mounted force platforms in order to eval- uate GRF patterns and to compute the joint torques using either the inverse dynamic approach 1-4 or the angular momentum cal- culus 5-7 . Traditionally, to obtain valid measurement the subject must place two consecutive steps on individual force platforms with no foot contact outside the surface of the devices (usually 50×50 cm). This condition can lead to an important number of trials before achieving valid measurements because the sub- jects must not voluntarily ‘target’ the force plate with their footsteps. In fact, in target conditions, the participant is prone to change his gait pattern which could impact GRF measure- ments. A change in step length, as little as 10%, has been shown to affect GRFs 8 . Thereby, targeting has been cited as a major limitation of gait studies 9 . Gait studies have intuitively attempted to reduce the effect of targeting, and simultaneously reduce the number of redun- dant trials, by fine-tuning each subject starting position 9 in order to make the footsteps coincide with the force platforms. Nevertheless, the number of trials made to obtain valid GRF measurements could remain important because starting posi- tion definition is especially uncertain with people with im- paired gait, as well as in children, because of the spatiotemporal gait parameters variability 10 . Thereby, record- ing several trials across the force platforms, until a series of adequate tests have been performed, impose a lot of trials and can be very time consuming. The duration of gait evaluation is particularly problematic in children because their compli- ance to a research protocol is limited in time 11 . Moreover, in- creasing the number of trials with a participant prone to muscular fatigue can result in major methodological bias be- J Musculoskelet Neuronal Interact 2013; 13(2):236-243 Decomposition of the vertical ground reaction forces during gait on a single force plate L. Ballaz 1,3 , M. Raison 2,3 , C. Detrembleur 4 1 Department of kinanthropology, Université du Québec à Montréal, Montreal (Qc), Canada; 2 Department of mechanical engineering, École Polytechnique de Montréal, Montreal (Qc), Canada; 3 Research Engineering Chair Applied in Pediatrics (RECAP), Marie Enfant Rehabilitation Centre (CRME) – Research Center – Sainte-Justine UHC, and École Polytechnique de Montréal, Montreal (Qc), Canada; 4 Institute of NeuroSciences (IoNS), Université catholique de Louvain, Bruxelles, Belgium Abstract Davis and Cavanagh (1993) have proposed a solution to avoid the footstep targeting by using a large force plate but several points of Davis and Cavanagh’s method remain unclear and hardly computable. Objective: to develop a method that decomposes left and right GRF profiles from the GRF profile recorded on a single platform. This method aims to include a systematic detection of the single to double stand-phase-instants in order to lead to accurate measurement of the vertical GRF component in typically developing children. Methods: Six children were asked to walk without targeting their footsteps on a set-up composed of inde- pendent force platforms. The vertical GRF component, independently measured on the different platforms, was numerically summed to obtain the corresponding global vertical GRF, to which the decomposition method was applied. Then, the validation consisted in comparing the vertical GRF computed from this decomposition to the independently measured vertical GRF. Results: the mean relative error between the computed vertical GRF and the corresponding measured vertical GRF of 36 double stances (6 double stances × 6 children) is equal to 3.8±2.6 %. Conclusion: implemented a new method to assess with known accuracy the vertical GRF component under each foot using a unique large force platform. Keywords: Centre of Pressure, Double Stance, Ground Reaction Force Decomposition, Single Platform, Children Original Article Hylonome The authors have no conflict of interest. Corresponding author: Dr Laurent Ballaz, CRME – Research Center, 5200, East Bélanger Street, H1T 1C9, Montréal, QC, Canada E-mail: laurent.ballaz@uqam.ca Edited by: F. Rauch Accepted 22 April 2013