Pediatric Walking

Motion data acquisition using wearable sensors and their processing form an essential research topic with extensive applications in physiological motion monitoring, gait disorders detection, and motion patterns classification important both in pediatrics and neurology. This paper presents the use of accelerometric data and the global navigation satellite system (GNSS) for the body motion symmetry evaluation of children of different age, diagnosis, body mass index (BMI), and gender. Signals were recorded in natural conditions using mobile sensors as an alternative to data acquisition in the clinical environment. The datasets include signals acquired during walks of 35 children with motion trajectories monitored by the GNSS system. The proposed computational methodology based on accelerometric data analysis uses general methods of digital signal processing, feature extraction both in the time and frequency domains, and their classification. Selected evaluation methods include support vector machine, Bayesian, k-nearest neighbour methods, and neural networks. The goal of the study is in the comparison of the symmetry level estimated by the specialist and an analysis of the relation between the BMI and walking symmetry. The highest accuracy was achieved by a two-layer neural network for classification into two classes based on left and right body walking features to analyze different symmetry levels. Proposed methods demonstrate the abilities of general signal processing tools in rehabilitation, fitness movement monitoring, and neurology.