Deep Learning for Accelerometric Data Assessment and Ataxic Gait Monitoring

Ataxic gait monitoring and assessment of neurological disorders belong to important multidisciplinary areas that are supported by digital signal processing methods machine learning tools. This paper presents the possibility using accelerometric data optimise deep convolutional neural network systems distinguish between ataxic normal gait. The experimental dataset includes 860 segments 16 patients 19 individuals from control set with mean age 38.6 39.6 years, respectively. proposed methodology is based upon analysis frequency components signals simultaneously recorded at specific body positions a sampling 60 Hz. system uses all in range (0,30) Our classification results compared those obtained standard methods, which include support vector machine, Bayesian two-layer features estimated as relative power selected bands. show appropriate selection sensor can increase accuracy 81.2% for foot position 91.7% spine position. Combining input five layers increased 95.8%. suggests artificial intelligence efficient motion they have wide further applications.