Qualitative Risk of Falling Assessment Based on Gait Abnormalities

Walking in an unfamiliar environment may include some risks of falling. For frail seniors, these risks can be significantly increased according to their ability to maintain balance. Among several factors, the user's balance can be affected by several risks including the characteristics of the user's gait. To evaluate this issue, this paper presents three methods. The first uses a statistical model while the two others exploit an Artificial Neural Network (ANN). The latter two can be differentiated by the use of constraints applied onto the raw data. Centered on non-invasive augmented shoes, our proposed system uses mobile technology to provide an on-site assistance to users, replacing the bulky equipment usually needed for clinical gait analysis. The experimental framework is based on visual disturbances to induce variation in the parameters of the user's gait. Preliminary results obtained from this framework suggest that our models enable a risk level classification.