Prediction of the falling occurrence during gait using the information from angular position of knee joint and upper extremity

Falling in the people such as the elderly and patients who suffer from balance problems is a serious risk that in most cases results in irreparable damage including bones fracture and death. Identification of abnormal gait in such people and prediction of a fall before impact with the ground is very important. This article aims to provide a solution to anticipate falling based on the hidden Markov chain. Motion analysis cameras were implemented to obtain data of upper body and knee joint.14 healthy volunteers were tested under a special protocol to record data. In order to evaluate the performance of the proposed system in terms of error rate and exact separation of abnormal movements from normal activities, HMM approach, a method based on support vector machine as well as multilayer neural networks were used and the results were compared. The results indicate that the proposed approach, predicted falling in the period from 200 to 550 milliseconds before impact with accuracy of 92.63 ± 0.62. Mentioned time is appropriate for providing preventive mechanisms in order to protect different parts of the body including the femur, pelvis and head from injuries and fractures caused by falls.

Keywords:

hidden marcov chain , lack of balance , gait , prediction of falling