Evaluation by accelerometry of walking pattern before falls in hemiplegic patients

Hemiplegic patients often fall because of a lack of balance-during walking. They can become bed-ridden, or suffer failing syndrome after falling. The aim of this study was to evaluate the walking pattern just prior to falls among high-risk patients in a rehabilitation setting. A triaxial accelerometer was fixed to the subject's waist, and the triaxial acceleration signals were recorded. Thirty-one subjects walked down a corridor under supervision. The data were digitized at a sampling rate of 200 Hz, and analysed using a discrete wavelet transform. The variables required to evaluate falls were related to the reconstructed signal at level -3. We classified falls into three types. In Type 1 falls, the impact acceleration signals in the vertical direction were larger in walking just prior to a fall. In Type 2 falls, small impact acceleration signals in the vertical direction were observed. In Type 3 falls, the walking cycle changed irregularly just prior to a fall and the impact acceleration was larger and smaller before the fall. From these results, falls were evaluated. The next step will be predicting falls.