LAUSR

Gait quantitative evaluation plays an important role in indicating gait ability, comparing the effects of different rehabilitation programs, and indicating rehabilitation plans. However, traditional gait quantitative evaluation methods are mostly based on a variety of gait subscales and rely on doctors’ evaluations. Moreover, patients need to cooperate to complete a variety of movements, which is highly complex. To achieve objective gait assessment and reduce the difficulty of gait assessment and the dependence on doctors, we attempted to use inertial measurement units (IMUs) and plantar pressure insole sensors for gait quantitative assessment. We collected information on various joint angles and plantar pressure insole of stroke patients and healthy people during walking and scored the patients with the Mini Balance Evaluation System (Mini-BES) test. Then, based on the collected data, we performed gait division and feature extraction and used the random forest to obtain the feature importance. Finally, we used the importance-weighted Euclidean distance to characterize the walking ability and compared it with the Mini-BES test score. The experimental results demonstrate that the wearable device has good accuracy for gait assessment, and the collected data has good repeatability. The linear correlation between the extracted distance score and the Mini-BES test score is −0.94. Thus, it is feasible to use wearable devices to evaluate lower limb rehabilitation through a walking test, which could potentially replace the complex Mini-BES test.