The evaluation of gait impairment by inertial measurement unit(IMU)attracts extensive attention.The existing methods ignore the importance of movement posture during walking and cannot effectively represent the overall impairment state.An IMU gait feature model derived from multi-source adaptive Kalman data fusion is proposed.This model uses adaptive Kalman data fusion of acceleration and angular velocity to estimate the posture angle

and further segments the gait phase by zero-velocity update technology.Seven lower limb spatiotemporal and kinematic features are extracted from two IMUs attached to the lower leg to quantify gait differences

and gait impairment index is designed to characterize individualized overall gait ability.To this end

the researchers recruit patients with lumbar disc herniation(LDH)and healthy subjects for walking experiments

and the classification accuracy reaches 92.86% using machine learning method.The analysis of gait impairment index shows that the new gait variable set and gait impairment index provide a more comprehensive means for clinical evaluation and diagnosis of gait impairment.