作者: Marzieh M. Ardestani , Mehran Moazen , Zhenxian Chen , Jing Zhang , Zhongmin Jin
DOI: 10.1016/J.NEUCOM.2014.12.005
关键词: Kinematics 、 Gait (human) 、 Gait analysis 、 Mathematics 、 Structural engineering 、 Finite element method 、 Thrust 、 Surrogate model 、 Time delay neural network 、 Simulation 、 Trunk
摘要: Knee contact pressure is a crucial factor in the knee rehabilitation programs. Although can be estimated using finite element analysis, this approach generally time-consuming and does not satisfy real-time requirements of clinical set-up. Therefore, surrogate method to estimate would advantageous.This study implemented novel computational framework wavelet time delay neural network (WTDNN) provide estimation at medial tibiofemoral interface implant. For number experimental gait trials, joint kinematics/kinetics resultant were computed through multi-body dynamic explicit analyses establish training database for proposed WTDNN. The trained was then tested by predicting maximum implant two different patterns; "medial thrust" "trunk sway". WTDNN predictions compared against calculations from an analysis (gold standard).Results showed that could accurately calculate thrust ( R M S E ? =1.7MPa, N =6.2% =0.98) trunk sway =2.6MPa, =9.3%, =0.96) much faster than method. methodology therefore serve as cost-effective model evaluation retraining programs terms pressures.