Real-Time Analysis of the Dynamic Foot Function: A Machine Learning and Finite Element Approach - Aix-Marseille Université Accéder directement au contenu
Article Dans Une Revue Journal of Biomechanical Engineering Année : 2021

Real-Time Analysis of the Dynamic Foot Function: A Machine Learning and Finite Element Approach

Tristan Tarrade
  • Fonction : Auteur
Nawfal Dakhil
  • Fonction : Auteur
Dorian Salin
  • Fonction : Auteur
Maxime Llari

Résumé

Abstract Finite element analysis (FEA) has been widely used to study foot biomechanics and pathological functions or effects of therapeutic solutions. However, development and analysis of such foot modeling is complex and time-consuming. The purpose of this study was therefore to propose a method coupling a FE foot model with a model order reduction (MOR) technique to provide real-time analysis of the dynamic foot function. A generic and parametric FE foot model was developed and dynamically validated during stance phase of gait. Based on a design of experiment of 30 FE simulations including four parameters related to foot function, the MOR method was employed to create a prediction model of the center of pressure (COP) path that was validated with four more random simulations. The four predicted COP paths were obtained with a 3% root-mean-square-error (RMSE) in less than 1 s. The time-dependent analysis demonstrated that the subtalar joint position and the midtarsal joint laxity are the most influential factors on the foot functions. These results provide additionally insight into the use of MOR technique to significantly improve speed and power of the FE analysis of the foot function and may support the development of real-time decision support tools based on this method.
Fichier non déposé

Dates et versions

hal-03716430 , version 1 (07-07-2022)

Identifiants

Citer

Tristan Tarrade, Nawfal Dakhil, Michel Behr, Dorian Salin, Maxime Llari. Real-Time Analysis of the Dynamic Foot Function: A Machine Learning and Finite Element Approach. Journal of Biomechanical Engineering, 2021, 143 (4), ⟨10.1115/1.4049024⟩. ⟨hal-03716430⟩
29 Consultations
0 Téléchargements

Altmetric

Partager

Gmail Facebook X LinkedIn More