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

Area of Science:

• Biomechanics
• Computational modeling
• Medical engineering

Background:

• Finite element analysis (FEA) is crucial for studying foot biomechanics and evaluating therapeutic interventions.
• Traditional FEA for foot models is computationally intensive and time-consuming, limiting real-time applications.
• Developing efficient computational tools is essential for advancing foot function analysis.

Purpose of the Study:

• To develop and validate a novel method combining FEA with model order reduction (MOR) for real-time dynamic foot function analysis.
• To create a computationally efficient prediction model for the center of pressure (COP) path.
• To identify key biomechanical parameters influencing foot function.

Main Methods:

• A generic, parametric finite element (FE) foot model was created and dynamically validated during the gait stance phase.
• A design of experiments approach with 30 FE simulations was used to train the MOR technique.
• The MOR prediction model for COP path was validated using four additional random simulations.

Main Results:

• The coupled FEA-MOR method achieved real-time analysis, predicting COP paths with a 3% root-mean-square error (RMSE) in under 1 second.
• Subtalar joint position and midtarsal joint laxity were identified as the most influential factors affecting foot function.
• The MOR technique significantly enhanced the speed and efficiency of FE foot analysis.

Conclusions:

• Model order reduction (MOR) offers a powerful approach to accelerate finite element analysis (FEA) for foot biomechanics.
• This method enables real-time analysis, paving the way for dynamic decision support tools in clinical settings.
• Understanding the influence of subtalar and midtarsal joints is key for accurate foot function modeling.