Prediction of load in a long bone using an artificial neural network prediction algorithm

Area of Science:

• Biomechanics
• Orthopedics
• Artificial Intelligence in Engineering

Background:

• Bone's hierarchical structure complicates mechanical load analysis.
• Equine third metacarpal (MC3) bones experience complex, non-uniform surface strains under load.
• Quantifying bone loading from strain measurements presents an inverse problem challenging for conventional methods like finite element analysis (FEA).

Purpose of the Study:

• To investigate the efficacy of an artificial neural network (ANN) expert system for quantifying equine MC3 bone loading.
• To solve the inverse problem of determining load from experimental strain and displacement data.
• To develop a reliable method for predicting bone loading that surpasses FEA limitations.

Main Methods:

• Nine hydrated equine MC3 bones were subjected to compression loading in an MTS machine.
• Ex-vivo experiments recorded surface strain using rosette and single-element gauges, along with displacement and load exposure time.
• An artificial neural network (ANN) model was constructed using experimental data, including horse age and bone side, to predict loading.

Main Results:

• The ANN expert system demonstrated excellent reliability in predicting MC3 bone loading.
• The model achieved a high coefficient of determination (R² ≥ 0.98), indicating accurate load prediction.
• This approach successfully quantified load from experimental strain and displacement measurements.

Conclusions:

• Artificial neural networks provide a reliable and effective solution for the inverse problem of quantifying equine bone loading.
• The developed ANN expert system offers a powerful tool for understanding bone mechanics and predicting loads in MC3 bones.
• This study highlights the potential of AI in biomechanical analysis, offering an alternative to traditional FEA for complex inverse problems.