Machine Learning Prediction of Vaginal Tissue Tears Using Finite Element Simulations Informed by Planar Biaxial Testing.

purposeVaginal tissue tearing during childbirth is a prevalent injury with potential long-term consequences, yet the mechanical factors influencing tear behavior remain poorly understood due to the difficulty of conducting in vivo studies. This study aims to investigate how tear geometry and fiber orientation influence local stress and strain patterns, and whether machine learning (ML) can accurately predict tissue responses from finite element (FE) simulation data.

methodsPreviously characterized swine vaginal tissue, tested via planar biaxial loading, was used to calibrate the Holzapfel-Gasser-Ogden (HGO) hyperelastic constitutive model. FE simulations were conducted for a range of tear orientations, tear sizes, and fiber alignments to evaluate stress and strain distributions near the tear boundary. Four ML models, linear regression with Stochastic Gradient Descent, Random Forest, Support Vector Regression, and Extreme Gradient Boosting (XGBoost), were trained and evaluated on the FE-generated dataset.

resultsThe HGO-based FE simulations reproduced experimental results with high fidelity, validating the modeling approach. XGBoost achieved the highest predictive accuracy across all mechanical and geometric outputs. Sensitivity analysis via permutation tests revealed that the initial orientation of the elliptical tear was the most influential predictor of high local stress and strain at the tear during extension.

conclusionThis integrated FE-ML approach enables rapid, accurate prediction of vaginal tissue behavior under varied tear conditions without additional simulations or experimental tests. By leveraging synthetic data, it provides a powerful tool for injury risk evaluation and supports clinical decision-making in maternal health, particularly where in vivo studies are limited.