Adolescent idiopathic scoliosis (AIS) requires effective and personalized brace treatment strategies to prevent progression and the potential need for surgery. However, monitoring and prediction of the spinal column deformities during bracing is not always possible. Relying only on traditional methods or clinicians' experience may pose a complex challenge in orthopedic care, as unique biological characteristics of each patient make it difficult to decide on the optimal spinal brace treatment. This study introduces a comprehensive biomechanical modeling approach utilizing finite element analysis and neural networks to refine brace prescription and treatment outcomes. The Rigo Chêneau-type brace, known for its biomechanical principles targeting lateral displacement and transversal derotation, serves as the foundation for this study. A dataset of 120 diverse abnormal curvatures is analyzed to estimate the effectiveness of the proposed biomechanical brace mechanism prior to its design and fabrication. Through 3600 simulations across various curvature types and severity levels, the study evaluated brace performance in terms of coronal correction, sagittal plane stability, and stress distribution. Simulation findings indicate significant improvements in coronal alignment between 10% and 50% while preserving the physiological sagittal curves. Subsequently, simulation data were utilized for training a neural network model to estimate the spinal column position after using the prescribed brace. The trained scoliosis model demonstrated 90.6% accuracy in predicting spinal deviation changes. By leveraging advanced computational tools and patient-specific biomechanical data, the current simulations offer a promising approach for optimizing brace treatment in AIS patients by predicting the outcomes.
Talkhoncheh et al. (Fri,) studied this question.