Purpose: Personalized modeling of brace action has potential in improving brace efficacy in adolescent idiopathic scoliosis (AIS). Model validation and simulation uncertainty are rarely addressed, limiting the clinical implementation of personalized models. We hypothesized that a thorough validation of a personalized finite element model (FEM) of brace action would highlight potential means of improving the model.
Methods: Forty-two AIS patients were included retrospectively and prospectively. Personalized FEMs of pelvis, spine and ribcage were built from stereoradiographies. Brace action was simulated through soft cylindrical pads acting on the ribcage and through displacements applied to key vertebrae. Simulation root mean squared errors (RMSEs) were calculated by comparison with the actual brace action (quantified through clinical indices, vertebral positions and orientations) observed in in-brace stereoradiographies.
Results: Simulation RMSEs of Cobb angle and vertebral apical axial rotation was lower than measurement uncertainty in 79 % of the patients. Pooling all patients and clinical indices, 87 % of the indices had lower RMSEs than the measurement uncertainty.
Conclusions: In-depth analysis suggests that personalization of spinal functional units mechanical properties could improve the simulation's accuracy, but the model gave good results, thus justifying further research on its clinical application.
Keywords: Brace; Finite element model; Pediatrics; Spine deformity.