Preclinical validation of finite element models for predicting in vivo residual plate bending using continuous implant sensor data.

Background/objective: Plate failure, including bending, is a critical issue in orthopedic fracture fixation, with clinical failure rates of 3.5%-19%, burdening patients and healthcare systems. Preclinical ovine models have observed similar plate bending due to overloading. Finite element (FE) models could be capable of predicting failures but lack Methods: Tibiae of eleven sheep with osteotomy gaps (0.6 - 30 mm) were instrumented with stainless steel or titanium locking plates equipped with AO Fracture Monitors in a prior study. Residual plate bending angles were measured using co-registered CT scans at 0 and 4 weeks post-operation, with bending defined as ≥ 1°. Animal-specific FE models, incorporating virtual AO Fracture Monitors and non-linear implant material properties, were developed to determine sensor signals at the construct's yield point. Results: Within 4 weeks, plate bending angles ranged from 0.4° to 10.4°, with overloading bending observed in 6 animals. The FE methodology correctly predicted bending/no-bending outcomes in 9 of 11 animals, achieving 100% sensitivity and 60% specificity. Conclusions: This sensor-validated FE methodology robustly predicted The Translational Potential of this Article: This validated FE methodology, leveraging AO Fracture Monitor data, can be adapted for human use to tailor rehabilitation protocols immediately post-surgery and provide real-time feedback to patients and clinicians if loading exceeds safe thresholds. This approach could minimize implant failure, reduce revision surgeries, and enhance patient recovery.