Additive manufacturing and wearable sensors data-based machine learning modelling of optimized orthopedic wrist cast

Abstract More than 80% of the wrist load is applied on the distal radius, and distal radius fractures (DRFs) are the most traumatic wrist injuries. In the development of orthopedic wrist cast, customization, ventilation, weight, and recovery monitoring are the most important design targets. The CAD modelling and finite element analysis (FEA) of the cast using stiff polylactic acid (PLA) and soft thermoplastic polyurethane (TPU) layered composite are applied to develop customized wrist cast model with optimum strength and comfort. The ANSYS CFX-based topology optimization and computational fluid dynamics (CFD) analysis are used to develop optimum ventilation openings to dissipate heat generated at the cast-skin interfaces. The fused filament fabrication (FFF) based additive manufacturing (AM) is used to manufacture the optimized wrist cast model. The force and electromyography (EMG) wearable sensors data-based and genetic algorithm (GA)-tuned artificial neural network (ANN) machine learning modelling is used to determine integrated optimal cast system. The simulation results demonstrated that the cast is loaded with 26.7 MPa peak stress and 4.39 mm maximum deformation during typical wrist motions with a safety margin of 2.25. The optimized cast weighs 148 g and maintain the skin temperature to be below 33.85 °C. The GA-tuned ANN model is developed with 3000 datasets divided into (75%) training, (15%) testing, and (10%) validation tests. Finally, the optimal number of straps, and the tightening pressure are determined to be 3, and 11.24 mmHg respectively with R value of 0.9988.