Predicting the compressive strength of additively manufactured PLA‐based orthopedic bone screws: A machine learning framework

Abstract Additive manufacturing (AM) technology is an innovative technique that has shown potential in several surgical innovations such as the fabrication of cost‐effective orthopedic screws. It is imperative that the process parameters used during the fabrication of bone screws determine the strength of the screw to support the load‐bearing fracture sites. In this present work, the application of machine learning (ML) models was leveraged for predicting the compressive strength of additively manufactured orthopedic cortical screws. Different ML models such as k‐nearest neighbors, support vector regression, decision trees, and random forest were compared to offer a robust predictive ML model. During the study, fused deposition modeling based additive manufacturing technology is used to fabricate poly(lactic acid) based cortical screws and compressive strength was monitored using the universal testing machine. The predictive ML models were developed by various independent parameters such as infill percentage, layer height, infill pattern and wall thickness. The adequacy and robustness of different ML models were observed at different error metrics. The error metrics of the random forest model were comparatively lower for testing data with a mean absolute error of 2.06 ± 0.91, root mean square error of 2.37 ± 0.9 and coefficient of determination of 0.96 ± 0.042. The results highlight that the random forest is the most adequate ML model for predicting the compressive strength of additively manufactured cortical screws.