Additive manufacturing of cranial implants: Experimental evaluation and machine learning prediction of mechanical and dimensional performance

Abstract

This study investigates the additive manufacturing (AM) of defect-specific cranial implants produced by fused deposition modeling (FDM) using three different polymeric materials, namely polylactic acid (PLA), polyamide (PA), and polyether ether ketone (PEEK). The primary objective is to evaluate the effects of key printing parameters on the dimensional accuracy and mechanical performance of the fabricated implants as well as to develop machine learning (ML) models to predict the main output responses. For this purpose, material type (MT), printing speed (PS), infill density (ID), layer height (LH), and build orientation (BO) were selected as input parameters and organized according to an L27 Taguchi experimental design. The fabricated implants were evaluated in terms of maximum length, thickness, maximum compressive load, and impact resistance. Analysis of variance revealed that MT, ID, LH, and BO affected maximum compressive load, whereas PS was not significant for this response within the investigated range. Gaussian process regression (GPR), support vector regression (SVR), and decision tree regression (DTR) models were used to predict the mechanical responses of the implants. Among them, GPR showed the highest predictive accuracy, achieving (Formula presented.) for compressive load and (Formula presented.) for impact resistance.