Predicting the polyethylene wear rate in pin-on-disc experiments in the context of prosthetic hip implants: deriving a data-driven model using machine learning methods.

Abstract

Pin-on-disc (PoD) experiments are widely used to quantify and rank wear of different material couples for prosthetic hip implant bearings. However, polyethylene wear results obtained from different PoD experiments are sometimes difficult to compare, which potentially leaves information inaccessible. We use machine learning methods to implement several data-driven models, and subsequently validate them by quantifying the prediction error with respect to published experimental data. A data-driven model can supplement results from PoD wear experiments, and enables predicting polyethylene wear of new PoD experiments based on its operating parameters. It also reveals the relative contribution of individual PoD operating parameters to the resulting polyethylene wear, thus informing design of experiments, and potentially reducing the need for time consuming PoD wear measurements.