Triboinformatic Modeling of Wear in Total Knee Replacement Implants Using Machine Learning Algorithms

Abstract

Pin-on-disk (PoD) tests, the most prevalent studies, are being carried out in order to evaluate tribological behaviour of different bearing materials. However, the comparison of results obtained from the PoD tests is very difficult. In this present study, several machine learning models were developed and trained and then these trained machine learning models were validated by quantifying forecasting error against the experimental data reported in literature. These machine learning based models can be utilized as alternative solution of PoD trials in order to minimize time consumption and experiment complexity.