A framework for failure prediction models of medical electron linear accelerators

Among the available maintenance strategies, predictive maintenance seems to be the most promising for medical linear accelerators (linacs). Predictive maintenance predicts failures and allows timely reaction. Input data and model are important to implement predictive maintenance. The aim of this study is to provide a new framework including workflow, data and models that can be used for developing a predictive maintenance approach for medical linear accelerators. In this paper, 51 operational parameters and output performances data related to 15 systems of linacs, 8 environment data, processing data methods and 29 prediction models are identified. This work shows there is no standard failure prediction model apply to medical linacs.