An artificial intelligence-based model for cell killing prediction: development, validation and explainability analysis of the ANAKIN model

Abstract

The present work develops ANAKIN: anArtificial iNtelligence bAsed model for (radiation-induced) cell KIlliNg prediction. ANAKIN is trained and tested over 513 cell survival experiments with different types of radiation contained in the publicly available PIDE database. We show how ANAKIN accurately predicts several relevant biological endpoints over a wide broad range on ion beams and for a high number of cell-lines. We compare the prediction of ANAKIN to the only two radiobiological models forRelative Biological Effectivenessprediction used in clinics, that is theMicrodosimetric Kinetic Modeland theLocal Effect Model(LEM version III), showing how ANAKIN has higher accuracy over the all considered cell survival fractions. At last, via modern techniques ofExplainable Artificial Intelligence(XAI), we show how ANAKIN predictions can be understood and explained, highlighting how ANAKIN is in fact able to reproduce relevant well-known biological patterns, such as the overkilling effect.

Keywords: RBE model; cell survival prediction; deep learning; machine learning; radiobiological modelling.