Determination of genes potentially responsive to ionizing radiation through machine learning

Abstract

The prediction of cancer response to radiation therapy is one of the most important issues during the treatment of cancer patients. As a consequence, prediction of radiosensitivity is crucial for the improvement of clinical outcomes by optimizing the delivered doses and fractionation regimen. Moreover, the prediction of radiosensitivity can provide a better understanding of the underlying mechanisms responsible for radioresistance of cancer cells as well as the identification of biomarkers and potential drug targets of radiosensitivity. As experimental studies with the purpose of evaluating radiosensitivity are time-consuming and laborious, a computational approach which could predict radiosensitivity with high accuracy would be invaluable. We present here a novel machine learning-based computational approach that relies on network topology information of a gene to estimate its radiosensitivity.