Clustering of Largely Right-Censored Oropharyngeal Head and Neck Cancer Patients for Discriminative Groupings to Improve Outcome Prediction

摘要: Clustering is the task of identifying groups similar subjects according to certain criteria. The AJCC staging system can be thought as a clustering mechanism that patients based on their disease stage. This grouping drives prognosis and influences treatment. goal this work evaluate efficacy machine learning algorithms cluster into discriminative improve for overall survival (OS) relapse free (RFS) outcomes. We apply over retrospectively collected data from 644 head neck cancer including both clinical radiomic features. In order incorporate outcome information process deal with large proportion censored samples, feature space was scaled using regression coefficients fitted proxy dependent variable, martingale residuals, instead follow-up time. Two clusters were identified evaluated cross validation. Kaplan Meier (KM) curves between two differ significantly OS RFS (p-value < 0.0001). Moreover, there relative predictive improvement when label in addition features compared only where AUC increased by 5.7% 13.0% RFS, respectively.