Fig. 2From: Data-centric artificial intelligence in oncology: a systematic review assessing data quality in machine learning models for head and neck cancerGeneral characteristics and risk of bias rating for ML studies employing structured datasets. a Publication year trend. b Plot showing the number of patient datasets by the different head and neck cancer subtypes. c Plot showing the clinical utility of the models according to the different cancer subtypes. d Risk of bias rating for individual domains of the PROBAST toolBack to article page