Inter-labeler and intra-labeler variability of condition severity classification models using active and passive learning methods
作者：
Highlights：
• AL methods produce smoother Intra-labeler learning curves during the training phase.
• AL methods result in almost half of the mean Inter-labeler AUC standard deviation.
• The consensus label resulted in an AUC that was at least as high as that of the gold standard label.
• The consensus label resulted in an AUC that was higher than the mean AUC of any random labeler.
• AL methods reduce Inter-labeler performance variance, and the dependence on a particular labeler.
摘要
•AL methods produce smoother Intra-labeler learning curves during the training phase.•AL methods result in almost half of the mean Inter-labeler AUC standard deviation.•The consensus label resulted in an AUC that was at least as high as that of the gold standard label.•The consensus label resulted in an AUC that was higher than the mean AUC of any random labeler.•AL methods reduce Inter-labeler performance variance, and the dependence on a particular labeler.
论文关键词：Active learning,Electronic health records,Phenotyping,Condition,Severity,Variance,Labeling
论文评审过程：Received 1 March 2017, Accepted 3 March 2017, Available online 27 April 2017, Version of Record 6 October 2017.
论文官网地址：https://doi.org/10.1016/j.artmed.2017.03.003
