Table 4.

Held-out test results using different methods

Methods	P	R	F	AUC
Filtered MetaMap +Pathology Report Count (4151)	0.74	0.84	0.79	0.87
Full MetaMap (17897)	0.66	0.34	0.45	0.80
Filtered MetaMap (4150)	0.71	0.78	0.74	0.84
Bag of Words (57612)	0.53	0.43	0.48	0.74

The number in the parenthesis in first column is the number of features

Gray shade indicates baseline methods

P stands for precision, R stands for recall, F stands for f score, AUC stands for area under the receiver operator characteristic curve