Table 3.

The performance of baseline system with different combinations of training and test data (micro-averaged, at strict entity level). 2014: i2b2/UTHealth Challenge[14], 2016: CEGS N-GRID Challenge [15]

Training data	Test data	Precision	Recall	F-measure
2014 training	2014 test	94.95	89.17	91.97
2014 training + test	2016 training	85.54	65.98	74.50
2014 training + test	2016 test	83.94	66.57	74.25
2016 training	2016 test	92.34	83.23	87.55