Table 5.

Comparison of four state-of-the-art machine learning algorithms on Chinese admission and discharge summaries when optimized features were used

Algorithm	Admission notes		Discharge summaries
	Exact-match	Inexact-match	Exact-match	Inexact-match
SVM	90.54 (90.81/90.27)	93.70 (93.99/93.42)	85.56 (85.89/85.21)	89.87 (90.23/89.52)
ME	90.43 (91.07/89.80)	93.49 (94.15/92.84)	85.15 (86.01/84.30)	89.70 (90.61/88.80)
CRF	93.52 (93.77/93.26)	94.69 (94.95/94.43)	89.23 (90.29/88.20)	91.00 (92.08/89.94)
SSVM	93.53 (92.93/94.15)	95.35 (94.72/95.97)	90.01 (89.19/90.84)	92.65 (91.91/93.51)

Values are F-measure (recall/precision) (%).

CRF, conditional random fields; ME, maximum entropy; SVM, support vector machines; SSVM, structural support vector machines.