Table 2.

AUROCs (95% CIs) of nine possible combinations of finding-extraction method and BN classifier

BN classifiers	Finding-extraction methods
	Expert	Topaz	MedLEE
Primary analyses: BN classifiers using 31 influenza-related findings
Expert-defined BN	0.70 (0.63 to 0.77)	0.64 (0.57 to 0.71)	0.64 (0.57 to 0.72)
BN-EM-Topaz	0.79 (0.73 to 0.85)	0.73 (0.66 to 0.79)	0.71 (0.64 to 0.78)
BN-EM-MedLEE	0.77 (0.70 to 0.83)	0.70 (0.63 to 0.77)	0.66 (0.59 to 0.74)
Secondary analyses: BN classifiers using 17 influential findings*
Expert-defined BN	0.80 (0.74 to 0.86)	0.76 (0.69 to 0.82)	0.73 (0.66 to 0.80)
BN-EM-Topaz	0.82 (0.76 to 0.88)	0.75 (0.69 to 0.82)	0.74 (0.68 to 0.81)
BN-EM-MedLEE	0.79 (0.73 to 0.85)	0.73 (0.66 to 0.80)	0.70 (0.63 to 0.77)

*The 17 influential findings indicated in BN-EM-Topaz were arthralgia, cervical lymphadenopathy, chill, cough, fever, hoarseness, influenza-like illness, lab confirmed influenza lab order (nasal swab), malaise, myalgia, rhinorrhea, sore throat, suspected flu, viral infection, viral syndrome, and wheezing.

AUROC, area under the receiver operating characteristic; BN, Bayesian network.