Table 3.

The performance of various models for segregating ATB from LTBI in validation cohort

Parameters	Validation set (n = 263, 125 ATB, 138 LTBI)
	cforest	bart	gamboost	gbm	glmnet	lda	log_reg	svm
AUC (95% CI)	0.963 (0.940–0.986)	0.956 (0.932–0.981)	0.947 (0.919–0.975)	0.958 (0.935–0.981)	0.913 (0.876–0.950)	0.884 (0.841–0.927)	0.910 (0.872–0.949)	0.929 (0.896–0.962)
Sensitivity (95% CI)	92.80% (86.88–96.17%)	85.60% (78.38–90.69%)	82.40% (74.79–88.08%)	89.60% (83.02–93.82%)	78.40% (70.40–84.71%)	69.60% (61.05–76.98%)	80.80% (73.02–86.74%)	82.40% (74.79–88.08%)
Specificity (95% CI)	89.86% (83.69–93.86%)	92.03% (86.29–95.49%)	92.03% (86.29–95.49%)	89.86% (83.69–93.86%)	93.48% (88.07–96.53%)	94.93% (89.90–97.52%)	92.75% (87.18–96.02%)	93.48% (88.07–96.53%)
PPV (95% CI)	89.23% (82.73–93.48%)	90.68% (84.08–94.72%)	90.35% (83.55–94.53%)	88.89% (82.21–93.27%)	91.59% (84.78–95.51%)	92.55% (85.42–96.35%)	90.99% (84.21–95.03%)	91.96% (85.43–95.72%)
NPV (95% CI)	93.23% (87.64–96.40%)	87.59% (81.23–92.00%)	85.23% (78.66–90.04%)	90.51% (84.44–94.37%)	82.69% (75.99–87.82%)	77.51% (70.65–83.16%)	84.21% (77.58–89.15%)	85.43% (78.93–90.18%)
PLR (95% CI)	9.15 (5.55–15.07)	10.74 (6.06–19.02)	10.34 (5.83–18.33)	8.83 (5.36–14.56)	12.02 (6.35–22.76)	13.72 (6.61–28.50)	11.15 (6.10–20.38)	12.63 (6.68–23.89)
NLR (95% CI)	0.08 (0.04–0.15)	0.16 (0.10–0.24)	0.19 (0.13–0.28)	0.12 (0.07–0.19)	0.23 (0.16–0.32)	0.32 (0.24–0.42)	0.21 (0.14–0.30)	0.19 (0.13–0.28)
Accuracy (95% CI)	91.25% (87.22–94.10%)	88.97% (84.61–92.21%)	87.45% (82.90–90.92%)	89.73% (85.48–92.85%)	86.31% (81.63–89.95%)	82.89% (77.87–86.96%)	87.07% (82.48–90.60%)	88.21% (83.76–91.57%)

ATB: active tuberculosis; LTBI: latent tuberculosis infection; AUC: area under the ROC curve; PPV: positive predictive value; NPV: negative predictive value; PLR: positive likelihood ratio; NLR: negative likelihood ratio; CI: confidence interval