Table 2.

Performance comparison of our model with publicly available qualitative models.

Dataset	Independent test dataseta
Model	Our model (LogBB_Pred)	admetSAR	LightBBB	SwissADME	BBB predictor
Accuracy (%)	0.85	0.85	0.70	0.70	0.70
MCC	0.6	0.65	0.42	0.29	0.15
Sensitivity	0.42	0.85	0.70	0.57	0.28
Specificity	0.99	0.66	0.74	0.75	0.85
NPVb	0.83	0.94	0.45	0.44	0.80
PPVc	1	0.66	0.89	0.22	0.77
URL	http://ssbio.cau.ac.kr/software/logbb_pred	http://lmmd.ecust.edu.cn/admetsar2	http://ssbio.cau.ac.kr/software/BBB	http://www.swissadme.ch/	https://www.cbligand.org/BBB/index.php
Reference	This study	Yang et al. (2019)	Shaker et al. (2021)	Daina et al. (2017)	Liu et al. (2014)

^aThe independent test dataset used for the comparison of quantitative models was also used for qualitative model evaluation.

^bNegative predictive value: (number of true negatives)/(number of true negatives + number of false negatives).

^cPositive predictive value: (number of true positives)/(number of true positives + number of false positives).