Table 5.

Comparison to OS models trained on the 3-class problem (COVID-19 vs CP vs Control). Due to the difference in sample size/COVID-19 prevalence, in fact, models are not directly comparable

Model	COVID Sensitivity	Overall accuracy	COVID prevalence	#Test images
Ours (best)	90.80%	91.66%	20.00%	21191
ResNet50 [8]	85.90%	88.10%	46.84%	746
LightCNN [4]	88.23%	84.56%	25.39%	392
COVNet [2]	90.00%	89.04%	30.00%	434
ResNet18 [3]	81.30%	86.70%	35.79%	210
DarkCOVIDNet [33]	85.25%	87.02%	50.00%	1000
DreNet [17]	93.00%	87.00%	47.38%	57
WRE [11]	86.40%	86.12%	50.00%	29