Skip to main content
. 2021 Apr 29;4:553987. doi: 10.3389/frai.2021.553987

Table 1.

Classifier performance vs. training set size.

Training set size (in images) Sensitivity Specificity Accuracy
1,000 Pneumonia/1,000 Normal 0.89 0.96 0.92
300 Pneumonia/300 Normal 0.97 0.81 0.89
200 Pneumonia/200 Normal 0.90 0.86 0.88
100 Pneumonia/100 Normal 0.90 0.88 0.89
50 Pneumonia/50 Normal 0.98 0.80 0.89