Skip to main content
. 2021 May 20;4(2):109–118. doi: 10.1093/pcmedi/pbab013

Table 1.

Performance of the CNN1 model in different validation and test dataset.

Validation dataset Independent test dataset External validation dataset P-value
Precision 0.933 (0.928, 0.938) 0.898 (0.895, 0.901) 0.968 (0.966, 0.970) <0.001
Recall 0.934 (0.933, 0.939) 0.836 (0.832, 0.840) 0.965 (0.963, 0.967) <0.01
F1-score 0.933 (0.928, 0.938) 0.856 (0.853, 0.859) 0.965 (0.963, 0.967) 0.001
Accuracy 0.934 (0.929, 0.939) 0.836 (0.832, 0.840) 0.965 (0.963, 0.967) <0.01
Positive predictive value 0.869 (0.859, 0.880) 0.771 (0.761, 0.780) 0.923 (0.915, 0.930) <0.01
Negative predictive value 0.975 (0.971, 0.979) 0.956 (0.954, 0.958) 0.973 (0.971, 0.975) <0.001

95% confidence intervals are included in brackets.