A bootstrapping method40 was used to calculate the empirical distributions, and McNemar’s test41 was used to analyze the differences between the CNNCF and the experts.