Skip to main content
. 2022 Aug 17;13:137. doi: 10.1186/s13244-022-01276-7

Table 2.

Average mean absolute error (MAE) and rank correlation of the baseline CatBoost and naïve models, which only utilize clinical variables to predict segmentation quality. The naïve method predicts the mean target value for all samples. Parentheses indicate standard deviation. The values are aggregated from 64 repeated fivefold cross-validations

MAE Corr
CatBoost 0.016 (± 3·10−4) − 0.16 (± 0.02)
Naïve 0.016 (± 0) n/a