Fig. 2: Comparison of diagnostic performance of MAOSS, Unimodal models and TE-CAP on internal and external testsets for histology-proved hepatic steatosis. | Nature Communications

Fig. 2: Comparison of diagnostic performance of MAOSS, Unimodal models and TE-CAP on internal and external testsets for histology-proved hepatic steatosis.

From: Multi-modal AI for opportunistic screening, staging and progression risk stratification of steatotic liver disease

Fig. 2: Comparison of diagnostic performance of MAOSS, Unimodal models and TE-CAP on internal and external testsets for histology-proved hepatic steatosis.

Receiver operating characteristic (ROC) analysis of compared models for classification of dichotomized histologically determined hepatic steatosis grades: a on internal test dataset and c. on external test dataset; Individual classification metrics (BACC, F1, Sensitivity, Specificity) were calculated based on 1000 bootstrap replications of the test datasets and reported with mean and 95% CI for compared models: b on internal test dataset and d on external test dataset. BACC: balanced accuracy, F1: F1 score, CI: confidence interval, TE-CAP: transient elastography-controlled attenuation parameter. For more details, please check supplementary Tables S4--S10. Source data are provided as a Source Data file.

Back to article page