Skip to main content
. 2024 May 8;23:137. doi: 10.1186/s12944-024-02122-z

Fig. 4.

Fig. 4

Machine learning techniques employed to formulate diagnostic models for NASH. A The model built with the RF algorithm demonstrated the highest predictive accuracy, boasting a C-index value of 0.928. B-I The AUC values of the three model genes for the separate diagnosis of NASH in both the training set and the external validation set were relatively high