Fig. 2: Baseline model performances. | Schizophrenia

Fig. 2: Baseline model performances.

From: Classification of schizophrenia, bipolar disorder and major depressive disorder with comorbid traits and deep learning algorithms

Fig. 2

A SCZ; B BIP; and C MDD. The plots show the ROC curves for the baseline models that use only the bestPRSs from the targeted disorders (blue, Model B.I), bestPRSs from all traits (targeted disorders and comorbid traits, gold, Model B.II), and bestPRSs from only the comorbid traits (green, Model B.III). The results indicate that models with the inclusion of PRSs from comorbid traits have better performances than that of using only the PRSs from targeted traits. It is also clear that models using only the PRSs from comorbid traits can also have decent performance. D A comparison between the baseline elastic model (Model B.II.SCZ) and DNN model for SCZ classification (Model II.SCZ).

Back to article page