“Multi-loss Disentangled Generative-Discriminative Learning Model Improves Schizophrenia Data Analysis and Biomarker Identification”

https://pubmed.ncbi.nlm.nih.gov/38117620

The Multi-loss Disentangled Generative-Discriminative Learning (MDGDL) model, which represents and separates schizophrenia data into modality-specific and modality-common vectors, improves multimodal data analysis by achieving superior classification and regression task performance and identifying key multimodal biomarkers.