FedDep-GT-LLM: Federated Graph Transformer and Clinical Language Modeling for Multimodal Depression Analysis
Abstract
Abstract Early identification of depression severity remains a major challenge in precision mental healthcare because of the hetero-geneity of behavioral, linguistic, physiological, and neuro cognitive indicators. Existing artificial intelligence based depressionassessment methods are often limited by inadequate privacy preservation, interpretability, poor cross-domain generalization,and insufficient multimodal fusion. To address these limitations, we propose an explainable federated multimodal learningframework that integrates Graph Transformers(GTs), large language models (LLMs), and adaptive cross-modal attention forearly depression severity prediction. The proposed framework, FedDep-GT-LLM, leverages decentralized federated optimizationto preserve patient privacy while enabling collaborative learning from distributed multimodal datasets across institutions. Itfuses speech, facial expressions, textual transcripts, and EEG signals depending on dataset availability, using modality-specificencoders, followed by a heterogeneous graph transformer for relational representation learning. A domain-adapted clinicalLLM extracts contextual semantic embeddings from patient conversations and self-reported narratives, whereas a SHAPguided explainability module enhances clinical interpretability. We evaluate the framework on DAIC-WOZ, E-DAIC, AVEC2019, and MODMA datasets under cross-dataset and federated settings. FedDep-GT-LLM achieved an average accuracy of94.8%, sensitivity of 94.1%,…
Read the paper
Related studies
- Active Learning during Federated Learning for Object Detection
- Fairness-aware federated learning
- Decentralized Federated Learning
- Building a Personally Identifiable Information Recognizer in a Privacy Preserved Manner Using Automated Annotation and Federated Learning
- Preservation of Patient Level Privacy: Federated Classification and Calibration Models
- Efficient explainable federated learning
Go deeper in Evidence Hub
Evidence Hub adds a study data sheet (design, sample size, evidence level, key numbers), an AI summary and a chat grounded in the paper's text.