Experimental Validation of a Federated Explainable Multi-Modal Transformer–GAN Framework with AutoML-Optimized Ensemble Learning for Bias-Resilient, Privacy-Preserving and Resource-Efficient Real-Time Disease Prediction Across Heterogeneous Multi-Site Populations
Abstract
Abstract In recent years, the demand for real-time disease prediction has increased significantly, especially in environments where patient data are distributed across multiple healthcare institutions. Traditional centralized learning approaches often suffer from issues such as data bias, lack of transparency, inconsistent performance, and high computational requirements. These challenges reduce their effectiveness in practical healthcare scenarios. To overcome these limitations, this study presents a federated multi-modal framework designed for accurate and privacy-preserving disease prediction. The framework utilizes Vision Transformers to capture complex patterns from diverse medical data, while Generative Adversarial Networks are employed to improve data diversity and address class imbalance. Federated Learning allows multiple institutions to collaboratively train models without sharing sensitive patient information, ensuring data privacy and security. Additionally, an ensemble strategy integrates multiple models, including Convolutional Neural Networks, Random Forest, Support Vector Machine, and Bidirectional Long Short-Term Memory, to enhance prediction reliability. Interpretability methods are incorporated to provide clear insights into model decisions for healthcare professionals. Experimental results demonstrate improved prediction accuracy, stable performance across institutions, and reduced computational and communication costs.
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.