Related Experiment Video
Updated: May 9, 2026
05:36
STFEEG-Tool: A Spatial-Temporal-Frequency EEG Analysis Tool for Motor Imagery Brain-Computer Interfaces
Published on: March 10, 2026
CogniFuse and Multimodal Deformers: An Extended Study on Benchmarking and Modeling Biosignal Fusion
IEEE Transactions on Bio-Medical Engineering
|May 7, 2026
Summary
CogniFuse is a new benchmark for fusing multimodal biosignals from everyday activities. Our Multimodal Deformer models advance real-time health monitoring outside clinical settings.
Area of Science:
- Biomedical Engineering
- Signal Processing
- Machine Learning
Background:
- Human physiological signals offer insights into physical and mental states.
- Extracting this data during daily activities is challenging due to noise and artifacts.
Purpose of the Study:
- Introduce CogniFuse, a benchmark for multimodal biosignal fusion in unconstrained environments.
- Develop a robust benchmarking pipeline for comparability and reproducibility.
- Propose novel Multimodal Deformer models for biosignal analysis.
Main Methods:
- Developed CogniFuse, a multi-task benchmark for multimodal biosignal fusion.
- Created a comprehensive benchmarking pipeline emphasizing reproducibility and usability.
- Proposed Multimodal Deformer models to capture multi-level power features and temporal dependencies.
Main Results:
- The Multi-Channel Deformer model achieved the highest average benchmark score.
- Demonstrated robustness across various architectures, tasks, and model sizes.
- Successfully fused multimodal biosignals from unconstrained environments.
Conclusions:
- CogniFuse and the proposed models advance multimodal biosignal fusion for real-time monitoring.
- This work enables health monitoring outside controlled clinical conditions.
- All code and data are publicly available to ensure transparency and facilitate future research.