Multivariate Biosignals

Multivariate biosignals, encompassing multiple physiological measurements recorded simultaneously (e.g., EEG, EMG, ECG), are analyzed to extract meaningful information for diagnosis, prognosis, and treatment. Current research focuses on developing advanced machine learning models, including state-space models, diffusion models, graph neural networks, and optimal transport methods, to effectively handle the high dimensionality, temporal dependencies, and variability inherent in these data. These efforts aim to improve the accuracy and efficiency of applications ranging from dementia detection and sleep staging to gesture recognition and brain-computer interfaces, ultimately leading to more precise and personalized healthcare.