The field of electrophysiology and brain-computer interfaces is rapidly advancing with innovative approaches to address long-standing challenges. One notable direction is the development of models that can generalize across subjects and datasets, eliminating the need for subject-specific calibration data. This is being achieved through techniques such as feature disentanglement, multi-scale architecture, and instance-based transfer learning. Another significant trend is the creation of open-source evaluation frameworks and benchmarks, enabling fair comparison and driving progress in areas like sleep staging and speech detection from brain data. These advancements have the potential to improve the accuracy and reliability of brain-computer interfaces, enabling applications such as personalized cardiac monitoring and non-invasive communication restoration for individuals with speech deficits. Notable papers in this area include those proposing novel neural network architectures, such as CodeBrain, which offers biologically informed and interpretable EEG modeling, and the introduction of the LibriBrain dataset and pnpl library for speech decoding from non-invasive brain data. Additionally, the proposed Physiological-Model-Based Neural Network for heart rate estimation during daily physical activities demonstrates improved performance and interpretability.