EEG Emotion Grading Using AI and Human-Centric Lighting

The patent creates a mapping relationship between fMRI BOLD responses and EEG brainwave patterns, allowing EEG to serve as a simplified copy or proxy for fMRI measurements. By training a machine learning model on fMRI data and applying it to EEG data, the system replicates fMRI-level emotional detection accuracy using the simpler, more affordable EEG modality.

The patent replaces the complex mechanical and computational infrastructure of fMRI systems with a much simpler EEG-based system. By substituting the expensive, bulky fMRI machinery with portable EEG devices and software-based processing, the system achieves comparable functional outcomes with dramatically reduced complexity and cost.

The patent performs preliminary training using fMRI data to establish a reference mapping between neural activity patterns and emotional states. This pre-trained model captures the relationship between brain activity and emotions before applying EEG measurements, allowing the system to account for individual variations through the learned mapping rather than requiring subject-specific calibration.

The patent introduces a machine learning model as an intermediary that translates EEG brainwave patterns into emotional state predictions. This intermediary layer learns the complex relationship between neural activity and emotions from training data, enabling accurate emotional detection from EEG signals despite the indirect nature of the measurement and individual physiological variations.