A Network Physiology Approach to Brain-Heart Interaction for Affective State Characterization Using Photoplethysmography Features

Emerging evidence highlights bidirectional, complex, and nonlinear communication between heart and brain dynamics.While several studies have applied artificial intelligence to distinguish and classify emotions based on static EEG and PPG features, little research has focused on the network elements of these interactions across different emotional states.Using data from the DEAP dataset, which includes EEG and PPG signals recorded while participants viewed emotionally evocative music videos, this study applied a novel network analytic approach to examine dynamic interactions between brain rhythms and PPG features (amplitude, peak-to-peak interval, and pulse width amplitude).Time-delay stability of partial cross-correlation was used to identify emotional states.Significant correlations emerged between EEG rhythms and both PWA and PPI across affective states (p < 0.05).However, reverse relationships, where PPI or PWA influenced EEG bands, were not significant.Furthermore, the correlation between PPG amplitude and EEG rhythms, or vice versa, did not significantly differentiate between affective states, suggesting that PPG amplitude is less indicative of emotional states than PPI or PWA.The findings establish EEG-PWA and EEG-PPI connectivity as reliable indicators of affective states and provide insights for developing explainable graph-based emotion recognition systems.