Unveiling the influence of facial expressions on EEG-based biometric system performance in ADHD and healthy children

Abstract

The growing need for reliable biometric systems to identify children in contexts such as vaccination tracking, missing child recovery, and hospital safety has highlighted the importance of using robust physiological markers. EEG signals have emerged as promising biometric indicators due to their resistance to external manipulation and forgery. In this study, we investigated the influence of emotional facial expressions on EEG-based biometric identification in children with and without Attention-Deficit/Hyperactivity Disorder (ADHD). EEG data were recorded from 25 typically developing (TD) children and 22 children diagnosed with ADHD during exposure to four emotional facial expressions: happy, sad, angry, and neutral. To evaluate the robustness of biometric identification across emotional states, brain connectivity features were extracted using various directed connectivity metrics (Directional Transfer Function (DTF), ffDTF, dDTF, dDTF08, Partial directional coherence (PDC)) and analyzed through clustering techniques based on Riemannian and Euclidean distances. The DTF feature combined with Riemannian distance-based clustering achieved the highest identification accuracies across all emotional states, reaching up to 100% for both groups. Specifically, accuracies of 99%, 99.4%, 99.6%, and 100% for healthy children and 100%, 99.77%, 99.77%, and 100% for ADHD children were obtained for sad, happy, angry, and neutral emotions, respectively. Statistical analysis confirmed the emotional resilience of the biometric system, showing no significant differences in identification accuracy between emotional states or between ADHD and TD groups. These findings support the feasibility of emotion-insensitive EEG-based biometric systems for child identification and highlight the utility of brain connectivity features in enhancing performance across neurodevelopmental populations.