BACKGROUND: Daydreaming can be monitored either to avoid it while doing hands-on tasks or to enhance it to foster creativity. Although significant research has been conducted in Brain recordings and Machine learning, some problems have not received sufficient attention. One of them is the automated identification and classification of daydream states with emphasis on physiological signals and prandial states. Until now, researchers have been relying only on subjective questionnaire-based methods of daydream identification, neglecting neural hemodynamics. METHODOLOGY: In this study, EMG-based physiological triggers have been incorporated to detect self-induced daydream episodes in pre- and post-meal prandial states. For the AI-driven hemodynamic monitoring of the brain in relation to the analysis of the electrical activity of the stomach during self-induced daydreaming, fNIRS and EGG signals of 30 participants were recorded, preprocessed, and investigated simultaneously. Both the duration and frequency of the daydreaming episodes were analyzed using these two modalities, which were further subjected to a feature extraction and class label encoding process to facilitate a four-class classification of daydreaming and prandial state. RESULTS AND CONCLUSION: Machine learning models were incorporated for classification and resulted in the highest testing accuracy of 90.77% for daydream detection and gave insights into the connection between meal consumption and daydreaming. FUTURE WORK: In the future, this study could serve as the preliminary basis for multimodal monitoring systems used to assess the state of cognition in parallel with the analysis of meal intake patterns. This research can also lead to the development of person-specific treatments in the domain of mental and attentional health.
Ishtiaq et al. (Fri,) studied this question.