Large Models Driven Behavior Modeling to Enhance Effect Treatment of Arts in Psychological Consumer Electronics

The integration of psychological art therapy with consumer electronics has opened new avenues for improving mental health care. This study proposes a multi-feature user behavior modeling framework designed to enhance the effectiveness of art therapy delivered through consumer electronics by leveraging advanced artificial intelligence techniques. We utilize multi-modal data, including physiological signals such as heart rate (HR), skin conductance (SC), and electroencephalogram (EEG), to perform real-time emotion recognition. Artificial Intelligence (AI) has entered a new era of Large Models (LMs). With billions or trillions of parameters trained on vast amounts of corpus, LMs have achieved unprecedented successes in many challenging applications. In this paper, our approach realize the LMs-based deployment via employing a multi-scale convolutional neural network for extracting temporal and spatial features, followed by a hybrid fusion method combining early and late fusion strategies. The DEAP dataset is used to evaluate the proposed model, demonstrating significant improvements over state-of-the-art methods in terms of accuracy, precision, recall, and F1 score. The ablation study further validates the importance of multi-modal features and the effectiveness of the hybrid fusion approach. The results underscore the potential of our model in enhancing the delivery and efficacy of art therapy through advanced consumer electronics.