Wearable art therapy as a novel multisensory intervention for alleviating depressive symptoms

Abstract Depression is a main global cause of emotional disability, impacting cognitive function, social interactions, and quality of life. Traditional assessment methods depend on subjective, time-consuming self-reported surveys and clinical interviews, and challenging for continuous monitoring and intervention. To address these limitations, this research-proposed system integrates wearable sensing technologies, multisensory art therapy, and deep learning (DL) techniques to provide personalized and continuous mental health support. The Wearable Art Therapy Depression Dataset with 11,700 records collected using smart wearable devices and smartphone sensors, including heart rate variability (HRV), electrodermal activity (EDA), sleep behavior, physical activity, motion patterns, and mobile usage statistics. The collected raw sensory data undergo preprocessing steps such as Kalman filtering and Min-Max normalization to generate clinically relevant emotional indicators. Discrete Wavelet Transform (DWT) for time-frequency emotional feature extraction. A DL model based on Emperor Penguin Colony mutated Vanilla Long Short-Term Memory (EPC-Vanilla LSTM) models then employed to predict depressive indications. An EPC optimization algorithm is employed for adaptive hyperparameter tuning, while the Vanilla LSTM model is utilized to predict depressive symptom severity and emotional fluctuations. Based on the predicted emotional state, the proposed wearable therapeutic system delivers adaptive multisensory interventions through dynamic visual art patterns, calming audio feedback, tactile vibration stimulation, and color-changing smart textiles. Experimental evaluation implemented in Python demonstrated significant symptom reduction, achieving a recall of 96%. The obtained results confirm the effectiveness of the proposed wearable multisensory art therapy framework for personalized depression management and continuous emotional well-being support.