Key Technologies of Emotion Visualization and Art Therapy: Design and Validation of the ArtThera System
Wenhao Dai *
Chongqing Polytechnic University of Electronic Technology, Chongqing, China.
Chuandi Tang
Chongqing Polytechnic University of Electronic Technology, Chongqing, China.
Lühan Wu
Chongqing Polytechnic University of Electronic Technology, Chongqing, China.
Liping Chen
Chongqing Polytechnic University of Electronic Technology, Chongqing, China.
Xianfeng Liu
Chongqing Polytechnic University of Electronic Technology, Chongqing, China.
Yiman Liang
Chongqing Polytechnic University of Electronic Technology, Chongqing, China.
*Author to whom correspondence should be addressed.
Abstract
Aims: This study presents the design and pilot evaluation of ArtThera, a mobile system that integrates emotion visualisation, AI-generated art content, and multimodal emotional assessment for use among college students.
Study Design: A cross-platform mobile application was developed with four main modules: emotion-centric visual capture, a colour-emotion mapping engine, an AI-generated art engine, and a multimodal assessment module.
Methodology: The emotion capture module supports image-based and text-based emotional input. The colour-emotion mapping engine applies K-means clustering to extract dominant colour features from a curated collection of more than 1,000 classical artworks and links these features to the Pleasure-Arousal-Dominance (PAD) emotional model through statistical association analysis. The AI-generated art engine produces personalised colour palettes, minimalist drawing templates, and contextual audio content. The assessment module combines emotion diary records, heart rate variability (HRV) indicators, the Self-Rating Anxiety Scale (SAS), and the Positive and Negative Affect Schedule (PANAS). A four-week pilot evaluation was conducted with 20 undergraduate volunteers recruited through a university psychological association. The study protocol received institutional ethics approval prior to data collection.
Results: Participants completed an average of 3.4 sessions per week. The system demonstrated stable operation on Android and iOS platforms, with an average cold-start time of 1.3 seconds and an average content generation response time of 1.7 seconds. Pre-session to post-session comparisons showed a mean RMSSD increase of 12.8% (38.2 ms to 43.1 ms). Weekly HRV monitoring indicated a longitudinal trend of increasing RMSSD from baseline (Week 0: 28.6 ms) through Week 4 (44.1 ms). SAS scores decreased from a baseline mean of 48.6 to 41.3 at the four-week endpoint, while PANAS positive affect scores increased and negative affect scores decreased following individual sessions.
Conclusion: These preliminary findings suggest that ArtThera is technically feasible and acceptable for campus-based emotional support. However, given the absence of a control group, the small sample size (N = 20), and the lack of randomisation and formal statistical testing, the therapeutic outcome indicators must be interpreted with caution. No causal conclusions regarding clinical effectiveness can be drawn from this pilot study.
Keywords: Emotion visualisation, art therapy, AI-generated art, colour-emotion mapping, K-means clustering, pleasure-arousal-dominance model, heart rate variability, multimodal assessment, mobile mental health, college students