Exploring the Clothing Feedback Mechanism in the Interaction Between Medical Students’ Emotional States and Humanistic Education Using a Support Vector Machine Model

This study presents an analytical framework combining weighted multi-kernel Support Vector Machine (SVM) and multimodal feature fusion to quantify the impact of clothing on medical students’ emotional states in humanistic education contexts. From the perspective of textile psychology and sensory engineering, visual clothing attributes (HSV color space, LBP texture, and deep style features from ResNet-50) are integrated with physiological signals (ECG, galvanic skin response) to construct a high-dimensional multimodal feature vector. Principal Component Analysis (PCA) and Sequential Forward Selection (SFS) are applied for dimensionality reduction and optimal feature selection, while an improved Grey Wolf Optimizer (I-GWO) automatically tunes the weighted multi-kernel SVM hyperparameters. Experimental results demonstrate that the framework achieves an emotion recognition accuracy of 89.7%, with color brightness identified as the most critical driver of positive emotions and empathy. Analysis across different teaching scenarios—doctor-patient communication, ethical debate, and narrative reflection—reveals that context moderates the clothing-emotion relationship, providing precise quantitative evidence for optimizing educational strategies. The proposed approach is suitable for integration with antenna-enabled smart clothing systems, wireless sensing networks, and edge-computing platforms, enabling real-time monitoring of student emotional states and data-driven enhancement of medical humanities teaching environments.