Peripheral thermoregulation patterns of stress and relaxation assessed via infrared thermography and linear...

In recent years, the automatic identification of stress and relaxation has gained attention due to the substantial impact these states have on physical and mental well-being. While incorporating physiological responses that accurately reflect emotional behavior improves the efficiency of classification models, identifying this relationship is challenging. This often leads to the use of black-box approaches that complicate physiological interpretation. This study aims to characterize the thermal behavior of fingertips in female and male participants during psychological stress and relaxation using statistical features, and to classify these states based on the most significant features. The study also evaluates virtual reality as a relaxation tool by comparing it with mindfulness-based relaxation. To this end, a virtual reality environment was developed based on scientific recommendations to support relaxation processes. The results show that automatic feature selection through linear discriminant analysis can highlight thermographic correlates of peripheral thermoregulation associated with emotional states. Findings indicate that a reduced subset of highly relevant features is sufficient for high performance, achieving high classification accuracy with the 45, 19, and 55 most relevant features for the female, male, and global datasets, respectively. However, these results should be interpreted cautiously due to the limited sample size and the exploratory nature of the study. Additionally, the study demonstrates that virtual reality is associated with relaxation-like thermal responses, supporting its potential as a complementary tool in psychological interventions. Overall, the results may contribute to improving classification performance, furthering our understanding of psychophysiological mechanisms and aiding in the development of effective systems for identifying emotional states automatically.