EMPIRICAL STUDY ON MULTIMODAL EMOTION RECOGNITION TECHNOLOGY TO IMPROVE ATHLETES' TRAINING EFFECTIVENESS

Abstract

Objective: Emotional states of athletes have significant qualitative effects on their performance and training outcomes. This paper aims at investigating whether these multimodal emotion recognition technologies can allow providing delivery of emotional feedback in real-time which can be subsequently utilized to modulate the training interventions in a manner that leads to improving the effectiveness of the training. Methods: This study was an RCT with 150 athletes from the China Southwest Medical University training base, recruited via stratified random sampling to achieve a balanced distribution in terms of gender and level of experience. Based on a 1:1 ratio, participants from four sports (swimming, soccer, sprinting, and basketball) were randomly divided into the Experimental Group (EG) and Control Group (CG). The experimental group was provided with real-time emotional feedback using a multimodal emotion recognition system that monitored facial expressions, physiological signals, and voice. Analysis Sharing: Analyses were performed using R 4.0.3 to examine the effect of the intervention on primary and secondary outcomes. Results: The experimental group showed statistically significant pre to post improvement with Wilcoxon signed-rank tests on all primary outcomes. For Endurance, the effect size was small, with a Cohen’s d of 0.37 (95% CI [0.14, 0.59], p < 0.001), indicating a significant improvement. Skill showed a medium effect size (Cohen’s d = -0.646, 95% CI [-0.88, -0.41], p < 0.001), suggesting better performance in the experimental group. For PANAS Positive, a large effect size (Cohen’s d = -0.924, 95% CI [-1.16, -0.69], p < 0.001) was observed, reflecting improved emotional regulation. Lastly, PANAS Negative showed a large effect size (Cohen’s d = 0.959, 95% CI [0.72, 1.20], p < 0.001), indicating significant reduction in negative emotional states in the experimental group. Conclusion: These findings underscore the potential of emotion recognition technology to support athletes in managing their emotional and physiological responses during training.