Back to Evidence Hub
Case Study seismocardiography-for-emotion-recognition-a-study-on-emowear-with-insights-from-deap
2024 Release

Seismocardiography for Emotion Recognition: A Study on EmoWear with Insights from DEAP

Executive Summary

This study introduces seismocardiography (SCG) and accelerometry-derived respiration (ADR) as novel modalities for emotion recognition (ER) using wearable accelerometers. By leveraging the EmoWear dataset, the authors validated their methodology against the DEAP dataset and demonstrated that SCG achieves comparable performance to traditional cardiac signals like ECG and BVP for ER. Combining SCG with ADR enables a single chest-worn accelerometer to capture both cardiac and respiratory signals, paving the way for cost-effective, real-world ER applications.

This study shows that a single wearable accelerometer on the chest can track emotions by measuring heart and breathing vibrations, offering a simpler and cheaper way to integrate emotion recognition into daily life.

Answer Machine Insights

Q: Can SCG be effectively used for emotion recognition?

Yes, SCG achieved similar performance to ECG and BVP for emotion recognition.

Results show that SCG is a viable modality for ER, achieving similar performance to ECG and BVP.

Q: What is the significance of combining SCG with ADR?

Combining SCG with ADR enables emotion recognition using only a single chest-worn accelerometer.

These findings confirm the applicability of a single chest-worn accelerometer for ER, which is a significant step towards embedding ER in everyday life scenarios.

Key Results

  • SCG achieved comparable emotion recognition performance to ECG and BVP, with F1 scores of ~0.58 for valence and ~0.57 for arousal using SVM and LR classifiers.

  • Combining SCG with ADR demonstrated the feasibility of using a single chest-worn accelerometer for emotion recognition, achieving statistically significant results (p < 0.001).