[4C1-OS-27a] [Organized Session] OS-27

[4C1-OS-27a-03] Feature Extraction and Mood Estimation Using ECG Signals during Recalling Incidents from the Past

〇Akane Kitagawa¹, Shohei Kato^1,2 (1. Graduate School of Engineering, Nagoya Institute of Technology, 2. Frontier Research Institute for Information Science, Nagoya Institute of Technology)

Keywords:ECG Signals, Mood Estimation, Feature Extraction

The purpose of this study is to evaluate mood quantitatively using ECG signals during recalling incidents from the past for preventing depression and anxiety. For this purpose, we calculate indexes of the heart rate variability from ECG signals, and propose a method to classify seven kinds of mood using a combination of Random Forest (RF) and Support Vector Machine (SVM). The results show accuracy rate of more than 70% for any mood except mood of pleasantness. The results indicate the effectiveness of mood classification using ECG signals during recalling incidents from the past. In addition, we have discovered features to contribute mood changes by feature selection using RF.