**6. Conclusions**

An in-ear EEG device was developed. Earphone rubber was used as the in-ear EEG device main body. Silver-adhesive fabric was used as an in-ear EEG electrode. The in-ear EEG signals were verified to be close to T7 and T8 on MMN ERP responses, with a correlation of approximately 0.8530. The emotion classification results were approximately 71.07% for valence, 72.89% for arousal, and 53.72% for four emotions, compared to those of the DEAP emotion classification results using T7 and T8, which were about 69.85 % for valence, and 78.7 % for arousal, while the accuracy for classifying four simplified emotions was about 58.12%. Classification accuracies between in-ear EEG, and T7 and T8 electrodes, are not statistically significant. These results together with its earphone-like wearability, sugges<sup>t</sup> its potential for novel healthcare applications, such as home-based or tele-monitoring systems.

**Author Contributions:** Conceptualization, P.I. and S.P.; methodology, S.P. and C.A.; validation, S.P. and C.A.; resources, P.I.; data curation, C.A.; writing—original draft preparation, S.P. and C.A.; writing—reviewing and editing, P.I. and S.P.; supervision, P.I. and S.P.

**Funding:** This study was funded by Department of Computer Engineering, Chulalongkorn University, via a Graduate Scholarship for Alumni.

**Conflicts of Interest:** The Authors declare that there is no conflict of interest.
