Authors:
Vincenzo Ronca
1
;
2
;
Gianluca Di Flumeri
2
;
3
;
Andrea Giorgi
2
;
Alessia Vozzi
1
;
2
;
Pietro Aricò
2
;
3
;
Nicolina Sciaraffa
2
;
3
;
Luca Tamborra
1
;
2
;
4
;
Ilaria Simonetti
1
;
2
;
4
;
Antonello Di Florio
2
;
Fabio Babiloni
2
;
3
;
5
and
Gianluca Borghini
2
;
3
Affiliations:
1
Department of Anatomical, Histological, Forensic and Orthopaedic Sciences, Sapienza University, Rome 00185, Italy
;
2
BrainSigns srl, Rome 00185, Italy
;
3
Department of Molecular Medicine, Sapienza University of Rome, Rome 00185, Italy
;
4
People Advisory Services Department, Ernst & Young, Rome 00187, Italy
;
5
Department of Computer Science, Hangzhou Dianzi University, Hangzhou, China
Keyword(s):
Facial Video, Neurophysiological Assessment, Signal Processing, Heart Rate, Electrodermal Activity, Emotional State Evaluation.
Abstract:
Human emotions decoding and assessment is a hot research topic since its implications would be relevant in a huge set of clinical and social applications. Current emotion recognition and evaluation approaches are usually based on interactions between a “patient” and a “specialist”. However, this methodology is intrinsically affected by subjective biases and lack of objectiveness. Recent advancements in neuroscience enable the use of traditional biosensors and maybe commercial wearable devices, which lead to a certain grade of invasiveness for the subject. The proposed study explored an innovative low-invasive hybrid method, based on the use of video data and smart bracelet, to overcome such technological limitations. In particular, we investigated the capability of an Emotional Index (EI), computed by combining the Heart Rate (HR) and the Skin Conductance Level (SCL) estimated through video-based and wearable technology, in discriminating Positive and Negative emotional state during
interactive webcalls. The results revealed that the computed EI significantly increased during the Positive condition compared to the Negative one (p = 0.0008) and the Baseline (p = 0.003). Such evidences were confirmed by the subjective data and the classification performance parameters. In this regard, the EI discriminated between two emotional states with an accuracy of 79.4%.
(More)