Emotion Detection for Children on the Autism Spectrum using BCI and Web Technology

• Published in: 2022 IEEE/WIC/ACM International Joint Conference on Web Intelligence and Intelligent Agent Technology (WI-IAT) pp. 207 - 214
• Main Authors: Zaman, Akib, Tahsin, Anika, Rahman, Mostafizur, Akhter, Rabeya, Rahman, Hinoy, Mustary, Shobnom, Farid, Dewan Md
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.11.2022
• Subjects: Autism; Autism Spectrum Disorder (ASD); Brain modeling; Brain-Computer Interface (BCI); Brain-computer interfaces; Brainwave Data; Emotion recognition; Ensemble Learning; Feature extraction; Predictive models; Source coding; Web Application
• DOI: 10.1109/WI-IAT55865.2022.00037

Brain-computer interface (BCI) technology is getting popular to detect emotions of Autism Spectrum Disorder (ASD) affected children nowadays. Understanding the emotional state of the ASD affected children is very challenging due to their inconsistent behaviour. In this paper, we have introduced a location-guided web-based system to monitor and detect emotions of ASD-affected children employing BCI technology. We have collected raw brainwave data of 30 subjects during three instances: positively excited, neutral, and negatively excited. We have created a 30-sec interval sample to generate the dataset. We have extracted features from the data and applied data pre-processing techniques to find the informative features and remove the outliers. Then, we have built an ensemble model that achieved 93% F1-score and uses majority weighted voting to predict the emotional state of the ASD affected children. Finally, we have deployed the ensemble model into a web application so that guardians can find the emotional state of their child. The brainwave data and location of the child are collected by the headgear and uploaded to the cloud. We have integrated a headgear that can be easily worn by the child. The main objective of this work is to design and develop a location-guided web application so that guardians can access the emotional state along with the location of the child using the web. Our source code including feature extraction, model development and evaluation is available at https://github.com/akibzaman/ASD-children-emotion-prediction.