Real-Time Emotion Recognition Using Wearable Ubiquitous Devices for Enhanced User Interaction and Mental Health Monitoring
Dnyaneshwar Prabhakar BawaneAssistant Professor, Department of Applied Mathematics and Humanities Yeshwantrao Chavan College of Engineering, Nagpur, India. dnyanesh02@gmail.com0009-0009-2446-0240
Takveer SinghCentre of Research Impact and Outcome, Chitkara University, Rajpura, Punjab, India. takveer.singh.orp@chitkara.edu.in0009-0000-7255-2507
Debanjan GhoshAssistant Professor, Department of Computer Science & IT, ARKA JAIN University, Jamshedpur, Jharkhand, India. debanjan.g@arkajainuniversity.ac.in0000-0002-3255-6199
Dr. Satya Ranjan DasAssociate Professor, Department of Computer Science and Engineering, Siksha 'O' Anusandhan (Deemed to be University), Bhubaneswar, Odisha, India. satyadas@soa.ac.in0009-0009-5723-9665
I. KatharajAssistant Professor, Department of Mechanical Engineering, Faculty of Engineering and Technology, JAIN (Deemed-to-be University), Karnataka, India. kantharaj@jainuniversity.ac.in0000-0001-9497-0865
Rohit GoyalSchool of Engineering & Computing, Dev Bhoomi Uttarakhand University, Dehradun, India. socse.rohit@dbuu.ac.in0000-0001-7978-4079
Keywords: Wearable Devices, Ubiquitous Computing, Real-Time Emotion Recognition, Physiological Signals, HRV, EDA, Machine Learning, Deep Learning, Mental Health Monitoring, Iot-Based Systems, Affective Computing, User Interaction.
Abstract
Recognition of real-time emotions with the help of wearable ubiquitous devices is becoming increasingly popular as a data-driven solution to improve interaction between the user and monitor mental health around the clock. In this paper, I will introduce a lightweight multimodal model that interprets physiological measurements HRV, EDA, and skin temperature of low-power wearable devices with the help of sophisticated machine learning and deep learning algorithms. The statistical analysis indicates the high level of robustness of the system by scoring high classification accuracy as a whole and showing strong discriminative ability among different classes of emotions with low variance of prediction error. The validation of performance was done by cross-validation, confusion matrix analysis and significance testing to ensure the reliability of the model was consistent both in controlled environment and in the real world. The scalable deployment of the framework in the IoT ecosystems relies on its low-latency processing and statistically stable performance characteristics. The findings reinforce the possibility of wearable-based affective computing as a personalized digital health, adaptive interface, and as an early warning system of stress-related abnormal.