Copyright © 2026 Authors retain the copyright of this article. This article is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

@article{207580,
        author = {Ashly P Eldho and Nishanth R},
        title = {BEAM: Behaviour and Emotion-Aware Multimodal Framework for Digital Wellness},
        journal = {International Journal of Innovative Research in Technology},
        year = {2026},
        volume = {13},
        number = {3},
        pages = {2167-2179},
        issn = {2349-6002},
        url = {https://ijirt.org/article?manuscript=207580},
        abstract = {Excessive smartphone usage has become a growing concern due to its impact on mental wellbeing, sleep quality, productivity, and daily life. Existing digital wellbeing applications primarily focus on screen-time monitoring and often pro-vide limited support for personalised intervention and long-term behavioural assessment. To address these limitations, this paper presents BEAM, a multimodal and agentic AI framework for personalised digital wellness. The proposed framework com-bines behavioural analytics and facial emotion recognition to as-sess user wellbeing. Behavioural indicators, including screen time, phone unlock frequency, social media usage, sleep dura-tion, stress level, anxiety score, physical activity, and family in-teraction, are analysed using a Logistic Regression model to pre-dict addiction levels. Emotional context is obtained through an ensemble emotion recognition framework integrating a Convolutional Neural Network (CNN) and DeepFace. The system fur-ther incorporates Retrieval-Augmented Generation (RAG), persistent memory, and a LangGraph-based Agentic AI layer to generate personalised recommendations, analyse historical behavioural patterns, and provide contextual wellness insights. Experimental evaluation demonstrates that the behavioural prediction model achieves an accuracy of 91%, while the emo-tion recognition framework achieved approximately 77% grouped emotion classification accuracy and provides meaning-ful emotional context for recommendation generation. The integration of memory, knowledge retrieval, and autonomous reasoning enables personalised and longitudinal wellbeing assessment. The results indicate that BEAM provides a more comprehensive approach to digital wellness management than conventional screen-time monitoring systems and demonstrates the potential of multimodal and agentic AI techniques for personalised wellbeing support.},
        keywords = {Digital Wellness, Agentic AI, Multimodal Learning, Emotion Recognition, Behavioural Analytics, Retrieval-Augmented Generation.},
        month = {August},
        }