Authors : Annappa S. S.; Asha S.; Lohith C.

Volume/Issue : Volume 11 - 2026, Issue 7 - July

Google Scholar : https://tinyurl.com/y5jbwy86

Scribd : https://tinyurl.com/yevj6dv5

DOI : https://doi.org/10.38124/ijisrt/26jul1281

Note : A published paper may take 4-5 working days from the publication date to appear in PlumX Metrics, Semantic Scholar, and ResearchGate.

Abstract : Population ageing is increasing demand for continuous, low-burden monitoring of physical and emotional wellbeing in the home, yet most smart-mirror prototypes to date are single-user devices tuned to one fixed idea of a healthy expression or posture. This paper proposes an IoT-enabled Smart Mirror architecture built around three ideas that, to the authors' knowledge, have not previously been combined in a single mirror-based system: (i) identity-aware sensing that disambiguates several household or care-home residents sharing one device using fused facial and gait signatures; (ii) a peruser adaptive baseline that learns each resident's own resting facial-expression and posture pattern over time and flags deviations from that individual baseline rather than from a fixed population norm; and (iii) a large-language-model (LLM) coaching layer that turns the fused emotional-physical state into short, empathetic, context-aware spoken or on-screen guidance, with an explicit safety-guardrail stage before anything reaches the user or a caregiver. An edge-cloud split keeps the sensing-to-deviation loop local for low latency while the LLM coaching and longitudinal analytics run in the cloud. Because no hardware prototype has yet been built, this paper deliberately stops short of reporting experimental accuracy figures; instead it develops the mathematical formulation of the identity-aware fusion and adaptive-baseline mechanism and presents a small, clearly labelled synthetic simulation to illustrate how the deviation signal and the edge-cloud latency budget would behave. The paper is intended as a design and feasibility contribution that can guide a subsequent hardware implementation and user study.

Keywords : Smart Mirror, Multi-User Recognition, Identity-Aware Sensing, Adaptive Baseline, Large Language Models, Health Coaching, Elderly Care, Emotion Recognition, Posture Recognition, Internet of Things, Edge-Cloud Computing.

References :

1. P. Silapasuphakornwong and K. Uehira, “Smart Mirror for Elderly Emotion Monitoring,” 2021 IEEE 3rd Global Conf. on Life Sciences and Technologies (LifeTech), Nara, Japan, 2021, pp. 356–359, doi: 10.1109/LifeTech52111.2021.9391829.
2. S. Bianco, L. Celona, G. Ciocca, D. Marelli, P. Napoletano, S. Yu, and R. Schettini, “A Smart Mirror for Emotion Monitoring in Home Environments,” Sensors, vol. 21, no. 22, p. 7453, 2021, doi: 10.3390/s21227453.
3. L. Rachakonda, P. Rajkumar, S. P. Mohanty, and E. Kougianos, “iMirror: A Smart Mirror for Stress Detection in the IoMT Framework for Advancements in Smart Cities,” 2020 IEEE Int. Smart Cities Conf. (ISC2), Piscataway, NJ, USA, 2020, pp. 1–7, doi: 10.1109/ISC251055.2020.9239081.
4. C. S. T. Kedanjoth, M. Thomas, D. H. Pohren, A. D. S. Roque, and E. P. D. Freitas, “An IoT-based Multimodal AI System for Emotional and Behavioral Analysis,” 2025 12th Int. Conf. on Future Internet of Things and Cloud (FiCloud), Istanbul, Turkiye, 2025, pp. 151–158, doi: 10.1109/FiCloud66139.2025.00029.
5. L. V, K. K, K. S, G. D. K, H. R, and M. V, “Reflective Wellness: An AI-Powered Smart Mirror for Personalized Health Monitoring and Insights,” 2025 3rd Int. Conf. on Artificial Intelligence and Machine Learning Applications (AIMLA), Namakkal, India, 2025, pp. 1–5, doi: 10.1109/AIMLA63829.2025.11041129.
6. M. J. Santofimia, X. del Toro, C. Bolaños, J. Dorado, and S. Colantonio, “A Smart Mirror to Your Health: Personalized Virtual Coaching for Active and Healthy Ageing,” in Digital Health and Informatics Innovations for Sustainable Health Care Systems, Springer, 2025, doi: 10.1007/978-3-031-84158-3_10.
7. M. A. Kasno and J.-W. Jung, “Feasibility of an AI-Enabled Smart Mirror Integrating MA-rPPG, Facial Affect, and Conversational Guidance in Realtime,” Sensors, vol. 25, no. 18, p. 5831, 2025, doi: 10.3390/s25185831.
8. Y. Pu, J. Zhang, M. Li, and Y. Mou, “Design of smart elderly care emotion recognition and elderly care system based on transfer learning,” Health Informatics Journal, 2025, doi: 10.1177/14727978251337893.
9. M. Jörke, S. Sapkota, L. Warkenthien, N. Vainio, P. Schmiedmayer, E. Brunskill, and J. A. Landay, “GPTCoach: Towards LLM-Based Physical Activity Coaching,” in Proc. 2025 CHI Conf. on Human Factors in Computing Systems, ACM, New York, NY, USA, 2025, doi: 10.1145/3706598.3713819.
10. Z. Yang, X. Xu, B. Yao, E. Rogers, S. Zhang, S. Intille, N. Shara, G. G. Gao, and D. Wang, “Talk2Care: An LLM-based Voice Assistant for Communication between Healthcare Providers and Older Adults,” Proc. ACM Interact. Mob. Wearable Ubiquitous Technol., vol. 8, no. 2, pp. 1–35, 2024.
11. M. Al-Ratrout, P. Ravva, S. Sharmin, A. Raikwar, J. Y. Shin, and L. Barmaki, “AFA: Identity-Aware Memory for Preventing Persona Confusion in Multi-User Dialogue,” arXiv preprint arXiv:2604.25022, 2026.
12. “The role of LLM-powered chatbots in assisting elderly people: systematic review,” Network Modeling Analysis in Health Informatics and Bioinformatics, Springer, 2025, doi: 10.1007/s13721-025-00698-9.
13. “GrandGuard: Taxonomy, Benchmark, and Safeguards for Elderly-Chatbot Interaction Safety,” arXiv preprint arXiv:2605.20203, 2026.