| Title |
Design and Evaluation of an AI-based Sleep and Emotion Prediction Framework for Age-friendly Smart Home Systems |
| Authors |
전명임(Myeong-Im Jeon) ; 강문식(Moon-Sik Kang) |
| DOI |
https://doi.org/10.5573/ieie.2025.62.9.21 |
| Keywords |
Age-friendly smart home; Sleep prediction; Emotion recognition; LSTM-based deep learning; Autonomous environment control; Multimodal sensor data |
| Abstract |
With the global aging trend, the demand for intelligent living environments that support the health and quality of life of the elderly is increasing. This paper proposes an AI-based sleep and emotion prediction framework designed for an age-friendly smart home system. The proposed system integrates various sensor and wearable device data such as light intensity, temperature, heart rate, activity level, and sleep pattern and processes them based on time series. It then predicts the user's behavioral status, sleep status, and emotional status in real time through a Long Short-Term Memory deep learning model. The prediction results are used to autonomously control the environment through smart lighting, notification devices, and other IoT devices, contributing to improved sleep quality and emotional stability. Experimental results using public datasets (CASAS, SHiFE) confirmed the performance of 91.2% in behavioral prediction accuracy, 88.6% in sleep prediction, and 84.3% in emotional prediction, and the average response time was 2.3 seconds and the control success rate was 97.9%, demonstrating real-time performance and stability. These results demonstrate the feasibility and scalability of implementing customized smart homes for the elderly. |