Real-Time Multimodal Context-Aware System for Elderly Care Using Integrated Temporal-Spatial Reasoning in Smart Homes

With the growing elderly population, ensuring safety and independence at home has become more important than ever. Although ambient intelligence-enabled smart homes offer promising alternatives, data collection by itself won't give a comprehensive picture of how senior citizens behave on a daily basis. To monitor and assist senior citizens in smart home settings, this paper proposes a real-time, context-aware system that integrates deep learning and temporal-spatial reasoning. Using the publicly available CASAS dataset, we trained an LSTM-based model to recognize routine activities and spot anomalies based on deviations from typical behavior patterns. To improve the system's comprehension of spatial context, our method also incorporates semantic reasoning via ontologies.