Abstract The need for advanced health monitoring systems has grown with the world's fast-growing aging population. To foresee time-series health habits from wearable sensor data and trigger smart responses tailored for elderly citizens, this research proposes a novel architecture that leverages Long Short-Term Memory (LSTM) networks. Several layers constitute the proposed architecture: wearable sensors for data acquisition, edge devices for preprocessing in real-time, LSTM models for predictive analytics over time, and a logic-based intelligent reaction mechanism. The solution utilizes a complete dataset named "AI-Driven Elderly Care: Real-Time Monitoring & Assistance," which includes activity logs, emergency alarms and physiological information like heart rate, glucose level. An optimized two-layer LSTM model is the foundation of the approach that identifies long-range dependencies in healthcare data. A rule-based knowledge response system takes predictions such as poor mobility, abnormal heartbeat or disrupted sleep and employs cloud-edge synchronization based on MQTT and HTTP protocols to dispatch real-time alerts, nudges or caregiver warnings. With 96.8% accuracy, 95.6% precision, 96.2% recall, and a 95.9% F1-score, the model performed well. Through the local processing of insights, this approach significantly reduces latency, enhances real-time interactivity and ensures data privacy compared to traditional centralized systems. The proposed framework's early anomaly detection and customized intervention reflect how ideally suited this approach is for elderly care. This work introduces a wearable-based deep learning system-based scalable, low-latency, and smart scheme of monitoring the well-being of the elderly, which is an essential contribution in smart healthcare environments.
Ji et al. (Wed,) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: