The novel contribution is an end-to-end co-designed architecture that couples an edge-deployed deep-learning predictor with 5G transport for real-time remote patient monitoring. A hybrid CNN-LSTM model augmented with attention mechanisms processes multivariate vital-sign streams (heart rate, blood pressure, respiratory rate sampled at 100 Hz) to predict deterioration, and is optimized for edge inference; data transmission relies on 5G Ultra-Reliable Low-Latency Communication (URLLC) with network slicing to guarantee QoS on a dedicated healthcare channel. The integrated system reports 14.4 ms end-to-end latency and 96.5% multi-vital-sign prediction accuracy - a 47% latency reduction and 4.2% accuracy improvement over cited state-of-the-art - validated over three months on 1000 patients. The core idea is that jointly optimizing the model for edge deployment and the transport for URLLC/slicing is what unlocks sub-second, reliable vital-sign prediction in clinical settings.
This paper proposes a system architecture that integrates deep learning with 5G network capabilities for real-time remote patient monitoring and prediction of vital signs (heart rate, blood pressure,…