ABSTRACT
The integration of the internet of things (IoT) and electronic health records (EHR), which enables continuous monitoring and seamless data integration, offers numerous benefits such as remote patient monitoring, personalized healthcare, and improved patient outcomes. However, efficiently allocating resources in the IoT-EHR ecosystem poses significant challenges in decision-making. To address the problem, this work develops GraLSTM, a distributed model that combines graph neural networks (GNNs) with LSTM for efficient anomaly detection. The GraLSTM architecture effectively captures spatial and temporal data dependencies, outperforming GCN and GNN. Experimental findings after training with dynamic data showed that GraLSTM, with multiple IoT devices, achieved the lowest loss and batch loading time, resulting in fast computational speed and scalability to support clinical decision-making.
