Patient safety is in danger because healthcare networks are more susceptible to cyberattacks as they become more intricate and linked. By altering data transmitted between various system components, malicious actors can hack into these networks. As cloud, edge, and IoT technologies become more widely used in contemporary healthcare systems, this difficulty is predicted to increase. This study presents a Combined Hybrid Deep Learning Framework with Layer Reuse for Cybersecurity (CHDLCY) to address this issue. This system is built to detect malicious actions that modify the metadata or payload of data flows across IoT gateways, edge, and core clouds quickly and precisely. The CHDLCY's is a unique design demanding less training time, while bigger models at the core cloud profit from a cutting-edge layer-merging method. The core cloud model is partially pre-trained by reusing layers from trained edge cloud models, which drastically reduces the number of training epochs required from 35 to 40 to just 6 to 8. Thorough tests demonstrated that CHDLCY not only accelerates the training phase but also achieves remarkable accuracy rates, ranging from 98% to 100%, in identifying cyber threats. The proposed approach offers a significant improvement over previous models in terms of training efficiency and generalizability to new datasets.
Read full abstract