A novel end-to-end privacy preserving deep Aquila feed forward networks on healthcare 4.0 environment

Healthcare 4.0 is considered to be the most resilient technology with the integration of Internet of Things (IoT), Artificial Intelligence (AI), and 5G wireless communication. These Internet-enabled devices compile human data and deliver it to remote clinical centers for efficient diagnosis and treatment processes. Though sensor-driven devices have largely eased everyday lives, these healthcare systems have been suffering from various security breaches and data privacy problems. This evokes a need for designing intelligent systems to eradicate data breaches and privacy problems. This research presents a groundbreaking framework that uniquely combines privacy-preserving optimized deep learning to effectively diagnose cardiac troubles utilizing edge and fog computing devices. The recommended framework is tested with real-time datasets and employs federated learning for training the network. To ensure high prediction performance, Gated Aquila Optimized Deep Feedforward Networks (GAODP) are used to predict heart diseases. In the primary instance, information is acquired by the IoT sensors and stored in fog gateways. Subsequently, the federated mode of GAODP is adopted for the prediction of heart diseases. The ablation tests are conducted utilizing IoT nodes connected to medical sensors, with fog gateways implemented using Embedded Jetson Nano gadgets. Extensive evaluations are carried out using the gathered datasets, and performance metrics are examined and analyzed. The experimentation involves various traditional deep learning frameworks to prove the performance of privacy-preserving models. Moreover, the duration required to construct the model is calculated to confirm the recommended model's intricacy. The outcome indicate that the recommended model has attained notable predictive accuracy, including precision (0.98), recall (0.975), specificity (0.98), F1-score (0.99) and accuracy (0.99) with less computational overhead.