Analyzing the security of e-health data based on a hybrid federated learning model

Abstract

This research aims to provide an approach for analyzing the security of the e-health care system through the pre-processing of different deep learning models and the usage of federated learning. Infrastructural support for e-healthcare services is being gradually deployed by the health sector. This method increased the safety of patients and doctors through a protected platform. As a result, it is going to replace the current health service. Even if this technology is becoming more and more widespread, a number of data security threats need to be tackled. In this research, a CNN and MLP architecture with a classification-focused approach using a number of pre-trained feature extractors such as ResNet-50, VGG16, and Inception- v3 have been implemented. Additionally, various machine learning classification algorithms and Other methods of picture classification besides neural networks have been tried and compared, including Random Forest and Logistic Regression. Federated learning has also been incorporated to increase the security of healthcare data because it does not transmit actual data but models. The objective is to develop a hybrid federated learning model to analyze the security of e-health data. The core premise is to utilize a methodology like federated learning, which enables a technique for creating machine learning models while safeguarding user privacy and can maintain e-health data security without transferring real-world data.