Blockchain-Enabled Technique for Privacy- Preserved Medical Recommender System
Eric Appiah Mantey, Conghua Zhou, Joseph Henry Anajemba, Yasir Hamid, John Kingsley Arthur
Abstract
With the proliferation of privacy issues surrounding the Internet of Medical (IoMT) recommender system data, this study presents a Secure Recommendation and Training Technique (SERTT) which is contingent on a combination of both federated learning and blockchain approaches. Firstly, the study presents a new framework for recording, sorting, and transmission of IoMT data while incorporating blockchain to ensure that the IoMT data transmitted to cloud servers is not made vulnerable by the sharing of the original data. Secondly, by utilizing medical data, the study designs a Recommender Data Management Neural Architecture (REDMANA) which is based on federated learning and model searching training framework. The proposed technique guarantees that the model gradients which are trained by each node are not disclosed all through the universal training and modeling procedure. This makes the raw data inaccessible to either the IoMT data provider or the user. Considering that the model ensures that users can only obtain their necessary inquiries, neither medical data suppliers nor users can obtain access to raw data. Thus, it reduces the issues of safeguarding medical data sets to the issues of securing data processing. Using numerical analysis and experiments the proposed technique is compared with other existing techniques, the result shows that the proposed SERTT system is efficient and secures recommender data management training and modeling technique and that it performs previously designed techniques as compared.