Securitizing Patient Record and Access Using Ethereum Smart Contract Graph Embedded Pyramid Network Face Recognition

With more and more data being generated, identifying a secure and effective data access framework has become a crucial research issue. Technological evolutions have been made in numerous areas to name a few being, agriculture, industry and specifically healthcare systems. Healthcare has experienced several remarkable alternates as the comprehensive transformation from document-based storage to electronic health records (EHR). The centralized mechanisms utilized by medical institutions for the Electronic Medical Records (EMR) management and transfer can be highly susceptible to security and privacy menaces. Blockchain have been a fascinating research area over a long period of long time and the advantages it imparts have been utilized by a number of several industries. In a similar manner, the healthcare sector stands to ease extensively from the blockchain technology owing to security and privacy. In this work to securitize patient record and access using a method called Ethereum Smart Contract Graph Embedded and Pyramid Network (ESCGE-PN) is proposed. In this paper, we first design a Graph Embedding-based Neural Network and incorporate it into Differentiable Permutation Invariant Ethereum Smart Contract to securitize patient medical record. Our proposed Differentiable Permutation Invariant Ethereum Smart Contract Graph Embedding-based Neural Network makes managing healthcare records using Permutation Invariant operator to accept arbitrary samples and maps patient records with the corresponding face images to form blocks in blockchain. We then propose novel Feature Fusion Pyramid Network Cosine Similarity-based face recognition for patient information access, which uses blockchain along with the the Feature Fusion Pyramid Network to address security concerns and enables selective sharing of medical records between doctors and patients. To measure the ESCGE-PN methods performance, four different performance metrics, data confidentiality, data integrity, recognition accuracy and recognition error are validated and analyzed. As a result, ESCGE-PN method achieved a higher performance compared to other state-of-the-art methods.