Protecting Healthcare Data: Blockchain-based Solutions for Security, Privacy and Scalability

Every individual in our digitally advanced world requires adequate data protection. The transmission of healthcare information is increasingly focused on protecting data privacy and improving scalability. The rapid progression of communications technology has made the sharing of distributed information across several domains an inescapable trend, enhancing transmission of information with an emphasis on privacy and scalability. The fast expansion of blockchain systems sometimes requires the substitution of customized blockchains with more secure, scalable, and decentralized alternatives. Ensuring privacy in healthcare data is particularly tough due to stringent data laws and regulations. Permissioned blockchain solutions are gaining prominence due to their transparency, immutability, and customization features. A consortium-based permissioned blockchain provides better security to healthcare data. However, protecting privacy is crucial and demanding. Federated learning (FL) ensures data privacy as the federated devices retain the real data locally without sharing it. It serves as an advantageous instrument when used with blockchain technology. With the increasing usage of Internet of Healthcare Things (IoHT) devices, centrally managing them and ensuring the privacy of healthcare data is becoming increasingly difficult. A shard-based blockchain architecture is suggested for federated learning networks to enhance scalability and privacy, offering improved administration and privacy controls for data sharing within the network. We discussed an innovative and reliable access control mechanism for sharing healthcare-oriented FL models that utilize permissioned blockchain technology to solve access control and privacy concerns. A distributed federated learning-based model sharing architecture designed to safeguard the privacy and security of healthcare data through the implementation of an Attribute-based Access Control (ABAC) model, whereby the trust attribute is computed and used for access control decisions. Blockchain, as a type of distributed ledger technology, is inherently trustworthy; however, it lacks computing capability and has excessive delay due to its cumbersome consensus methods. We explore an alternative scaling strategy for a distributed healthcare federated learning-based secure model sharing architecture. The system employs state channels to minimize on-chain transactions, mitigate architectural delay, and decrease bandwidth use, therefore easing the strain on the blockchain. For a deeper understanding, we evaluate two layer-2 blockchain methodologies (ZK-Rollup and State Channels) on throughput, latency, complexity, and manageability to enhance scalability and privacy in healthcare data exchange. We found that state-channels become more complex to manage with the increasing number of participants in the consortium network and adopted the ZK- Rollup L-2 solution for further evaluation. To improve the performance of the ZK-Rollup L-2 method, we introduce a novel concept of PoA for the scalability of ZKP and the privacy of healthcare data. Commitment values are calculated over EHR using KZG polynomial and ZK-SNARK-based ZKP system (PLONK) is used to generate proofs with these commitment values. Instead of raw data, it significantly reduces the proving time in the ZK-Rollup system. This novel system not only provides the proof of availability of data but also correctness, data retrievability, consistency with the commitment values, and ensure maximum privacy of healthcare data with lower overhead on the ZK-Rollup system.