Sharded Blockchain-based Online Diagnostic System for Suspected Patients During Pandemics

During pandemics, diagnostic tests are essential to provide quick treatment of patients and limit the disease spread. The high demand for testing resources can stress the healthcare system. Thus, a remote collection of symptoms and reporting the results via an automated diagnostic system is highly desirable. However, such a system is challenged by privacy and scalability issues. Hence, we propose a sharded blockchain-based system that (a) introduces a set of shards that distributes the testing load among a group of local nodes (LNs), hence, offering high scalability for country-wide adoption, (b) uses ring signatures and unique random identifiers to ensure the anonymity of the users and the unlinkability of test requests, hence, supporting privacy-preservation, (c) deploys a detection strategy at the LNs based on deep neural networks, which is implemented on smart contracts, hence, enabling autonomous diagnosis, and (d) provides healthcare entities with authorized access to the symptoms and test results, hence, enabling efficient data sharing that supports future research. We provide an implementation of the proposed system and our experimental results demonstrate the high scalability and privacy of the system while achieving a testing accuracy up to 90%. We present a case study for U.S. wide deployment showing that a total daily test request of 2, 407, 462 can be performed and reported in 11 minutes compared to 63 days in absence of sharding. Moreover, sharding decreased the user storage requirement to be 0.18 MB at maximum instead of 723 MB without sharding.