Blockchain Technology for Ophthalmology: Coming of Age?

Abstract

Ophthalmology has been an early adopter of cutting-edge digital technology such as artificial intelligence (AI) and could be primed to integrate blockchain architecture in the management of AI and big data analytics. Blockchain technology has seen rapid development and maturation over the past few years. Invented in 2008 by Satoshi Nakamoto (a presumed pseudonym for an anonymous inventor/group of inventors), the blockchain ledger, forming the basis of Bitcoin, was the first decentralized cryptocurrency.1 Since then, different blockchain infrastructures, with different consensus frameworks have emerged, such as Ethereum2 and Hyperledger Fabric.3 In essence, blockchain is a decentralized ledger with immutable properties allowing secure verifiable transfer of data in a peer-to-peer fashion, utilizing a common consensus protocol to prevent single points of failure. A variety of use cases have been developed to harness the advantages of blockchain technology, largely in the financial sector and the business world. However, the health care sector, which requires tight control over confidential medical data, is well-poised to take advantage of the unique characteristics of blockchain technology. This could be of particular relevance in the management of big data and AI research, notably in the field of ophthalmology which has an abundance of numerical data and imagery. In this review, we introduce the basic concepts of blockchain technology, discuss its unique advantages and its use case in ophthalmology. BLOCKCHAIN TECHNOLOGY The concept of blockchain technology is built on a framework of an ever-growing list (or “chain”) of transactions, grouped into units called “blocks”, which in turn are linked to their immediate predecessors by a unique cryptographic “hash value”. Generated from specific mathematical algorithms such as the SHA-256 (256-bit Secure Hash Algorithm), hash values are characterized by their deterministic value, as well as pre-image and collision resistance, serving as “fingerprint” of a block and its content. Blockchain platforms depend on “consensus protocols” to approve, record, and validate each transaction. A consensus protocol is a form of rules to reach a common majority agreement on the present state of the ledger within a blockchain network. Once a consensus has been reached, the block containing the log of the data is added into the “chain”, and shared with every stakeholder in the network, known individually as “nodes”. The archetypal example would be a bitcoin transaction. When the transfer of coins is initiated between 2 parties, miners compete for the right to record the transaction by solving a difficult mathematical puzzle, expanding precious energy in the process. The node that solves the puzzle first is recognized for its efforts and given the right to record the transaction, hence reaching a consensus. The common transparency and traceability of each individual cryptoasset transaction prevent duplicative transactions, solving the well-documented “double-spend” problem—an inherent flaw in digital asset schemes where the same single-use digital token is spent more than once. This forms the basis of a hack-resistant, immutable distributed ledger. Current blockchain platforms can be broadly dichotomized into two main groups—permissionless or permitted (Fig. 1). Permissionless blockchain platforms such as Bitcoin and Ethereum provide unrestricted access to the public. Conversely, permitted platforms such as Hyperledger Fabric will retain a central approving authority. Hybrid or consortium-based blockchain platforms are derived from a combination of these two architectures, resulting in partial centralization with participation restricted through the private network. Table 1 provides a summary of the terms and definitions unique to blockchain technology.FIGURE 1: Permissionless blockchain targets at implementing a common platform that can involve anyone with anonymous identity into the network, which often comes with built-in currency and is public, open and fully decentralized; Permitted blockchain engages a few organizations with known identity (forming a consortium) that require collaborative operations to realize specific business logics, which are co-hosted/co-managed by the consortium. TABLE 1 - Summary of Common Terms and Definitions in Blockchain Technology Ledger Book or computer file for recording and totaling transactions Blocks Transactions cumulated and recorded into fixed sized blocks. Each block contains timestamp, a unique hash, the hash of the previous block and transaction data 23 Chain List of blocks linked cryptographically23 Hash Cryptographically generated fixed length string of values, based on random input of transactions/data, that is easily verifiable23 Mining Validating of transactions and recording onto the decentralized ledger24 Byzantine generals problem Computer science description of a situation where involved parties must reach a decision to avoid failure, but some parties are dishonest or malicious25,26 Consensus protocol A form of rules to reach a common majority agreement on the present state of the ledger within a blockchain network23 Fault tolerance Level that allows a system to continue operating normally in the event of failure of some components or nodes27 Nodes Communicating points that may perform different functions on the blockchain platform23 Permissioned Access control layer governed by a central authority23 Permissionless Public access without restriction to participation23 Hybrid/Consortium Blockchain platform governed by multiple organizations23 Immutable Unchangeable ledger23,24 Cryptoasset Digital assets that utilizes cryptography as a medium for transactions24 Smart contracts Automated executions of complex transactions based on computational logic when certain conditions are met24 Asymptotic security Security if and only if the adversary's advantage is a negligible function of the security parameters i.e. a secure scheme that is conditionally proven to be harder than any polynomial for the attacker to break28 Deterministic Same operation performed by different nodes will produce the same result29 Pre-image resistance Computationally infeasible to derive the original transaction data from a given hash function30 Collision resistance Computationally infeasible for two distinct inputs to result in the same hash output23 APPLICATIONS OF BLOCKCHAIN TECHNOLOGY Apart from application in finance, blockchain technology could be highly relevant in other industries such as health care, insurance, and supply chain management4 due to the inherent key advantages: immutable transaction records, decentralized peer-to-peer transaction, costless verification, reduction of incumbent market power, avoidance of single point of failure, and smart contracts – automated executions of complex transactions based on computational logic. Although it has yet to achieve mass-market adoption, blockchain technology has been heavily touted as a potential general-purpose technology, gaining traction across multiple industries such as finance, hedge fund management, and supply chain management.4 Years of reliance on antiquated digital systems have resulted in cumbersome, inefficient and resource-intensive processes. This results in significant resource wastage, and also renders systems susceptible to fraudulent attacks or system-wide failure.5 Financial institutions are therefore innovating with blockchain technology to address these concerns. Another frontrunner in the adoption of blockchain technology is supply chain management.6–8 The complex multi-faceted nature of supply chains places heavy demands on proper record keeping, quality control, and transaction monitoring. Current supply chains rely on centralized intermediation entities with little transparency across the entire chain. The supply chain thus suffers from vulnerability towards malicious modification or human errors and poor accountability. The application of blockchain technologies has the potential to disrupt the industry by effectively eliminating the trust required between involved parties. By transferring the onus of trust onto the algorithm and its immutable record, the issues associated with the need of verifying intermediaries can be eliminated.9 BLOCKCHAIN TECHNOLOGY IN HEALTH CARE In a 2019 technical report by International Telecommunication Union,10 the Telecommunication Standardization Sector identified the health care sector as one of the key sectors that could be a beneficiary of blockchain technology. The devastating COVID-19 pandemic, while unfortunate, has provided a significant impetus to accelerate this process.11 It is important to note that traditional distributed database management system (DDMS) can support the secure transfer of health data through encryption and data masking with the acceptance of several significant flaws: potential single point of failure, subject identification and tampering of data. In comparison, blockchain armed with asymmetric encryption and hash values can surmount these challenges albeit with a measured sacrifice of throughput rate and latency. At this point, adoption of blockchain in health care is still in its infancy with multiple proof of concepts but a limited selection of commercially available health care blockchain platforms. At present, most of these platforms are focused on electronic medical record management, such as patient-controlled electronic medical record (EMR) accessibility and immutable recording of clinical records. One of the most well-known is Medicalchain (Medicalchain SA, London) which is built on the Hyperledger Fabric architecture. Medicalchain's12 primary focus is to assign EMR access-granting rights to the patient, thereby returning control back to the patient. It provides a self-contained incentive system by rewarding data-sharing behavior with its native token (MedToken) which can be utilized in exchange for relevant services. Another example is the national rollout of the e-Estonia health care EMR built on Keyless Signature Infrastructure blockchain technology, allowing for verification of integrity of accessed medical records as well as immutable record of access logs.13 Since its inception in 2016, it has enabled digital permeation with 99% of health data digitized securely and handling up to 1.8 million patient queries every month, made possible through decentralized authenticated sharing of data. Notable examples of data sharing on the e-Estonia platform include physician retrieval of time-critical patient information during emergencies as well as patient monitored access of their medical data.14 Separately, blockchain could potentially be a disrupting technology in the health care supply chain and insurance field. The decentralized nature of blockchain provides a platform for cross-institution and cross-border collaboration, providing transparent check and balance to all stakeholders. This gave rise to initiatives such as Pharmaledger—a European Union blockchain consortium involving 12 global pharmaceutical firms such as Pfizer, Novartis and GSK.15 In the health care insurance field, peer-to-peer transaction of cryptographically-secured sensitive information between stakeholders would remove costly intermediaries and improve efficacy. Fraudulent activities would also be deterred by the algorithm and the immutable log. This could transform the entire patient-customer journey, from verifiable health declaration during policy purchase to transparent and traceable claims process. It is thus becoming apparent that the trust-less verification and immutable audit trail afforded by blockchain is exceedingly crucial for innovative applications in health care. BLOCKCHAIN TECHNOLOGY IN OPHTHALMOLOGY Ophthalmology as a field has been an early adopter of new evolving technologies, in particular the application of AI and deep learning (DL) for the automated analysis of medical images, such as retinal images and optical coherence tomography scans.16–18 DL in medicine (and ophthalmology), an area of active research, is highly reliant on the availability of large high-quality datasets as well as rigorous model validation and testing. However, the management of diverse datasets from different countries and centers for training and testing of algorithms in these studies poses significant challenges. This is attributable to extensive restrictions due to concerns over data security and patient confidentiality, preventing honest transfer of research medical data to support collaborative efforts.19 In addition, proper research community oversight over the multitude of novel AI and DL systems is unattainable due to a lack of transparency regarding model validation and testing. Recognizing these challenges, in a recent study, Tan et al20 proposed a permission blockchain-enabled platform (based on Hyperledger Fabric) to assist with the development and validation of DL algorithms to tackle the global myopia epidemic. They provide proof-of-concept, using this blockchain-enabled platform for secure handling of data transfer, model sharing, and auditable reporting of model validation and testing results across 3 separate sites in 2 countries, in the development of robust DL algorithms for automated detection of myopic macular degeneration and high myopia from retinal images. They suggest that this blockchain-based solution for the management of research datasets and model testing results provides advantages of data integrity and immutability, as well as automation in data consistency and a shared ledger promoting easier collaboration. They also suggest that widespread adoption of this novel method could increase validity and transparency of AI studies in medicine, and may allow health regulators (eg, US Food and Drug Administration) a means of effectively auditing and verifying the diagnostic performance of AI algorithms for regulatory approval. In conjunction, the immutable transaction log replicated across all nodes provides the ideal digital infrastructure to track each iteration of the AI-model training, improving collaborative efficiency and trust. Further applications of blockchain in the health care industry could likewise impact the field of ophthalmology (Table 2). Supply chain transformation is particularly valuable for perishable products21 and would likely play a significant role in health care, where tight monitoring of labile high-value medications is crucial to guarantee safety and efficacy. Scarce or costly products that require highly regulated storage conditions would be ideal candidates, examples which include total parenteral nutrition, mRNA COVID vaccines, intra-vitreal anti-vascular endothelial growth factors (anti-VEGF), blood products or biologics. During the COVID-19 pandemic, Lin et al established a blockchain-based platform to provide virtual clinical service for ophthalmology patients. They proposed a proof-of-concept to verify and efficiently monitor online prescriptions, creating a blockchain-based online pharmacy for prescription renewals and remote drug delivery.22 The significance of this lies in the fact that telemedicine is a highly visual-dependent service, making it particularly well-suited for visual-oriented specialties like ophthalmology. Another potential application of blockchain in health care, which will be highly relevant in ophthalmology, is the use of blockchain technology to monitor and improve patient treatment adherence as well as to automate medication support programs. This would be particularly valuable for high-cost treatment regimes that require considerable patient compliance, such as recurring intra-vitreal anti-VEGF treatments. TABLE 2 - Blockchain Characteristics and Potential Use Case Immutability31–38 AI algorithm training and testing, health care insurance, data transfer, medication distribution supply chain, patient support programme, medical licensing, patient disease monitoring, clinical drug trial Traceability and provenance31,32,34,37,39 Data transfer, medication distribution supply chain, patient support programme, medical licensing De-centralized data security33–37 Patient support programme, health care insurance, patient disease monitoring, data transfer Peer to peer transaction31–35,37 Health care insurance, data transfer, patient support programmes Cost-less verification33,35,37–40 Health care insurance, patient disease monitoring, patient support programmes, medication distribution supply chain, clinical drug trial Smart contracts/De-centralized autonomous organisations33,35,40,41 Medication distribution supply chain, patient support programmes, health care insurance, Anonymity31,32 Data transfer, patient support programme CHALLENGES Although we expect greater innovative and disruptive use cases for blockchain technology in ophthalmology to materialize, implementation and integration could remain a challenge. First, selecting the appropriate blockchain platform will be critical which, under most health care circumstances, will exclude permissionless blockchains. Researchers, clinicians, and hospital administrators will need to be cognizant of the clinical and operational workflow changes required if blockchain is adopted. Second, mindsets deeply rooted in the traditional DDMS will need to be changed. In addition, switching from DDMS to blockchain will entail greater digital automation, integration of application programming interfaces (API) and distributed applications (Dapps), off and on-chain event connection, wholesale conversion to digitized data collection and upgrading of the information technology infrastructure. Third, there are significant costs. Investments will be required for dedicated digital hardware, networking and storage overheads as well as maintenance. Fourth, from a clinician and provider perspective, immutability of the blockchain platform will prevent amendments of erroneous entries whereas latency could be a source of frustration when contrasted against highly efficient traditional client-server databases. Fifth, blockchain relies on a flawless algorithm to create an asymptotic security, hence cryptographic flaws could leave vulnerabilities within the platform. Finally, data security issues need to be addressed. The decentralized peer-to-peer transaction could potentially compromise patient's data if it has not been appropriately anonymized. The sensitive nature of health care data hence demands that the algorithm undergo extensive trials and penetration tests to guarantee that the patient's privacy and confidentiality is upheld. Finally, the democratization of data sharing at the patient-level might not materialize. It is highly possible that the lure of incentives would be nullified by heightened senses towards privacy preservation, hence failing to convince and motivate patients to proactively share their data. Such inertia could be further compounded by a lack of understanding and trust of the reliability of blockchain platforms. It might be more realistic to consider monetization of big data in an institutional level, yet even that faces significant resistance for fear of privacy breaches or the loss of autonomy over valuable data (Supplementary table: https://links.lww.com/APJO/A87). CONCLUSIONS AND FUTURE DIRECTIONS In conclusion, the health care sector faces a pressing need for new digital technologies that allow secure and efficient sharing of data to address the inefficiencies and demands in current systems. There is increasing recognition that blockchain technology could deliver the novel digital platforms that are required to address these requirements. In the field of ophthalmology, blockchain technology can help to monitor data and results integrity, enabling greater research collaboration to support continued AI development. In addition, newer protocols and algorithms are likely to further improve the functionality and scalability of blockchain for widespread health care applications beyond eye care.