Ethical hazards of health data governance in the metaverse

Abstract

Metaverse-enabled healthcare is no longer hypothetical. Developers must now contend with ethical, legal and social hazards if they are to overcome the systematic inefficiencies and inequities that exist for patients who seek care in the real world.

In their Nature Machine Intelligence article, Wang and colleagues¹ present the metaverse as a virtual space “enriched by effectively unlimited data” and capable of driving substantial innovations in medical technology and AI (MeTAI). The metaverse is prime for developing an interactive forum for patient care. It is also where ‘digital twin’ data will, according to the authors and others in the field², proliferate and provide the basis for simulating diagnoses and opportunities for virtual clinical trials that help advance medical technologies and improve care. Wang and colleagues¹ envision that the metaverse will be populated with data involving patients, their avatars and phantoms that will be “seamlessly” integrated and shared broadly across virtual worlds. Such data availability would support dynamic evaluation of AI-based software as a medical device and innovations in medical device development tools.

As computational bioethicists who have investigated blockchain applications^3,4 and applications of related tools such as nonfungible tokens (NFTs)⁵ in healthcare and medical research, we share the authors’ curiosity about ways to leverage the power of modern computing and reimagine a more equitable, state-of-the-art healthcare system. We argue that realizing Wang and colleagues’ vision for responsible MeTAI requires first contending with real ethical, legal and social hazards that are likely to limit data liquidity in the metaverse without scrutiny. At worst, irresponsible MeTAI policy could exacerbate informational injustices that many patients currently experience in some healthcare systems in the real world⁶.

In this Commentary, we expand on four key areas that are ripe for ethical–legal analysis and sustained public consultation for responsible health data governance in the metaverse: democratization of health information exchange infrastructures; data ownership and control; compensation for data contribution(s); and metaverse literacy to enable informed participation. We directly comment on the ethical implications of innovations discussed in the article by Wang and colleagues¹ and are therefore not exhaustive. Myriad other ethical, legal and social issues⁷ associated with delivering metaverse-enabled healthcare, or the ‘medaverse’, is the focus of our ongoing work. In brief, they include issues of medical licensing, the quality of patient–provider relationships, bias and algorithmic injustice, equity and access to care.

Privileged versus democratized health information exchange

The first area that developers and prospective users of MeTAI need to consider is how existing health information exchange (HIE) infrastructures are likely to limit the open flow of personal health information across virtual worlds. Implicit in Wang et al.’s arguments¹ is that health data will be more readily available and accessible in the metaverse, and that users will be able to access infrastructure technologies that power the metaverse. For example, the metaverse requires greater compute power than many household broadband internet services can feasibly support. This creates access barriers for many prospective end users and contributes to what has been called the great digital divide⁸. Despite legislative reform in the United States under the 21^st Century Cures Act, personal health information remains siloed, electronic record vendors deprioritize interoperability⁹ and researchers and other covered entities are incentivized to hoard rather than share data. HIE ecosystems, particularly in the United States, often lack the trusted data-processing environments needed to support secure access, use and exchange of sensitive health information. This problem is formally recognized by the European Union’s recent Data Governance Act¹⁰, which attempts to encourage ‘data altruism’ by investing in trusted data-processing ecosystems and technologies, including privacy-enhancing technologies. Insofar as data are resource-intensive to generate and expensive to store in large volumes without exclusive, proprietary access, data will remain paywalled and profitable for a privileged few in the HIE market. We expect that HIE in the metaverse will mirror these same challenges, absent reform in how providers, payers, patients, researchers and health data brokers prioritize data sharing.

Patient data ownership and control

Wang and colleagues also suggest that federated learning, blockchain and NFTs can help “to preserve the integrity of patients’ data” such that that patients decide when and with whom their own digital healthcare properties can be shared. Yet, US privacy laws do not recognize patients as legal owners of their health data, nor as arbiters of the data’s downstream use in many cases. Indeed, the courts ruled against an individual in Dinerstein v. Google, for example, who claimed unauthorized use of their health data but could not demonstrate damages incurred as a result¹¹. We expect that claims to health data ownership in the metaverse will similarly remain ambiguous, and privacy violations even more difficult to litigate given the jurisdiction-less virtual environment. Debates about data ownership further complicate the use of NFTs to compensate individuals or distribute ‘royalties’ for their data contributions¹², particularly given that myriad actors are involved in generating, collecting, storing and securing these data.

Compensation for patient data contribution(s)

The metaverse offers a unique opportunity to reward data sharing or exchange among individuals directly to authorized users through decentralized ledger and encryption technologies. Such technologies open the door for future patient-controlled access to data using computational ‘tokens’ or NFTs that digitally attribute ownership. Individuals could actively seek out data transactions in metaverse marketplaces or in the context of decentralized autonomous organizations (DAOs), which incentivize members to crowdshare assets (for example, funds, data and computer processing power) in exchange for a stake (that is, token) in the DAO’s returns. Stakes could be monetary (such as, returns on crowdsourced investments) or informational (for example, insights into effective treatments for rare diseases derived from pooling geographically dispersed data).

DAOs are quickly becoming the ‘companies’ of the metaverse¹³, but with co-constructed smart contracts, rather than traditional business executives, outlining the terms of engagement. These novel, virtual entities and communities allow patients to engage in transactions that are automatically executed using smart contracts. The contractual terms are then documented in a tamper-resistant, peer-to-peer fashion without a trusted intermediary or escrow service paid to receive or disburse assets. The revolutionary nature of this scheme lies in the disintermediation of personal health information with other traditional covered entities and their business associates who may also claim ownership rights or control over patient data. Rather than relying on digital health platforms to broker health data, individuals could select from pre-scripted or bespoke smart contracts to fit their data-sharing preferences, allowing them to donate, exchange or sell data involving themselves or their online avatars.

Wang and colleagues¹ suggest that in theory, patients could take greater control over their own data and avatars and rely less on centralized data brokers, who rarely solicit patient preferences or consent. In practice, many patients may find the added responsibility to review and consent to every new request for data use intimidating or even burdensome. In a recent proof-of-concept study, Zichichi et al.¹⁴ illustrated what such a blockchain-based personal information management system could entail. In order for individuals to truly ‘own’ and control their own data, those data need to be stored in a decentralized file storage system (such as the InterPlanetary File System) to which data requesters gain access only by meeting certain criteria outlined in a shared access control list. Data access requests and permissions are automatically processed using smart contracts, with all transactions documented in a publicly accessible and auditable decentralized ledger. Even when authorization servers are leveraged to relieve data owners from the burden and computational costs of completely handling data access and distribution, patients are solely responsible for managing access to their own data. This includes managing their private keys, or else risk losing access to their own data, much as investors who have lost access to their crypto assets due to misplaced private keys¹⁵.

Literacy and informed participation in MeTAI

The complexity and responsibility of participating in such a multi-ledger system places considerable barriers on truly democratized control over patient data. Such systems can also reintroduce a reliance on experts who conceptualize, encode or broker access to these systems on behalf of patients. The implications for the metaverse are that until these systems are worked out and made accessible to ordinary citizens, the metaverse will continue to be a place where patients have no more or less control over their patient data than in the ‘real’ world — which is to say, not much. Inaccessibility of the metaverse due to infrastructure barriers (for example, broadband connectivity), low technical literacy or other causes is likely to have disproportionate effects on certain users in terms of engaging with the metaverse and inputting data into the AI/ML systems in the metaverse, and in regard to what data-driven health interventions stem from these data. Patients in the metaverse must continue to rely on (that is, trust) centralized entities to be responsible stewards of their data. These ‘trustless’ environments for data sharing, though promising, must be perceived as ‘trustworthy’ and feasible if they are to play a role in encouraging data altruism and liquidity in the metaverse. Indigenous scholars in our own field of genetics and genomics have proposed compelling models for blockchain-enabled sharing of genomic and related health data that follow sovereignty-by-design principles and that address issues of both community trust in data projects and trustworthiness in the computing environments that process these data¹⁶.

Conclusion and future directions

The more granular these data become, and the more ‘twinned’ to our human selves, the more urgent it becomes to find meaningful data privacy solutions, including those that permit inferences about data to be shared while the data remain private and unduplicated. The promise of the metaverse is not just to become a forum for new forms of data generation, sharing and discovery, though this is undoubtedly important. Instead, the metaverse’s great challenge is to conceptualize new technological infrastructures and frameworks that effectively balance two opposing needs: to democratize patient privacy and agency over health data while simultaneously facilitating ever more data liquidity to feed data-hungry AI/ML systems in ways that effectively and equitably enhance human health. More normative and empirical research on how to inclusively develop, ethically deploy and rigorously evaluate healthcare applications for the metaverse is required to reach the above-mentioned goals.

Future research agendas should explicitly address the ethical tensions between the need for privacy-preserving MeTAI and the equitable distribution of healthcare benefits for all patient populations. We also currently lack data about what different stakeholders perceive as the key advantages and technological pitfalls of MeTAI. Public consultations with both prospective end users and regulators could fill existing knowledge gaps to provide better understanding of the needs of patients, providers and payers. Researchers working on practical solutions will likely need to assemble multidisciplinary teams that leverage expertise in bioethics, computer science, medicine and law, among other relevant fields. Although the patient care impacts of MeTAI-driven healthcare may still be hypothetical, the emerging realities of a virtual healthcare universe powered by such technologies are no doubt real.