Science - USA (2022-02-04)

500 4 FEBRUARY 2022 • VOL 375 ISSUE 6580 science.org SCIENCE

By Kristin Kostick-Quenet^1 , Kenneth D.

Mandl2,3, Timo Minssen^4 , I. Glenn Cohen^5 ,

Urs Gasser^6 , Isaac Kohane2,3, Amy L. McGuire^1

P

ersonal (sometimes called “pro-

tected”) health information (PHI)

is highly valued ( 1 ) and will be-

come centrally important as big

data and machine learning move

to the forefront of health care and

translational research. The current health

information exchange (HIE) market is

dominated by commercial and (to a lesser

extent) not-for-profit entities and typically

excludes patients. This can serve to under-

mine trust and create incentives for shar-

ing data ( 2 ). Patients have limited agency

in deciding which of their data is shared,

with whom, and under what conditions.

Within this context, new forms of digital

ownership can inspire a digital market-

place for patient-controlled health data.

We argue that nonfungible tokens (NFTs)

or NFT-like frameworks can help incentiv-

ize a more democratized, transparent, and

efficient system for HIE in which patients

participate in decisions about how and

with whom their PHI is shared.

NFTs grew out of the concept of “tokens”

in gaming, whereby a “fungible” token can

be used to purchase a thing of value (e.g.,

a gold coin to spend on a superpower) but

a NFT can only be traded, given its intrin-

sic value that is not directly comparable

to that of another token (e.g., a specific

sword versus a specific tapestry). NFTs

have evolved into digital contracts com-

posed of metadata to specify access rights

and terms of exchange. Their nature as

metadata means that NFTs point to digi-

tal content but are not the content itself.

The use of NFTs as a tool for digital artists

to prevent the unsanctioned circulation of

artwork online has since bled into sports,

entertainment, and even health care, com-

modifying digital information and creating

a multi-billion-dollar market.

NFTs are made up of a unique 40-digit

identification code (“hash”) and a uniform

resource locator (URL) linking to the con-

tent online, forming a “smart contract”

that can range from just a few lines of

computer code to a more elaborate set of

instructional code. These contracts desig-

nate a patient-controlled copy of the digi-

tal data and the terms under which they

can be accessed and used, using pseud-

onyms that permit deidentification while

ensuring transparency and accountability.

NFTs are created by “minting” digital

content on a blockchain. Minting involves

uploading and having other computers

verify and time-stamp the content, loca-

tion, and originator of digital informa-

tion, and all subsequent transactions

are recorded on a digital ledger that is

distributed across a network of comput-

ers. Redundancy, along with the compu-

tational difficulty and processing energy

required, makes it difficult to tamper with

the transaction record. Blockchains, some-

times referred to as “trustless” systems,

provide a verifiable infrastructure to man-

age digital assets.

There are many situations in which the

distinctive features of NFTs could pro-

vide potential advantages over, and help

address gaps in, the existing HIE sys-

tem. For example, in 2020, the US Office

of the National Coordinator of Health

Information Technology (ONC) issued

a rule as part of the 21st Century Cures

Act that provided a technical basis for

an individual to assert the right of access

to a computable version of their medi-

cal record. By the end of 2022, all certi-

fied PHI technology will need to support

Substitutable Medical Applications and

Reusable Technologies (SMART) on the

Fast Healthcare Interoperability Resources

(FHIR) application programming inter-

face (API), which allows third-party apps

to connect with and request data from

health care provider electronic health

records (EHRs) ( 3 ). To ease the cumber-

some manual process of extracting data

from EHRs, the SMART on FHIR API sup-

ports an ecosystem of patient-facing apps

that could serve to enhance patient agency

in directing sharing of their data. Although

this represents a leap forward in engaging

patients in HIE, the largely commercial na-

ture of the marketplace may still serve to

undermine trust and create disincentives

for sharing PHI.

POTENTIAL BENEFITS

Automating data access and control

At least two scenarios could be realized

under an NFT or NFT-like framework for

personal control of health data. In the

first, PHI would be uploaded (or “minted”)

as a distinct, “original” version, with the

generator or custodian (e.g., EHR com-

pany, hospital, biobank, etc.) required to

register each new datum (e.g., diagnosis

of illness, prescription) on a public block-

chain. The encrypted PHI would only be

accessible to those given explicit permis-

sion in a smart contract. Mann et al. ( 4 )

proposed such a scheme using blockchain

and smart contracts to “prosent” (proac-

tively consent) pseudonymously to data

release or exchange for certain uses. In

this way, patients could specify in advance

with whom they agree to share data with-

out needing to consent to every transac-

tion, enabling greater patient control and

more timely and efficient data exchanges.

In cases where patients might prosent to

the sale of their PHI or to participation in

clinical trials offering compensation, the

smart contract could allow for automated

distribution of funds to the patient. If the

patient ever wanted to modify the terms of

the contract, each change of terms would

be immutably stored as distinct, time-

stamped events in the blockchain ledger.

In a second, but not mutually exclusive,

scenario, a patient’s data could stay right

where it is (for example, a hospital data-

base or on a patient’s phone), and the smart

contract could “push” an algorithm to sum-

marize or analyze the data with automated

permission from a smart contract. This ap-

proach is highly compatible with federated

learning approaches that train machine-

learning algorithms on multiple local data-

sets without explicitly exchanging data

samples or compiling them into a central-

ized server. Smart contracts could help to

realize data privacy and security goals of

federated learning, recently credited as be-

ing the “future of digital heath” ( 5 ).

DATA

How NFTs could transform

health information exchange

Can patients regain control over their health information?

(^1) Baylor College of Medicine, Houston, TX, USA. (^2) Computational Health Informatics Program, Boston Children’s Hospital, Boston, MA, USA. (^3) Department of Biomedical Informatics, Harvard
Medical School, Boston, MA, USA.^4 Centre for Advanced Studies in Biomedical Innovation Law (CeBIL), University of Copenhagen, Denmark.^5 Harvard Law School, Boston, MA, USA.^6 School of
Social Sciences and Technology, Technical University of Munich, Munich, Germany. Email: [email protected]
POLICY FORUM
INSIGHTS