DOKCHAIN: INTELLIGENT AUTOMATION IN HEALTHCARE

TRANSACTION PROCESSING

W. Bryan Smith, PhD; Chief Scientist, PokitDok; [email protected]

Contents
Summary 1
Motivation 1
Cost Savings Estimates 2
New Market Opportunities 4
DokChain Alliance 6
Protocol 6
Use Cases 6
DokChain Currency: Cure Coin 10
Cure Regulation and Minting 12
Closing 15

Summary
DokChain is a distributed network of transaction processors operating on financial and
clinical data in the healthcare industry. Our goal is to deploy distributed ledger technology
across a broad range of industry participants to bring intelligent and dynamic automation to
four core use cases that span healthcare encounters: context-relevant identity management;
autonomous transaction validation and processing; prior authorizations; and event-driven
supply chain management. The initial implementationi employs proof of elapsed time (PoET)
consensus among hosting nodes to generate an encrypted, immutable log of every transaction
in the system, using on-chain pointers to an off-chain, distributed file system for data storage,
access, and analysis. The result of our implementation is a new kind of healthcare economy,
in which data and services are quantifiable and exchangeable, with strong guarantees around
both the security and privacy of sensitive information as well as the longitudinal auditability
of transaction history.
Motivation
The healthcare industry is encumbered by operational inefficiencies and prone to costly
errors, resulting in tremendous loss of both financial as well as human capital. While a matter
of global importance, the issue is particularly acute in the United States, where the national
expenditure on healthcare is disproportionately large (Figure 1). Note that moving the U.S.
to the nearest contour in the distribution would require more than half a trillion dollars
in cost reductions. Among the many reasons often given for the waste seen in healthcare
Date: September 26, 2017.
i
implementation details, including available consensus algorithms, minting protocol, etc are subject to change
1
2 W. Bryan Smith, PhD; Chief Scientist, PokitDok; [email protected]

spending, it is clear that poor information access and management across the ecosystem is
a common factor. The current state of affairs is, however, not for lack of trying. Indeed,
a tremendous amount of regulatory and policy-driven effort has been spent attempting to
solve the issues that plague the industry. The list of such initiatives includes “meaningful
use” mandates around legacy technologies, payment incentive models based on “value” versus
“volume,” and large scale insurance market intervention (i.e. the Affordable Care Act, ACA).
However, by many measures, these efforts have largely failed to deliver on their respective
objectives. While it is clear that these failures are due in part to behaviorally complex (and
often misaligned) incentives throughout the existing system, we believe that a truly connected
and intelligent information infrastructure will cure a lot of what ails healthcare.

Figure 1. Distribution of global healthcare spend as a percent of GDP (vertical axis), by

GDP per capita. Each colored circle represents a country: size is proportional to population;
color shows life expectancy at birth, as indicated by the scale at right. The three oval contours
represent 1, 2, and 3 standard deviations on the covariance distribution. An interactive
version is available here. Data from worldbank.org.

Cost Savings Estimates

In a very real sense, the simple statement “healthcare in the U.S. is a complete clus-
terf#@k” is sufficient motivation for creating DokChain. However, to try and more formally
frame the potential value of the DokChain network, we begin with an estimate of the ef-
ficiency gains the system is expected to realize across the healthcare industry. Healthcare
spending is rapidly approaching 20% of the U.S. GDP, with the average healthcare spend
globally at only about 6.7% of GDP. Yes, you read that correctly: the US spends nearly
three times the global average on healthcare, even after adjusting for population size and
 DOKCHAIN: INTELLIGENT AUTOMATION IN HEALTHCARE TRANSACTION PROCESSING 3

economic productivity, often with inferior outcomes. What’s more, the U.S. National Health-
care Expenditure report from the Centers for Medicare and Medicaid Services (CMS) projects
that healthcare spend will grow 1.3% faster than GDP over the next decade.
Although these statistics are alarming, it is difficult to define exactly how large healthcare
spending should be, given the combined size and relative wealth of the US population. Some
argue that perhaps healthcare spend should be much larger in the U.S. compared to other
countries for myriad reasons, including the extensive R&D activities and technology-heavy
processes that comprise significant aspects of the U.S. ecosystem. Such debates notwith-
standing, it is our position that there remain obvious sources of waste in the U.S. healthcare
system. Depending on which sources of waste one intends to tackle, estimates of the total
potential cost savings range from the low hundreds of billions to a trillion dollars or more.
We focus here on three categories of waste for our initial estimates: manual transaction
processing; care coordination; and administrative complexity ii.
Eliminating Manual Transactions. According to the most recent statistics from the Council
for Affordable Quality Healthcare (CAQH Index report, 2016), while the industry has seen
widespread adoption of electronic transaction systems over the last 20 years, manual transac-
tion processes still account for roughly $10 billioniii in unnecessary spend. The ‘gate keeper’
transaction of this set, eligibility and benefit verification, accounts for more than half of the
savings potential, as summarized briefly in Figure 2.

Figure 2. CAQH estimates of savings potential due to eradicating manual transaction

processes.

Close the Gaps in Care Coordination. The waste associated with failures in care coordina-
tion is attributed to patients “falling through the cracks” in the fragmented care delivery
processes common throughout the system today. This typically manifests as a loss of con-
tinuity of both patient data and provider assessment as a consumer moves from a primary
care provider to one specialist and then the next throughout a complex episode of care. The
ii
The text in this section borrows heavily from the 2016 CAQH Index and Berwick and Hackbarth, 2012
iii
All dollar values are adjusted to 2017 levels assuming 5% annual growth in healthcare spend
4 W. Bryan Smith, PhD; Chief Scientist, PokitDok; [email protected]

consequences include increased readmissions, declines in functional status, and increased de-
pendency, especially among the chronically ill. For example, 30-day readmission rates among
fee-for-service Medicare recipients have been estimated at about 20%, with roughly 75% of
those determined to be “potentially avoidable.” In aggregate, the estimated savings to be
realized by eliminating this class of waste range from $32 to $58 billion.
Automate Away Administrative Complexity. The result of governments, accreditation agen-
cies, payers, and others entities actively creating rules and processes that actually decrease
efficiency is among the largest sources of waste to be addressed. Administrative complexity is
the stuff that makes the everyday consumer and provider experiences in the U.S. healthcare
system as terrible as they tend to be. This includes such factors as providers being forced
to use electronic health record (EHR) software that conflicts with their preferred workflows,
consumers not understanding out of pocket contributions at time of service, and official re-
quirements to conduct business via legacy technologies that stifle innovation. It is estimated
that solving the problems related to administrative complexity will reduce waste by $137 to
$500 billion.
In summary, by addressing only these three classes of waste in the existing system, we
estimate a total addressable market at somewhere between $180 and $570 billion. The six
types of waste, adapted from the Berwick and Hackbarth paper, are shown in Figure 3 for
reference.

Figure 3. CAQH estimates of savings potential due to eradicating manual transaction

processes.

New Market Opportunities

While it is clear that there are massive efficiency gains to be made, effectively and safely
realizing those gains remains a significant economic challenge. To quote Berwick and Hack-
barth:
The potential economic dislocations are severe and require mitigation through
careful transition strategies.
And so, while we continue working diligently with corporate, legal, and regulatory partners to
ensure that appropriate transition strategies are being pursued, we have also identified some
key opportunities that exist now, and that will emerge as a result of large-scale distributed
network adoption in healthcare. A few of these opportunities are listed briefly here, some of
which are discussed in detail in the Use Cases section of this paper:
 DOKCHAIN: INTELLIGENT AUTOMATION IN HEALTHCARE TRANSACTION PROCESSING 5

Use Case Prioritization and Governance Definition. Founding members of the DokChain
Alliance will have a unique role in defining the specifics of initial use cases, and will work
together to determine the optimal governance structure for overseeing the development and
operation of the Network. Because the Network will only be open to Founders during the
initial period of testing and optimization, the ability to directly have a large scale impact on
the outcome of the system will be the sole privilege of these first adopters. While provisions
for modifying the Network have been incorporated into the protocol, such modifications will
be minor once the initial, closed period of optimization has lapsed.

Information Asset Monetization. As is often described in other blockchain implementations,

one of the most intriguing opportunities to be explored is the different ways information as-
sets can be defined and monetized. One example of this, in which an individual user’s identity
provides monetization opportunities both for the individual as well as the participating en-
tities who provide identifying information from their respective domains, is described in our
identity management use case below. There will also be significant value creation in the
eventual emergence of smart contract markets, wherein users will bid on the right to use
different contracts to conduct their transactions, relying on full transparency around the rel-
evant attributes of the various contracts being used to set a fair price. Other information
asset classes and mechanisms for monetization are left as an exercise for the reader.

Outcomes Based Optimization. An obvious feature of an integrated health information net-

work as described here is that the potential for outcomes-based optimization may finally be
realized. Not only will all the relevant data to conduct such optimizations be available at
a population level, but true personalized medicine approaches at an individual level can be
approached with confidence, knowing that the user identities and the data surrounding their
episodes of care have been verified and reliably logged in the system via distributed consen-
sus. The industry-horizontal nature of the DokChain network will enable outcomes based
optimization in a much broader business context as well. One such example, event-driven
supply chain optimization, is defined as a primary use case below.

Consumer-Driven Healthcare Markets. As consumers take charge of their personal health

information, and begin to understand the value of that information as an asset, having access
to a platform where that value can be exchanged for goods and services provides tremendous
opportunity for the emergence of true markets for health services. As discussed elsewhere,
rising plan deductibles and a growing consumer awareness of varying levels of quality of care
are already providing massive incentives to drive these markets. The DokChain network
provides the information access and transparency required in order to capitalize on those
existing incentives. We see these markets evolving in such a way that the very nature of
healthcare service payment will change dramatically, shifting from the existing model based
on premiums and deductibles to one where “payers” make real-time bids to pay for services
on behalf of consumers in exchange for interest payments over time.

Cryptocurrency Sale. The initial role of the in-Network cryptocurrency is to function as a

B2B token used as a means of “paying” for transactions and rewarding Founding Members
for their efforts during the initial, closed period of Network optimization. Once the Network
is open to a larger group of participants, however, the currency accrued by Founders may
be sold as part of the onboarding process for new network participants, or traded for other
currencies on a public exchange.
6 W. Bryan Smith, PhD; Chief Scientist, PokitDok; [email protected]

DokChain Alliance
The DokChain Alliance was established in early 2016 with the intent of creating a fully
interconnected health information economy, focused on the security, efficiency, verac-
ity, and transparency of information transfer across the network. The Alliance has pulled
together an unprecedented breadth of industry representatives, spanning not only the entire
healthcare ecosystem, but also financial services, consumer electronics and software, hard-
ware manufacturers, credit bureaus, third party data providers, and others. The group has
convened quarterly to discuss the Network protocol, use case prioritization, and governance,
which are summarized here.

Protocol
The initial implementation of DokChain is a private, permissioned, distributed ledger of
transactions, using on-chain references to an off-chain distributed file system to store the
actual data payloads associated with each transaction. Currently the data store is imple-
mented using IPFS/IPLD, though the system can flexibly accommodate any similar content-
addressed storage model. This approach trivially addresses content scalability and data
privacy concerns, as the distributed ledger is not used directly as a data storage mechanism.
For the testnet implementation, transactions are processed via execution of smart contracts
by a Proof of Elapsed Time (PoET) selected leader. Every contract — including the network
protocol, governance structure, membership agreements, et cetera — exists as a signed ob-
ject on the DokChain ledger, removing any ambiguity as to the authenticity and reliability
of these objects.
At a high level of abstraction, a transaction contains the following required elements:
{‘sender’:DKID, ‘recipient’:DKID, ‘payload’:blob, ‘contract’:address, ‘coin’:float}

The processes governing issuance and authentication of each identity (DKID) are described
in detail in the Use Case section below. The payload of a transaction as it is posted to
the network may contain the data upon which the designated smart contract will operate,
although only a hash of the data is written to the distributed ledger. Figure 4 outlines a
high level description of the protocol as currently defined. Additional details of the current
protocol have been previously described in a blog post by Ted Tanner, PokitDok co-founder
and CTO.

Use Cases
A fundamental consideration driving the initial DokChain use cases is a strong notion
of information asset ownership and rights management. In the DokChain Network,
digital asset ownership is conferred at time of issuance, based entirely on the identities (i.e.
pubkeys) of the transacting parties, with partial asset ownership defined explicitly in the
parameters of the contract being used. For example, protected health information (PHI) is
owned by the individual consumer from whom the data are collected, and the components
of the data to be shared with a provider or payer are defined explicitly, both at the time of
initially writing the data to the system, but also upon subsequent access for use in analysis
or auditing. It is our expectation that this more “appropriate” information asset ownership
model will incentivize a shift toward a more central role for consumer-driven digital asset
management in optimizing care coordination and delivery.
 DOKCHAIN: INTELLIGENT AUTOMATION IN HEALTHCARE TRANSACTION PROCESSING 7

Figure 4. DokChain Pseudocode

Contextually Relevant Identity Management Protocol. The cornerstone of the Dok-

Chain network is a multi-party, contextually-relevant identity management protocol (CRIMP)
that will ultimately incorporate active key management and key recovery functionality as im-
plicit attributes of each identity in the system. Multi-party is as expected: multiple third
party entities in the network can confirm the identity of an individual user. Contextual
relevance here refers to the fact that different elements of a user identity may be needed in
different contexts. A typical example given is the context of ordering an alcoholic beverage:
the only identifying information the vendor needs to know is that the person requesting the
drink is older than what the local age limit requires. Yet today, the customer surrenders an
identification card that contains their full name, address, weight, height, and other pieces of
sensitive information that have no relevance to their immediate request. We have developed
an identity management solution that, initially, allows the user to decide what they are will-
ing to share of their personal record based on the context of each encounter. As the system
evolves, intelligent agents will evolve in parallel, learning how to manage the complexity of
this information on behalf of the human user. The system is described in a high level of de-
tail here, with implementation specifics made available in the DokChain code base to hosting
nodes as appropriate.
8 W. Bryan Smith, PhD; Chief Scientist, PokitDok; [email protected]

(a) Multiparty Identity (b) Contextual Identity

Figure 5. A user identity (A) is defined as the entirety of all information known about
them from many different sources. Subsets of identity information are combined together(B)
to match different contexts.

An identity in the DokChain Network comprises a set of key-pairs for each individual,
where each key-pair can be used to unlock a distinct subset of information to be used in
different transaction contexts. Given the extent to which consumer engagement is a driving
factor in Network adoption, it is essential that users of the system have access to the most
robust identity verification and validation system in existence, without having to remember
any complex passphrases. We also believe users should not have to concern themselves with
the potential consequences of losing their private keys.
To that end, the DokChain identity management system takes full advantage of the fact
that, at least in the digital world, you truly are what you do. The entirety of your digital inter-
actions, not just within any single industry vertical, creates the strongest possible signature of
who you are as an individual. Medical records, e-commerce clickstreams, government-issued
identity cards, biometrics: anything that increases confidence in determining the uniqueness
of an individual can be incorporated. Our solution leverages the horizontal nature of the
DokChain Alliance directly, in that the protocol integrates information about users from
myriad third party sources (see Figure 5). The power of our approach becomes increasingly
obvious as compromised legacy systems continue to leak personal information about users
worldwide. We assume that at some point strong identifiers, as well as their links to sensitive
account numbers, will effectively be public knowledge as a consequence of these hacks. Our
system provides mechanisms to ensure identity verification and proofing even in this scenario.
We are pursuing a variation of a proof-of-work consensus component in the validation of
each identity, whereby the private key is partitioned among a group of user-defined trusted
parties, and according to a process such as Shamir’s Secret Sharing Scheme, the key is
regenerated from a subset of those partitions. This mechanism is initially being used for key
 DOKCHAIN: INTELLIGENT AUTOMATION IN HEALTHCARE TRANSACTION PROCESSING 9

recovery in the case of accidental loss, but will eventually function as an implicit component
of identity verification that occurs each time an identity is invoked in a transaction. The
work done is a combination of signing, encrypting, and reassembling the private key, as well
as moving the requisite subset of key partitions across the network in real time for each
validation and verification event.
The consensus identity that emerges from this process is given a confidence score that is
interpreted as a multi-party attestation that the attributes in a given user vector accurately
and uniquely identify the individual. Transacting parties can confirm one another’s iden-
tity via a consensus score alone as a base case, or may choose to request access to additional
identifying information, including the identities of the validating parties, as their preferences,
and budget, dictate. In addition to providing unprecedented identity security, this creates a
completely new information asset monetization channel as both users and their ‘validating
sponsors’ can expect to be compensated for providing such detailed, verifiable identity infor-
mation. Placeholder examples of how such scores may be computed are shown for reference
below.
Define a user vector, u, an m-dimensional vector of identity query attributes, q ⊆ u,
and a set of N identity service providers, p. Then a consensus score, Sqp , representing
the confidence over p that q corresponds to u, may be generally stated as
1 X
N
Sqp = F n(pi , q)
N i=1
Where, in a naive implementation, F n(·) simply represents each identity provider’s confidence
that q = u. Alternatively, we may consider a model that computes a weighted aggregate
over each element of q:
1 X
N
Sqp = P r(q | pi ) · wT
N i=1
where w is an m-dimensional weight vector (i.e. some information-theoretic prior) over the
elements of q that is updated over time, and as a function of the (inferred) context of the
query.

Efficient Prior Authorization and Referral. One of the more costly, complex, and to
date overwhelmingly manual transactions in the healthcare industry is the prior authorization
and referral process (of which the X12 278 transaction is a component). Today, it is estimated
that these transactions cost the payers several tens of dollars per transaction, a cost that can
be cut in half with basic process improvement and automation. We are developing smart
contract-based automation workflows that may reduce costs by an additional factor of 5 or
10.

Autonomous Claims Adjudication. Medical claims processing is a notoriously compli-

cated process, with coding, billing and reimbursement processes representing significant per-
centages of the business overhead for providers. Somewhat spurred on by antiquated regu-
latory policies, the system today sees clearinghouses, third party administrators, and other
intermediaries operating under the assumption of risk management on behalf of their respec-
tive customers. As things have evolved, however, these intermediaries are not proving their
value proposition, tending instead to increase the complexity and reduce the transparency of
conducting business.
10 W. Bryan Smith, PhD; Chief Scientist, PokitDok; [email protected]

To some extent, this issue is currently being addressed in the form of auto-adjudication
software and protocols. Auto-adjudication refers to computational claim processing, without
any human role in reviewing or paying the claim. Great strides have been made in this
arena, with United Healthcare (the largest private health carrier in the U.S.) claiming that
86% of electronically submitted claims to their systems are auto-adjudicated, while only 70%
of manually submitted claims achieve this status.
While this is a solid start, it is worth noting that 14% of claims at United Healthcare is
still a massive volume of claims, and that United is far ahead of other major carriers in this
regard. In addition, auto-adjudication rule sets and the software used to manage them often
require human intervention, which is simply deferring the manual portion of the task to a
different step in the process.
The DokChain Alliance members have agreed that an artificially intelligent, smart contract-
mediated claims adjudication protocol is a highly desirable use case to pursue. Such a system
will allow for near-realtime adjudication and remittance of submitted claims, saving time and
money for payers and providers, and reducing both the uncertainty as well as latency of out
of pocket expenses for consumers. Further, because the system will be running within the
DokChain distributed Network, updated rules and optimal payment schemes can be learned
in real-time over a broad range of plans and health systems.

Event-Driven Supply Chain Management. In the fragmented care delivery systems

today, consumers are often shuffled from one office to another, sometimes carrying their own
diagnostic test results (e.g. MRI images) along with them by hand, being forced to drive to
a pharmacy or equipment supply house that is far away from where they’ve been seen by
their provider, only to find that what they need is out of stock or not exactly what is needed
for their specific condition. Imagine instead the benefits of a situation where everything a
consumer needs after a doctor’s office or hospital visit is already at their door by the time they
arrive home. This improves not only convenience and consumer satisfaction, but adherence
to prescribed treatments will increase dramatically as well.
Making this situation a reality is the goal of this DokChain use case. Again leveraging
the horizontal nature of the distributed network, a consumer’s preferred retail vendor can
be granted access to their Personal Health Record to know when they have been seen by a
provider (an ‘event’), as much detail about the nature of that encounter as the consumer
wishes to share, and therefore what drugs and other consumables as well as durable medical
goods they need to continue treatment at home.

DokChain Currency: Cure Coin

We are defining an internal currency, herein denoted as Cure, as opposed to using Bitcoin
or Ethereum, in an effort to isolate the currency value from market forces outside of the
DokChain network. For example, the exchange rate of Bitcoin with respect to other currencies
is extremely volatile, subject to regulatory statements by governments, and protocol decisions
of a small group of developers and/or miners, none of which have any relevance to the
DokChain Network. While the currency may be listed and traded on an exchange at a later
date, thereby creating additional monetization strategies, we do not intend to focus on this
aspect of the currency in our MVP implementation.
The role of a virtual currency in the DokChain Network is three-fold:
(1) To signal transaction value in ongoing operations of the Network
 DOKCHAIN: INTELLIGENT AUTOMATION IN HEALTHCARE TRANSACTION PROCESSING 11

(2) A method for seeding the Network with relevant assets

(3) To incentivize Founding Members to participate in optimizing the Network
In order to bootstrap the Network, we are initially conducting a private, B2B token sale to
ten founding partners selected from participants in the DokChain Alliance. Each founder
will contribute a minimum of $1M in the form of some combination of cash, press, and other
assets. While minimum and maximum asset contributions are negotiated on an individual
basis, a $250k minimum cash contribution is a strict requirement. A non-exhaustive example
of assets considered is shown in Figure 6. Asset exchange for Cure in a future public coin
sale will be evaluated in an information theoretic framework, and priced according to the
relative increase in system functionality and/or information content provided by that asset
at that point in time. As with all other governance components of DokChain, such exchange
contracts will be encoded on-chain for reference.

Figure 6. CAQH estimates of savings potential due to eradicating manual transaction

processes.

DokChain members are granted access to the Network by purchasing Cure to be used in
transaction valuations. In addition to coins awarded for transaction processing as described
in the protocol above, each hosting node is issued a stream of coins at a variable rate that is
determined by that node’s participation in the Network. “Participation“ is defined by being
a source of transaction activity, processing transactions, validating user identities, and other
general functional attributes of the system.
In our testnet implementation, there is no distinction between nodes operating as trans-
action processors and nodes operating as validators, though a future implementation is in-
tended to accommodate such a distinction. As we move toward such a system, the sum of
the available coins in a block of transactions will be differentially shared among the processor
12 W. Bryan Smith, PhD; Chief Scientist, PokitDok; [email protected]

(PoET-selected leader) and any participating validators, with the processor being compen-
sated slightly more than each of the validators. A trivial differential payout scheme is included
here to illustrate a simple version of how such payouts could be easily managed:
Let N be the number of validators participating in mining any given block, rs the sum of
coin reward associated with the block, and mp the multiple over each validator payout that is
paid to the transaction processor. Then the reward for the processor, rp , and each validator,
rv , can be computed as:
mp rs rs − rp
rp = and rv =
N + mp N

The processor bonus, mp , can vary randomly according a known distribution whose param-
eters are determined as a function of the “Network Health.” Ideally the payout bonus for
processors asymptotically approaches 1.0 over time, as the tasks of processing and validating
approach parity due to maturation of a smart contract marketplace within the system.

Cure Regulation and Minting

The DokChain system requires an in-network currency, herein referred to as cure, in order
for transactions to be processed: some non-zero amount of cure is posted with, and spent
in the processing of, every transaction. Cure can be purchased directly via fiat currencies on
an exchange, acquired via like-kind asset exchange according to on-chain asset valuation
contracts, or earned over time by hosting a node in the DokChain Network. Here we
describe in detail the guiding principles underlying the rate at which cure is issued to hosting
nodes over time.
Because the DokChain protocol does not rely on proof-of-work mining, we do not define a
priori the total amount of currency that will be available in the future, but instead define a
monetary policy by which currency is minted. The currency issuance at any point in time will
be a function of three metrics: (1) the performance of nodes operating the network; (2) the
performance of the network as a whole; and (3) a model of external economic factors. The
goal is to modulate the rate of cure minting to reflect the value being created and captured
by the operation of DokChain, at the scale of both individual nodes as well as the entire
Network. We further explicitly regulate currency flow at a rate that is inversely proportional
to the derivative of an external economic factors model. This has the effect of incentivizing
the purchase of cure when the larger-scale economy is in decline, but induces scarcity in
times of strong economic growth outside the system.

Sources of Supply.

Cure Pre-Sale. All cure coin is issued via a contract included in the initialization of the
DokChain Network, modifiable only via a four-fifths majority vote among Network mem-
bers. The initial 1e9 cure coins, herein denoted as C0 , are distributed to the initial 10
DokChain Alliance Founding Members in exchange for the assets they each contribute to the
initialization of the Network. This is designated as the cure pre-sale. The current proposed
allocation in the pre-sale is 35.2% to PokitDok, Inc, with the remaining 64.8% split equally
among the other 9 Founding Members, for a stake of 7.2% per Founder.
 DOKCHAIN: INTELLIGENT AUTOMATION IN HEALTHCARE TRANSACTION PROCESSING 13

Transaction Fees. The smart contracts that broker each transaction include provisions for
cure disbursement, including a minimum required cure quantity to be paid to the hosting
nodes of the DokChain Network. In the simplest version of disbursement, the sum of this
currency is split equally among all transaction processors operating the Network. This dis-
bursement model illustrates at least one reason we are considering a consensus algorithm that
does not require all processors to validate every block: as the number of hosts grows, the
share of cure each node expects to obtain via this supply channel will decrease, potentially
dis-incentivizing network expansion.
Cure ‘Minting’. In addition to the initial currency allocated at the time of joining the net-
work, all hosting nodes will accrue cure according to a monetary policy or minting function,
which is described in detail below. For the time being, and until it is determined whether it is
practical (or desirable) to distinguish between transaction processors and transaction
validators, as well as whether we pursue a quorum-style consensus algorithm, it is assumed
that the total amount of cure minted at any point in time is disbursed equally among the
hosting nodes.
Node Performance Score. Given that the hosting nodes are by definition the entities
operating the network, there is clearly a relationship between the global network score and
the node performance score, but we parameterize them separately. The normalized node
operations score, Snd ∈ (0, 1], is a measure of how well the nodes are performing the tasks
expected of them, such as the percentage of transactions in a block that they participated in
processing/validating, etc. We expect nodes to behave properly, if not ideally, even at time
zero, when the network is initialized. Snd will be initialized at a fairly high value, increasing
at an unknown rate over time, currently modeled as a linear ramp from 0.67 to 1.0 over a
ten year period (Figure 7, green).
Snet
1.0
Snod

0.8

0.6

0.4

0.2

0.0

0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5 6.0 6.5 7.0 7.5 8.0 8.5 9.0 9.5 10.0 10.5

Figure 7. Examples of Snt and Snd used in this paper.

Network Operations Score. The currency minting rate needs to reflect the overall func-
tion of the DokChain Network in ways that are easily understood by all participants and
interested parties. The normalized network operations score, Snt ∈ (0, 1], captures aspects
of the function of the network as a whole. This metric is initially dominated by global net-
work statistics such as the total transaction volume processed by the network, normalized to
the 20B transactions per year in the healthcare ecosystem as estimated by CAQH. The rate
of cure minting is relatively large when this value is small, decreasing to zero as the ratio
14 W. Bryan Smith, PhD; Chief Scientist, PokitDok; [email protected]

Snd /Snt approaches 1. The network will be initialized with Snt set to 0.01, a value that we
expect will increase at an unknown rate over time, currently modeled as a gamma cumulative
distribution function ramping from 0.01 to 0.97 over a ten year period (Figure 7, black). We
define a requirement that Snt < Snd , which makes intuitive sense given that the network
performance score should not be higher than the average score of the nodes operating the
network.
External Economic Factors Score. We think it is important to explicitly include a term
in the cure minting function that takes into account external economic factors. This is a vital
component to the idea that the amount of cure being minted is in some way proportional
to both the value being created by the system (the node and network scores) but also the
“captured value” of, for example, the eroding market share and/or reduced economic output
of entities that are disintermediated by the success of the DokChain Network. Candidate
features in such a model include the labor participation rate, the equity prices of companies
that currently operate as intermediaries, etc. For the bootstrap analysis included in this
paper, the daily diff computed over 10 years worth of historical NASDAQ closing price data
was modeled with a Laplacian distribution and randomly sampled to provide a working
example of an Sec function, as shown in Figure 8.

1.0
NASDAQ Diff (10yrs) NASDAQ Diff (10yrs)
1.0
Laplacian Model Laplacian Model

0.8
0.8

0.6 0.6

0.4
0.4

0.2
0.2

0.0
0.0
−6 −4 −2 0 2 4 −4 −3 −2 −1 0 1 2 3 4

(a) Normalized Histograms (b) Cumulative Distribution Functions

Figure 8. Distributions of the NASDAQ daily diff data and a maximum likelihood-
estimated Laplacian model.

Putting it all together. As shown in Figure 9, we can simply compute the amount of cure
to be minted at any point in time t as
 
Sndt
Ct = C0 − 1 e−dSect
Sntt
Another possibility to consider is that we define an ideal currency accumulation function
with predetermined values at any point in time, CI , and then mint cure according to the
difference between the ideal cure and the sum of currency influx via direct purchase and
like-kind asset exchange, denoted as Cx :
Ct = CI − Cx
To actually be implemented, however, this approach requires definition of a max currency
amount, a path we currently do not favor.
 DOKCHAIN: INTELLIGENT AUTOMATION IN HEALTHCARE TRANSACTION PROCESSING 15

   
Snd Snd
C0 Snt − 1 e−dSect C0 Snt − 1 e−dSec
×1013
median
mean 0.8
16
cumulative sum of currency (log10)

15 0.6

cumulative sum of currency

14

0.4

13

0.2
12

11
0.0

0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5 6.0 6.5 7.0 7.5 8.0 8.5 9.0 9.5 10.0 10.5 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5 6.0 6.5 7.0 7.5 8.0 8.5 9.0 9.5 10.0 10.5
time (years) time (years)

(a) Time included in the exponent. (b) Time excluded from the exponent.

Figure 9. Two versions of cure accumulation. Dashed lines represent 95% confidence
intervals based on 10k random samples from the Sec Laplacian model.

Closing
Healthcare in the U.S. is a $3.5 trillion industry, growing at a rate expected to exceed GDP
growth for at least the next decade. Much of the waste in the industry is directly attributable
to intermediaries such as clearinghouses and outdated ‘technology’ vendors, making it ripe for
blockchain-based disruption. Here we have described a distributed network implementation
that eliminates these irrelevant intermediaries and empowers consumers, providers, payers,
and other interested parties to create a new healthcare information economy. Our goal is not
to change healthcare, but to replace it completely with something that we know will work
better. Readers interested in participating in the DokChain Alliance can visit DokChain.com
to learn more.