This is an open-access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in the Journal of Medical Internet Research, is properly cited. The complete bibliographic information, a link to the original publication on https://www.jmir.org/, as well as this copyright and license information must be included.
Despite the fact that the adoption rate of electronic health records has increased dramatically among high-income nations, it is still difficult to properly disseminate personal health records. Token economy, through blockchain smart contracts, can better distribute personal health records by providing incentives to patients. However, there have been very few studies regarding the particular factors that should be considered when designing incentive mechanisms in blockchain.
The aim of this paper is to provide 2 new mathematical models of token economy in real-world scenarios on health care blockchain platforms.
First, roles were set for the health care blockchain platform and its token flow. Second, 2 scenarios were introduced: collecting life-log data for an incentive program at a life insurance company to motivate customers to exercise more and recruiting participants for clinical trials of anticancer drugs. In our 2 scenarios, we assumed that there were 3 stakeholders: participants, data recipients (companies), and data providers (health care organizations). We also assumed that the incentives are initially paid out to participants by data recipients, who are focused on minimizing economic and time costs by adapting mechanism design. This concept can be seen as a part of game theory, since the willingness-to-pay of data recipients is important in maintaining the blockchain token economy. In both scenarios, the recruiting company can change the expected recruitment time and number of participants. Suppose a company considers the recruitment time to be more important than the number of participants and rewards. In that case, the company can increase the time weight and adjust cost. When the reward parameter is fixed, the corresponding expected recruitment time can be obtained. Among the reward and time pairs, the pair that minimizes the company’s cost was chosen. Finally, the optimized results were compared with the simulations and analyzed accordingly.
To minimize the company’s costs, reward–time pairs were first collected. It was observed that the expected recruitment time decreased as rewards grew, while the rewards decreased as time cost grew. Therefore, the cost was represented by a convex curve, which made it possible to obtain a minimum—an optimal point—for both scenarios. Through sensitivity analysis, we observed that, as the time weight increased, the optimized reward increased, while the optimized time decreased. Moreover, as the number of participants increased, the optimization reward and time also increased.
In this study, we were able to model the incentive mechanism of blockchain based on a mechanism design that recruits participants through a health care blockchain platform. This study presents a basic approach to incentive modeling in personal health records, demonstrating how health care organizations and funding companies can motivate one another to join the platform.
Precision medicine aims to define diseases at a higher resolution using genomic data, electronic health records, and life-log data by providing new therapies to each targeted subgroup [
As such, health care blockchain has been implemented to solve these various problems. Blockchain is known as a distributed ledger technology—it records given information into small chunks of data sets called
For the 2 main advantages of blockchain—namely auditability and identification—many proof-of-concept studies have been conducted [
To examine the effectiveness of incentivization, one study [
Mathematical studies have been performed to model incentivization for blockchain. For decentralization, designing protocols that make it difficult to tamper with transactions is essential, and a mathematical approach to designing blockchain protocol that achieves secure information exchange has been examined [
Why is a new theoretical foundation needed for health care blockchain cryptoeconomics? The purpose of cryptoeconomics is to create internet services, but why do we need new theories and methodologies? In short, it is because the design of decentralized networks is completely different from that of traditional business/service planning. Existing business planning is given certain rules (market conditions) and aims to ensure that companies make the best choices to maximize profits; however, the design of a decentralized network aims for the opposite. Assuming that each entity acts strategically and selfishly, the design of a decentralized network makes rules to achieve desired results.
Despite the pressing need for further research on the cryptoeconomics of health care blockchain, there has been a lack of research on token economy design to induce patient engagement using health care blockchain automated incentive programs. Thus, we aimed to explore parameters to consider when designing incentive programs that can be embedded in the smart contract of blockchain by using experiments based on 2 representative virtual scenarios with mechanism design.
Blockchain-based networks have no central principal. Therefore, we need a system that uses tokens (as the medium) and market principles to help individuals grow the network even if they act in pursuit of their own interests. This is called the
To design the token economy, we considered 2 well-known economic theories: game theory and mechanism design. In game theory, existing business planning is given rules (market conditions) and aims to ensure that companies make the best choices to maximize their profits. The theory examines the best strategy in a given game. It explains how to draw conclusions in situations in which several rational decision-makers compete to win over one another. It is called a
Relationship between game theory and mechanism design theory.
However, the design of a decentralized network is the reverse. Mechanism design uses an engineering approach in which designers act rationally toward desired goals in a strategic environment by applying economic mechanisms and incentives to design strategies. It is also called
With the principle of mechanism design, we can reduce cost caused by trial and error and provide a platform for solving real-life problems as a theoretical model. Combined with blockchain, mechanism design has been adopted in various types of research. The auction is a representative example of mechanism design in a blockchain environment with randomness and information disclosure [
When recruiting participants to a vitality program in the real world, a life insurance company provides incentives to those involved in the program. This mechanism is easily applied to the blockchain using smart contracts. In addition, the blockchain protects personal health information and is able to conduct universal recruitment through the app; therefore, efficient and safe recruitment of participants can be carried out through health care blockchain.
Meanwhile, the compensation that companies provide to participants on the blockchain is cryptocurrency. Therefore, it is essential to discuss how to calculate the value of this cryptocurrency, which can be performed in many ways. The value of bitcoin is determined by the free market. Suppose a vitality program participant is recruited on a blockchain; the life insurance company must have a certain amount of cryptocurrency. A participant who received the cryptocurrency must be able to exchange it for their benefit. In a conventional blockchain, operators obtain cryptocurrency in exchange for maintaining a blockchain system. However, unless the life insurance company participates in the blockchain operation, it must purchase the cryptocurrency from another party to secure a certain amount of it. In addition, assuming that the value of this cryptocurrency changes, there could be some concern that the total amount of money in the blockchain system would become negative. To solve these problems, we define a concept called
Several members exist in a blockchain system with a currency exchange. First, there is a blockchain operation party that operates the blockchain and acts as a currency exchange. In this study, it is assumed that the blockchain operating party does not affect the token economy because the party only generates cryptocurrency, and the value of cryptocurrency is fixed via the exchange. Second, there is a data provider that has patients’ health information. Members who store personal health information include entities such as hospitals or genetic companies. They receive cryptocurrency from members who request information when there is a transaction of the information they store. The reason for receiving the cryptocurrency is the cost of storing the information. The third party is a user on the medical information platform. Users can be patients requesting their genomic information, healthy people uploading their life-log data, or life insurance companies wanting to recruit participants. In this study, we assume a life insurance company prepares a vitality program from the perspective of mechanism design by recruiting participants and finding out how long the recruitment period will be, depending on the amount of the incentive.
Basic scenario. PHR: personal health record.
We established the validation scenario by modifying the basic scenario (
Validation scenario. PHR: personal health record.
The participants will decide to participate in the research based on the opinion of doctors who are treating their cancer. In our study, we assumed that the smaller the amount of clinical information that participants are required to send, the higher the probability of consent. This is based on the general assumption that the more information participants have to provide, the more careful they are because normally people care about privacy and security.
For example, if participant X in group A agrees to consent, the institution storing the data of participant X shares the data with the research company and the institution obtains cryptocurrency compensation. In the validation scenario, the companies transfer tokens to data providers that have clinical information for candidate patients in exchange for compensation for storing the data, which is different from the basic scenario in that data providers are added in the token flow
Mathematical models for both scenarios were constructed; a methodology that minimizes the cost of the organizers was modeled. The essential members were participants and organizers. In both scenarios, the reward was the main parameter that affects participation—we assumed that the higher the reward, the greater the probability of consent. In the case of an organizer, an assumption is needed about which information can be obtained from the participants and the organizers’ cost policies. The cost was divided into 2 categories: reward-related parameters and time cost–related parameters.
We assume that the organizer's cost function is divided into 2 parts—reward and cost. Rewards are provided to participants; therefore, the reward portion is multiplied by the total number of participants. The time is also included in the time cost, and because the rewards and time are different in units, it is necessary to balance both parts by giving one part a weight. In our model, we give a weight to the time cost part (the time weight). The time weight may vary in different recruitment programs. A higher time weight means that an organizer values recruitment time more and wants to recruit participants quickly. We express the cost function as
When minimizing the cost, if the time weight α is small, cost will be largely dependent on the reward. The unit of time
In addition, the recruiting time calculation method should be defined. The organizer sends consent requests to those who meet the requirements, and it is assumed that each participant has a probability of consent. Under the assumption, the time taken for recruitment is calculated as follows. On day 1, the organizer sends consent requests to the people in the participant pool. Some participants agree based on their probability, and others disagree. If the number of people who agree does not satisfy the required number of recruits, on day 2, the organizer will send the consent request again to those who did not agree. This process is repeated to calculate the recruiting time. Recall that the probability of consent to participate in research is dependent upon the reward, and the higher the reward, the greater the probability of consent. Increasing the probability of consent means it will be easier to recruit participants, and the time
We assumed that participants can access information about how long the program lasts and the reliability of the insurance company. They also consider inconvenience that may be caused by joining the program.
The welfare of a participant is separated into 2 parts: (1) monetary value and (2) labor. In order to have unbiased results, the socioeconomic status of participants should be considered. The lower the status, the more significant impact on the monetary part, which results in biased recruitment; therefore, we introduced a concave function to represent monetary welfare. Participants have information about their socioeconomic positions, the degree of annoyance for recording tasks, and the reliability of life insurance companies, which are not disclosed to the research companies. Let the socioeconomic status of the
The participant's welfare is defined using the square root (which is a concave function), which means the higher
The former term is the weight given to the welfare according to the socioeconomic status, and the latter term is the fatigue of continuously uploading the log by multiplying the basic quadratic function for
Socioeconomic status is likely to be biased because it provides compensation for participation. Therefore, at initial recruitment, a certain number of the desired population is selected to ensure that socioeconomic status is as evenly matched as possible. In labor-related welfare, fatigue builds up faster over time, and each person’s degree of annoyance is different. Hence, the labor-related cost is proportional to the degree of annoyance and the square of the duration.
In addition to consideration of financial benefits, the probability of participating in the program was introduced (to consider the psychology of real-world participants). The probability will be large when the reward is large, the reliability of the organizer is high, and the socioeconomic status is low.
Assuming that the welfare function of the
The normalizer is a buffer weight to allow the rewards to grow. If the right term of equation 2 is greater than 1 for most of
We modified the previous assumptions and added an intermediate data provider in the validation scenario to ensure that the model is robust after we change variables.
For data providers, assumptions are not necessary because there is no separate strategy available; however, there is a higher chance of making a profit in proportion to the amount of information held by the data providers. A research company can define its cost function, which sets the compensation and time weight per participant. Participants must decide whether to agree or not for the consent request they received. Theoretically, participants decide to agree when their economic gains are greater than 0. In real life, however, other factors play a role in determining whether a patient will be able to participate in clinical trials. Thus, we assume that the agreement probability is based on the expert advice which is provided through the blockchain platform. Normally, patients get expert opinions from their oncologists. If
Participants in research must disclose whether their genes have been tested, genotype (if they have been tested), and which data provider is storing their data. Moreover, some information on the medical blockchain is open to companies: (1) whether one’s genes have been tested and (2) genotype. A few properties are needed to describe this. First,
In the basic scenario, the probability of consent of participants is the expert advice. The greater the rewards, the smaller the number of protocols available to participants and the greater the expert advice. The probability can be defined as a simple product in the basic scenario, but in the validation scenarios, the sigmoid function is used,
The sigmoid function converts a real value to a value between 0 and 1. Therefore, this value has the advantage of being used directly as a probability. When
We chose the convex optimization model. For each scenario, we suppose that the expected recruitment time can be found when the clinical trial is given a reward. For each reward and time of completion of the recruitment pair, we obtain the cost of applying the pair. The reward and time
To find the expected recruitment time for a given reward amount, the probability of each participant agreeing to participate in the clinical trial is determined. Then using Bernoulli implementation, we obtain the expected number of participants that agreed until
Therefore, if these expectations are added for all
Aggregating the expected values, we determine the expected number of participants upon
The reward affects the probability of consent of each participant, and this probability of consent affects the expected recruitment period
In
(A) Time–reward and (B) cost–reward trade-offs in the basic scenario when the time weight is equal to 1500.
Single-case (A) time–reward and (B) cost–reward and multiple-case (C) time–reward and (D) cost–reward trade-offs in the validation scenario when the time weight is equal to 100.
We compared the simulation results with optimization results. Since
Basic scenario mean values and standard deviation of errors (expected days=13).
Basic scenario.
Experiment | Error, mean (SD) |
1 | –0.57 (0.77) |
2 | –0.46 (0.69) |
3 | –0.33 (0.68) |
4 | –0.56 (0.77) |
5 | –0.49 (0.73) |
6 | –0.49 (0.61) |
7 | –0.67 (0.68) |
8 | –0.44 (0.80) |
9 | –0.61 (0.67) |
10 | –0.56 (0.66) |
Validation scenario mean values and standard deviation of errors.
Validation scenario.
Experiment | Expected days, n | Error, mean (SD) |
1 | 9 | –1.70 (4.51) |
2 | 9 | –1.37 (5.01) |
3 | 9 | –1.02 (4.16) |
4 | 10 | –1.84 (7.31) |
5 | 10 | –1.39 (4.82) |
6 | 9 | –1.73 (5.22) |
7 | 8 | –1.31 (4.28) |
8 | 9 | –0.89 (4.59) |
9 | 10 | –2.41 (6.45) |
10 | 10 | –1.30 (4.50) |
The company can adjust the number of participants
In this paper, we propose 2 token economy scenarios of health care blockchain with mechanism design. We set basic components in each scenario, which were the number of participants and recruitment time, constructed mathematical models to explain the 2 scenarios, and simulated changes in recruitment time and the number of expected participants. Through mechanism design, we demonstrated that the recruiter is able to set a desired and expected number of participants and recruitment time by adjusting the amount of incentive. This study is the first, to the best of our knowledge, to apply mechanism design to health care blockchain for real-world problems.
In classic game theory, designers of the game are not able to determine expected results of games quantitatively. They can expect rational participants to compete with each other for the best results by assuming that participants are reasonable, therefore, act to maximize their profits; however, it is not realistic to apply this assumption directly in the health care blockchain with token economy because it is crucial for a funder of health care blockchain to know expected time, cost, and the number of recruited participants.
In our models that employ mechanism design, the recruiter can set a quantitative outcome of programs by adjusting the relationship of number of participants, recruitment time, and rewards. In addition to the predictability of outcomes, the blockchain system can provide participants more secure environments with immutability of the system, although the system is more open to the public compared to conventional ways of recruiting participants.
Traditionally, the issue of whether direct incentives for health care participants are acceptable has been debated because direct incentives can bias medical research and treatment outcomes [
From a mathematical perspective, information about each participant is different depending on whether recruiting companies have access to it or not. While recruiting healthy individuals, we assumed that information about each participant is unobtainable by the company because they usually do not need strict inclusion criteria, which means that they can exclude some of applicants after closing the recruitment. When collecting life-log data, the degree of annoyance for each participant over a long period is unobtainable and we can only see the distribution for the degree of annoyance. In this study, this distribution is assumed to be a uniform distribution. In this case, a clinical trial company uses the expected value of the distribution (ie, the uniform distribution's average value). Meanwhile, in the validation scenario, each participant should provide gene data to the research company. From the gene data, the company can infer the probability of agreement of each participants and can adjust model to fit real-world data.
Recruiting participants is one of the most difficult parts of conducting a clinical research or trial. However, the token economy, through mathematical modeling and simulation techniques proposed in this study, will enable companies to obtain important insight into whether they can recruit participants within the desired period by setting appropriate rewards for recruiting participants. Moreover, due to difficulties in recruiting participants suitable for clinical trials, companies may make a request to hospitals for patients information. However, hospitals do not feel the need and reason to provide data they hold, or even if hospitals are willing to provide data, the consent process for the use of patient data is complicated, and many patients are concerned about privacy and security [
Our assumptions did not reflect complex circumstances for recruiting participants in the real world. Thus, if the results of our models are different from the expectations of recruiting company, recruitment would not be successful. Models with basic mandatory requirements such as recruitment time, amount of incentives, and number of participants, and without complicated assumptions, can be useful for other researchers as a starting point of their own modeling—they can use our framework as a basic scenario to design their own sophisticated token design in health care blockchain. Another limitation is that comparison and analysis with actual data could not be carried out to provide a more robust framework.
There have been few business models adopting blockchain technology in health care. Token economy of blockchain can be a powerful driver by incentivizing health care consumers with an immutably trackable token transaction system. This research is a starting point of designing the token economy in real-world health care settings; we modeled 2 possible scenarios, optimized the cost of the company, and compared the results with simulation data. This study can be used as a basis for designing an incentive system for companies recruiting clinical trial participants or health care program participants within a set period of time with a blockchain-based patient recruitment platform.
This research was supported by the Seoul National University Bundang Hospital Research Fund (grant 02-2020-0031), the Korea Health Industry Development Institute of the Ministry of Health and Welfare (HI19C0152), the National Research Foundation of Korea grant funded by the Korean Ministry of Science and ICT (2017R1E1A1A03070105), and by the Information Technology Research Center support program supervised by the Institute for Information and Communications Technology Promotion (IITP-2018-0-01441). We are especially thankful to Ilha Yune for editing the manuscript.
SYJ and THK drafted the manuscript. KPH contributed to the discussion. As the corresponding author, HJH supervised the process of writing the manuscript.
None declared.