Electronic medical record distributed sharing model based on hybrid alliance chain and stackelberg game

Abstract

The invention discloses an electronic medical record distributed sharing model based on a hybrid alliance chain and a stackelberg game, a hybrid alliance chain architecture is constructed, and credible and safe sharing of EMRs data is realized through cooperation of multiple participants. Then, decision conflicts possibly faced by participants in the sharing process are analyzed by applying a Stackelberg game theory, and a corresponding incentive mechanism is designed, so that each party can obtain a reasonable return while sharing data. In addition, the decomposition characteristic and the rapid convergence of the alternating direction multiplier method are utilized, and the profit maximization of the EMRs data provider and the effectiveness maximization of the EMRs data consumer are realized in a distributed mode. And finally, clustering the consensus nodes in the alliance chain network by using a condensation hierarchical clustering algorithm to improve the traditional PBFT consensus algorithm, improve the efficiency of EMRs data sharing in the alliance chain network, and further reduce the existence of data islands.

Description

Technical Field

[0001] The present invention relates to the field of digital medical technology, and in particular to an electronic medical record distributed sharing model based on a hybrid alliance chain and Stackelberg game. Background Art

[0002] At present, the construction of medical digitization is an important measure to promote the modernization and efficiency of the medical industry. By introducing advanced information technologies such as artificial intelligence, cloud computing, and big data, it is possible to realize the electronicization of medical data, the interconnection of information, and improve the quality and efficiency of medical services. With the continuous advancement of medical digitization, electronic medical records (EMRs) have received much attention and are widely used because they can provide convenient and superior electronic medical services [1]. First, as digital medical records, case records, and personal health records, electronic medical records can integrate patients' core health data and conduct comprehensive analysis with advanced technical means to obtain more accurate diagnosis and treatment plans. Second, unified electronic medical records are conducive to improving the integrity and traceability of data records, improving hospital management level and doctor work efficiency, and providing strong support for remote diagnosis, hierarchical diagnosis and treatment, and medical insurance cost control. At the same time, electronic medical records contain rich clinical data and experience. If they can be shared and interconnected, they will hopefully alleviate the current problems of insufficient medical resources, uneven distribution, and poor service homogeneity. China's exploration of building a national unified electronic medical record system began around 2000. After more than 20 years of efforts, the construction of basic information systems such as HIS, LIS, and PACS has been basically completed in hospitals above the second level nationwide, and most tertiary hospitals have established information systems such as telemedicine and regional imaging diagnosis. More than 7,000 public hospitals above the second level nationwide have access to the provincial regional coordination platform, and more than 2,200 tertiary hospitals have initially achieved information intercommunication and sharing within the hospital.

[0003] However, since electronic medical records are stored in various medical institutions, this not only hinders data sharing, but also increases the risk of patient privacy leakage. Therefore, it is a huge challenge to design an open and shared medical and health service system to improve diagnostic capabilities and medical quality.

[0004] At present, many hospitals or individual institutions in China still use traditional databases to store patient information. When a doctor needs to view a patient's medical records, the system extracts the required data from the database. However, this method cannot achieve data sharing with other medical institutions, resulting in the problem of information islands. Summary of the invention

[0005] The purpose of the present invention is to provide an electronic medical record distributed sharing model based on a hybrid alliance chain and Stackelberg game to solve the problems raised in the above background technology.

[0006] To achieve the above-mentioned purpose, the present invention provides the following technical solution: an electronic medical record distributed sharing model based on a hybrid alliance chain and Stackelberg game, characterized in that it includes:

[0007] S1, based on blockchain technology, record and manage electronic medical records in the medical and health data management system, establish a mathematical programming model for the interaction between ordinary nodes and on-duty nodes, establish an EMRs data management system based on a hybrid alliance chain, and model the interaction between on-duty nodes and ordinary nodes by introducing Stackelberg game;

[0008] S2, when the electronic medical record sharing process is executed, a pricing mechanism is established to charge for the sharing process, a secure transaction and sharing model between ordinary nodes and on-duty nodes is constructed, and the multi-leader and multi-follower Stackelberg game theory is used to balance the supply and demand relationship of electronic medical records;

[0009] S3,implementation of electronic medical record secure transaction and sharing model strategy based on ADMM algorithm;

[0010] S4, train the PBFT consensus algorithm based on agglomerative hierarchical clustering.

[0011] Furthermore, in S1, a medical and health data management scenario based on the alliance chain is established, including four heterogeneous blockchain networks: community residents blockchain network, fitness club blockchain network, community hospital blockchain network, and large hospital blockchain network, forming a medical and health public alliance chain network;

[0012] The medical and health data management scenario based on the alliance chain mainly includes the following roles: certificate authorities, community residents, fitness clubs, community hospitals, large hospitals, on-duty nodes, data owners, data users, and medical coins.

[0013] Furthermore, in S1, an EMRs distributed sharing model based on a hybrid alliance chain and Stackelberg game is established, in which ordinary nodes in the blockchain network are regarded as followers in the Stackelberg game model, and on-duty nodes are regarded as leaders in the Stackelberg game model. Information is circulated through transactions of EMRs resources between leaders and followers in the Stackelberg game model.

[0014] Furthermore, in S2, the EMRs data resource transactions between ordinary nodes / data users and on-duty nodes are modeled based on the multi-leader and multi-follower Stark game theory, and the on-duty nodes set the price of EMRs data resources with the goal of maximizing profits;

[0015] Based on the price set by the on-duty node, ordinary nodes determine the demand for EMRs data resources with the goal of maximizing utility.

[0016] Furthermore, in the multi-leader and multi-follower EMRs distributed sharing model based on the hybrid alliance chain, ordinary nodes i and on-duty nodes j are assumed to be rational agents, and an incentive mechanism is established.

[0017] Furthermore, in S3, the Stackelberg game based on the ADMM algorithm;

[0018] Phase 1: Each ordinary node observes the prices published by all on-duty nodes at the beginning of each iteration and decides to purchase EMRs data resources from one of the on-duty nodes;

[0019] The on-duty node repeatedly updates the value of the EMRs data resource based on the augmented Lagrangian method:

[0020] Phase 2: The on-duty nodes in the healthcare public alliance chain network know the behavior of ordinary nodes and predict the demand for data resources. Each on-duty node updates the price in the strategy space through the ADMM algorithm.

[0021] Furthermore, in S3, based on the AHC-PBFT consensus algorithm, the two most similar data points among all data points are combined and this process is repeated. AHC determines the similarity between the data points of each category by calculating the distance between them and all data points.

[0022] Furthermore, in AHC, a part of network consensus nodes is artificially and randomly generated, and each common node has EMRs transaction request information. By standardizing indicators and similarity measurement criteria, any number of clusters specified by the user is obtained.

[0023] Compared with the prior art, the present invention has the following beneficial effects:

[0024] (1) Combining multiple independent fields such as Stackelberg game theory, blockchain, medical big data, and clustering algorithms, a feasible solution to the distributed sharing of EMRs data is explored. First, a distributed sharing model of EMRs based on a hybrid alliance chain and stackelberg game is proposed for non-single medical network scenarios, which allows users to freely access the required EMRs in different medical institutions. Then, considering the high multidimensionality of the EMRs data consumer strategy space, the ADMM algorithm is introduced to maximize the profit of EMRs data providers and the utility of EMRs data consumers in a distributed manner, and the convergence of the ADMM algorithm is verified by numerical derivation. Finally, the PBFT consensus algorithm is improved based on AHC to improve the efficiency of EMRs data sharing. This sharing model is more conducive to improving the efficiency of EMRs data sharing and circulation among users in the medical scenario of this embodiment, promoting the interoperability of medical systems and related departments, and alleviating the problem of data islands.

[0025] (2) The combination of game theory and blockchain can promote many unexpected advantages. On the one hand, blockchain requires a large amount of computing resources to carry out distributed data recording. On the other hand, market-oriented blockchain applications require a large number of users to join, form consensus, and achieve the stability of the blockchain network. Game theory can effectively allocate resources, improve user enthusiasm, and make up for the shortcomings of blockchain technology. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Figure 1 is a framework diagram of the EMRs of the present invention;

[0027] Figure 2 This is a scenario diagram of medical and health data management based on a hybrid alliance chain in the present invention;

[0028] Figure 3 It is a Stackelberg game diagram of multiple leaders and multiple followers of the present invention;

[0029] Figure 4 It is a game tree diagram based on the reverse induction method of the present invention;

[0030] Figure 5 It is a schematic diagram of the agglomerative hierarchical clustering algorithm of the present invention; DETAILED DESCRIPTION

[0031] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0032] Example:

[0033] See also Figure 1-5 ,The present invention provides a technical solution: an electronic medical record distributed sharing model based on a hybrid alliance chain and Stackelberg game.

[0034] In the field of medical health, data is as valuable as gold, but due to the limitations of "information islands" and the fact that data owners always have concerns about data privacy and security, these "scattered treasures" are difficult to be mined and shared, and the rich value of data cannot be brought into play. Therefore, how to separate the ownership and use rights of data on the basis of ensuring data privacy and security, make data "available but invisible", and promote the sharing of medical data across hospitals and regions is a hot issue that researchers at home and abroad are paying close attention to.

[0035] 1.1 Hospital Information System

[0036] Hospital Information System (HIS) is a centralized system that provides comprehensive automated management and various services for the overall operation of the hospital. Electronic medical records are the core component of the hospital information system. They replace traditional paper medical records and record patients' medical information in a digital form. With the explosive growth of electronic medical record data, some new methods based on the Internet of Things technology have been proposed to improve the robustness and security of the system.

[0037] However, these IoT-based HIS are centralized and cannot meet the growing security and efficiency requirements. In these centralized systems, medical institutions control the entire system. Once the public-key generator (PKG) or third-party auditor (TPA) becomes malicious, it is easy to cause data deletion, tampering and other problems. Therefore, the centralized cloud storage structure cannot provide adequate protection for electronic medical records. In addition, the integrity of electronic medical records can also be easily destroyed by inevitable software / hardware failures and human errors in the cloud. At the same time, due to privacy considerations and competitive pressures, different medical institutions are reluctant to share their own data, and the consistency and interoperability of different data types from different medical institutions are also important issues in data sharing.

[0038] 1.2 Blockchain Technology

[0039] Blockchain is a data structure that organizes and maintains a large amount of data in a decentralized manner based on cryptographic technology. It can be understood as a distributed ledger or database. In order to achieve distributed consistency between unfamiliar and untrusted users, blockchain networks have some consensus algorithms, such as Proof of Work (PoW), Proof of Stake (PoS), and Delegated PoS. At the same time, a signature algorithm is also needed to verify the legitimacy of users and protect the privacy and security of users and transactions. Blockchain is generally considered to be a public, decentralized, distributed, and reliable super ledger with high Byzantine fault tolerance. It has been widely used in finance, medical care, digital evidence storage, cloud computing, Internet of Things systems, and other fields.

[0040] In general, blockchain technology provides a new paradigm for data sharing, changes the traditional way of data management and sharing, improves the security, transparency and reliability of data, and promotes more extensive data sharing and cooperation. Therefore, this embodiment uses blockchain technology to build a distributed platform between different medical institutions to achieve safe and efficient EMRs sharing.

[0041] 2.3 Blockchain-based EMRs sharing

[0042] HIS is a large framework or platform, while EMRs are systems used within this framework to record and manage patient medical information. EMRs are the key to realizing medical informatization and an important tool for improving medical quality and efficiency. The two complement each other and jointly promote the informatization of the medical industry. However, centralized organizations are prone to data leakage due to human errors and hardware damage. In addition, due to privacy issues and competitive pressure, medical institutions are reluctant to share their EMRs. Therefore, cross-institutional sharing of EMRs is a major challenge facing the current centralized HIS.

[0043] With the development of blockchain technology, its characteristics of decentralization, traceability, anonymity, etc. have attracted wide attention in the application of medical industry. At present, many scholars are paying attention to the problem of EMRs sharing based on blockchain technology. Some people have proposed a conceptual design of personal continuous dynamic health data sharing based on blockchain technology, supplemented by cloud storage, to share personal health-related information in a secure and transparent manner. Some people have proposed a cloud-assisted secure electronic health system, which uses blockchain technology to protect outsourced EMRs in the cloud from illegal modification. The key idea of ​​this system is that EMRs can only be outsourced by authenticated participants, and each operation of outsourced EMRs is integrated into the public blockchain as a transaction, which has higher security performance than that of blockchain. Based on the existing technology, there is also a privacy-preserving data sharing scheme based on blockchain, namely BPDS. In BPDS, the cloud is used to securely store the original EMRs, and a tamper-proof consortium blockchain is designed to share the EMRs index. There is also a blockchain-based searchable encryption scheme for EMRs sharing to improve the searchability of data. In this case, the construction of the EMRs index stored in the blockchain is a complex logical expression, which can be used by data users to search the index. In order to further improve the security of EMRs sharing, there is also a storage solution and service framework for medical data storage, sharing and use based on blockchain and cloud. In this solution, personal medical data applications based on blockchain can provide patient EMRs sharing services without infringing privacy.

[0044] The above works have proposed various EMRs sharing schemes from different aspects, but there is no detailed solution for specific application scenarios. In particular, in the process of cross-domain sharing of EMRs, how to ensure mutual trust between institutions in different scenarios and ensure the secure sharing of data is an urgent problem to be solved. In addition, the communication overhead of the traditional consensus algorithm based on blockchain is large and cannot meet the scalability requirements of large-scale networks. At the same time, for a single-chain network, large-scale user access will cause tremendous pressure on the blockchain network, thereby affecting the operating efficiency of the system.

[0045] Therefore, this embodiment designs a medical and health data management system for multiple scenarios based on the alliance chain, uses the Stackelberg game theory to model the interaction between ordinary nodes and on-duty nodes, and uses agglomerative hierarchical clustering to group consensus nodes in the blockchain network to improve consensus efficiency, ultimately achieving the purpose of promoting EMRs sharing, such as Figure 1 shown.

[0046] Table 1, commonly used symbols

[0047]

[0048] 2. EMRs data management system based on hybrid alliance chain

[0049] In this section, this embodiment will describe the medical health system in detail. In this system, users can create and upload their own EMRs, and can also verify whether their EMRs are recorded. At the same time, a number of different institutions act as consensus nodes to maintain the entire system. They are responsible for collecting and verifying EMRs, generating and ultimately recording them in the public blockchain. The transparent and immutable data storage structure makes EMRs data more secure and reliable. First, this embodiment sets up a medical health data management scenario based on a hybrid alliance chain, such as Figure 2 ; Then, this embodiment describes the EMRs sharing model based on the hybrid alliance chain, and models the interaction between the on-duty node and the ordinary node by introducing the Stackelberg game to obtain the optimal EMRs price of the on-duty node and the optimal demand for EMRs of the ordinary node. 2.1

[0051] Medical and health data management scenarios based on alliance chain

[0052] The medical and health scenario ( Figure 2 ) involves four heterogeneous blockchain networks, namely, the Blockchain network of Community residents (BNCR), the Blockchain network of fitness clubs (BNFC), the Blockchain network of community hospitals (BNCH), and the Blockchain network of large hospitals (BNLH). These four heterogeneous networks together constitute the Public consortium-chain network of medical and health (PCNMH). The medical and health scenarios based on the consortium chain mainly include the following roles: certificate authority (CA), community residents, fitness clubs, community hospitals, large hospitals, duty nodes (DN), data owners (DO), data users (DU), and medical coins.

[0053] 1. Certificate Authority (CA): Responsible for managing various institutions, generating unique identities for nodes in the blockchain network, and generating public and private key pairs. CA is the authoritative institution in this system and is completely trustworthy.

[0054] 2. Community residents: In daily life, each resident is a family unit, and each family belongs to the community. This embodiment regards the community as a consortium chain node, and multiple communities together form a community resident consortium chain network. Each consortium chain node generates a unique identity for the patient, allocates public and private keys, manages the patient's family health data, and saves the data to the corresponding edge server.

[0055] 3. Fitness club: The fitness club can generate a unique identity for the user, assign public and private keys, manage the user's exercise data, and save the data to the corresponding edge server. In this embodiment, the fitness club is regarded as a consortium chain node, and multiple consortium chain nodes together constitute the fitness club consortium chain network. When a fitness coach makes a personalized exercise plan for a user, he can view the user's other health data with the user's authorization.

[0056] 4. Community Hospital: Community hospitals can generate unique identities for users, assign public and private keys, manage patients’ medical and health data, and save the data to the corresponding edge servers. A community hospital is regarded as a local alliance chain node, and multiple community hospitals together form a community hospital local alliance chain network.

[0057] 5. Large hospitals: Large hospitals with relatively abundant medical resources, such as municipal hospitals and above. Large hospitals mainly generate unique identities for patients, allocate public and private keys, manage patients' medical and health data, and save the data to the corresponding edge servers. A large hospital is regarded as a local alliance chain node, and multiple large hospitals together form a large hospital local alliance chain network. Compared with community hospitals, the diagnosis results of large hospitals are more convincing and can provide higher reference value for patients' secondary visits.

[0058] 6. Duty Node (DN): As a node manager in a single network, the duty node is trusted by the nodes in the network, is responsible for communication between heterogeneous networks, and handles internal and external access requests.

[0059] 7. Data Owner (DO): Data owners are the producers of health data. They collect users’ health data (physical indicators such as blood pressure, blood sugar, heart rate, etc.) through smart devices. Users can not only upload, view, and share their own health data, but also view their own data access records.

[0060] 8. Data Users (DU): Data users mainly include doctors, fitness coaches and other data accessors. Every access record of a data user will be stored in the blockchain. Among them, the internal access records of the network are stored and shared in the log local alliance chain, and the cross-domain access records are stored and shared in the medical and health public alliance chain.

[0061] 9. Medical Coin: When EMRs are shared between different nodes, compensation needs to be paid. In order to distinguish general medical expenses, we introduced a token called Medical Coin for EMRs data exchange. The providers of EMRs will be rewarded with some medical coins by sharing EMRs, and the consumers of EMRs data need to pay some medical coins. The medical coin is used to evaluate the EMRs sharing behavior and balance the supply and demand of EMRs resources. In addition, these medical coins will be stored in the wallet, and through appropriate conversion relationships, they can be used to deduct part of the total medical expenses, which can incentivize the sharing of EMRs between institutions. At the same time, the wallet is an important part of the EMRs sharing process of medical nodes. Users will generate multiple wallet addresses for different transactions. It can store the attributes of medical nodes, protect the privacy and security of medical nodes, and prevent malicious attacks.

[0062] The nodes in the medical and health alliance chain are composed of consensus nodes in four heterogeneous networks. The master nodes in the consensus process are taken turns by the on-duty nodes (DN) in the four networks. According to the network type, the consensus nodes of the medical and health alliance chain are divided into five consensus groups. The first main group is composed of on-duty nodes in four networks, and the remaining four groups of consensus nodes are composed of consensus nodes in four heterogeneous networks. The network conducts intra-group consensus through multicast to improve the efficiency of reaching consensus between nodes. With the continuous increase in the number of nodes (institutions) in the system, the number of ordinary nodes in the alliance chain increases accordingly. In order to ensure the consensus efficiency of nodes in large-scale networks, the number of consensus nodes remains unchanged regardless of whether the number of ordinary nodes increases.

[0063] 2.2 EMRs distributed sharing model based on hybrid alliance chain and Stackelberg game

[0064] Figure 3 Based on the following assumptions: we regard ordinary nodes in the blockchain network as followers in the Stackelberg game model, and the on-duty nodes as leaders in the Stackelberg game model. Through the transaction of EMRs resources between leaders and followers in the Stackelberg game model, we promote the sharing of EMRs in the entire blockchain network and the interoperability of medical systems and related institutions, improve the efficiency of information circulation, and alleviate the problem of information islands.

[0065] 2.2.1 Parameters and definitions

[0066] There are N data users (DU) as ordinary nodes in the entire medical health public alliance chain network, where ordinary nodes i∈N=={1,2,......,N}. There is a duty node (DN) in each local alliance chain network, and there are M duty nodes in the entire medical health public alliance chain network, N>M, where duty nodes j∈M=={1,2,......,M}. Each DU is both the creator of the electronic medical record data of the local blockchain network and the user of the electronic medical record data in the heterogeneous blockchain network. However, each DU can only share access to the local electronic medical records (L-EMRs) within the blockchain network to which it belongs. In order to cross-domain access to electronic medical records in heterogeneous blockchain networks, the duty node will charge a certain price to the ordinary node, and then the duty node will share the electronic medical records in its medical health public alliance chain with the corresponding ordinary node. It is worth noting that: when the ordinary node / data user (DU) and the on-duty node (DN) belong to the same local area network, even if the on-duty node (DN) has the global electronic medical records (G-EMRs) in the entire medical health public alliance chain network, it cannot share the global electronic medical records (G-EMRs) with the ordinary node / data user (DU) for free. They can only share the local electronic medical records (L-EMRs) in the local area network. Table 1 provides commonly used symbols in this embodiment.

[0067] In the first stage of the Stackelberg game, the on-duty node (DN) aims to maximize profits and, as a leader, sets the unit price p of the global electronic medical records (G-EMRs) in the entire healthcare public alliance chain network. j In the second stage of the Stackelberg game, ordinary nodes / data users (DUs) determine their demand for data resources as followers based on the price of electronic medical record data resources. ij When d ij = 0, p ij = 0. Therefore, the demand matrix of all common nodes / data users (DU) is:

[0068]

[0069] The corresponding payment matrix for all ordinary nodes / data users (DU) is:

[0070]

[0071] Since all the on-duty nodes have the same data, which are all authentic and reliable global electronic medical record data on the medical and health public alliance chain that has passed the PBFT consensus, ordinary nodes / data users (DU) can achieve the goal of sharing global electronic medical records (G-EMRs) in the global blockchain network no matter which on-duty node (DN) they make an electronic medical record sharing request to. In other words, ordinary nodes / data users (DU) will only make a global electronic medical record (G-EMRs) data sharing request to one on-duty node (DN). As for which on-duty node ordinary nodes / data users (DU) will choose, it depends on the price of global electronic medical record (G-EMRs) data set by the on-duty node (DN). Since ordinary nodes / data users (DU) are rational, the lower the price of global electronic medical record (G-EMRs) data, the stronger the sharing demand of ordinary nodes / data users (DU); the higher the price of global electronic medical record (G-EMRs) data, the weaker the sharing demand of ordinary nodes / data users (DU). Therefore, the sharing request matrix between ordinary nodes / data users (DU) and on-duty nodes is:

[0072]

[0073] in When x ij = 1, ordinary node i makes a sharing request d to the on-duty node j ij , the price paid is p ij ; When x ij = 0, d ij =0, accordingly, p ij =0.

[0074] In order to more intuitively reflect the request of ordinary nodes / data users (DU) to the duty node (DN), we use the above chessboard form as an example. Assume that there are 5 leaders and 8 followers: The grid colored as a green circle represents x ij =1, and all other cases indicate x ij =0, that is, each ordinary node will eventually make a sharing request to only one on-duty node. It is worth noting that the size of the circle represents the tendency of ordinary node i to make a request to on-duty node j. For example, the fourth ordinary node has a greater tendency to make a sharing request to the first leader and the fifth leader. However, due to the influence of externalities, the fourth ordinary node finally chooses to make a sharing request to the fifth leader. 45 =1,x 41 =0.

[0075] 2.2.2 Stackelberg Game Model

[0076] In the Stackelberg game model, the on-duty node (DN) acts as the leader and the ordinary node / data user (DU) acts as the follower. In order to improve the efficiency of electronic medical records (EMRs) data sharing and circulation in the public alliance chain network of medical health, the interaction between the on-duty node and the ordinary node is modeled. The model sets the probability for each ordinary node i to select the on-duty node j:

[0077]

[0078] Among them, b ij is the highest unit price of the on-duty node j The unit payment price p of ordinary node i ij Similarly, b ik It is the motivation for price reduction when the on-duty node provides global electronic medical records (G-EMRs) data sharing for ordinary nodes.

[0079]

[0080] Proposition 1: According to (1) and (2), when other on-duty nodes (DN) fix the price of EMRs data, ω ij The value of is equal to the price p paid by ordinary node i to on-duty node j ij Negative correlation.

[0081] Proof: First, according to (1) and (2), ω can be ij Converted to the following form:

[0082]

[0083] Then, when x ij =1, we find ω based on formula (3) ij (p ij )’s first and second derivatives:

[0084]

[0085]

[0086] It can be seen from this that ij (p ij ) is a concave function, ω ij (p ij )∈[0,1], when x ij =0,ω ij (p ij )=0.

[0087] in:

[0088]

[0089] Therefore, in order to encourage more ordinary nodes / data users (DU) to participate in the sharing of global electronic medical records (G-EMRs) in the entire alliance blockchain network, promote the interoperability of medical systems and related institutions, improve the efficiency of data circulation, and alleviate the problem of information islands. Based on the Stackelberg game model, each on-duty node j∈M=={1,2,......,M} tends to offer a lower price p to ordinary nodes / data users (DU). ij , the reason is that when ω ij When increases, ordinary node i is more likely to make data sharing requests to on-duty node j, which leads to price competition among on-duty nodes (DN).

[0090] The request of the electronic medical record data resource of the ordinary node / data user (DU) to the on-duty node (DN) is regarded as the data sharing access capability of the ordinary node / data user (DU). Before the ordinary node / data user (DU) makes a data sharing request to the on-duty node (DN), the ordinary node i has the initial data sharing access capability l i Therefore, ordinary node i purchases electronic medical record data resource d from on-duty node j. ij The total shared access capability is:

[0091]

[0092] The data sharing and access capabilities of ordinary node i can affect its decision and behavior in the process of purchasing electronic medical record data resources of on-duty node j, thus affecting its satisfaction with ordinary node i. Make an impact:

[0093]

[0094] Since satisfaction is usually considered as a positive feedback process, we set the impact factor λ of the shared access capability of ordinary node i i > 0, and use the impact factor λ i and the total shared access capability of ordinary node i jointly describe the satisfaction of ordinary node i For each ordinary node i∈N as a follower, it follows the price p set by the on-duty node (leader) j To determine the EMRs data resource request policy as d i ={d i1 , d i2 ,......,d iM}, thereby maximizing utility, the formula is:

[0095]

[0096] For each on-duty node j∈M as a leader, in addition to the revenue from computing services, there are also service costs for operation and maintenance. The cost of the on-duty node (leader) is related to power consumption or hardware loss, expressed as cd ij Indicates, where c represents the cost coefficient, 0<c<p ij Therefore, each on-duty node j in the strategy space The price of electronic medical records (EMRs) data resources is determined to maximize profits, and the formula is:

[0097]

[0098] In order to improve the efficiency of electronic medical records (EMRs) data sharing and circulation in the medical and health public alliance chain network and promote the interoperability of medical systems and related institutions, we introduce multi-leader and multi-follower Stackelberg game theory to model the electronic medical records (EMRs) data resource transactions between ordinary nodes / data users (DU) and duty nodes (DN). Game theory is a promising resource management tool for studying the interaction between strategic entities. The goal of the game is to find the optimal strategy between participants to maximize their own interests or reach a certain equilibrium state. In this embodiment, in the first stage of the Stackelberg game: the on-duty node (leader) takes profit Π j Maximize the price p of electronic medical records (EMRs) data resources ij ; In the second stage of the Stackelberg game: ordinary node (follower) i sets the price p at the on-duty node j ij Based on the utility u i Maximize the demand for electronic medical records (EMRs) data resources ij .

[0099] A. The second stage of the Stackelberg game:

[0100] Due to the effect of externalities (externalities refer to the fact that ordinary nodes / data users (DU) in the medical and health data management system based on the hybrid chain will affect each other's utility due to the connection and interference between users and the price mechanism of the on-duty node (DN), resulting in the existence of network effects or congestion effects), the data resource demand of electronic medical records (EMRs) of ordinary nodes / data users (DU) is d ij will be affected by the price of the on-duty node j, P = {p 1 , p 2 ,......,p M} and other data user requirementsD -i Each DUi determines its own EMRs data resource requirements by solving the following utility maximization problem: ij :

[0101]

[0102] The subgame problem of ordinary node (follower) i can be written as the following mathematical programming problem:

[0103]

[0104] Due to its financial burden, each ordinary node has no motivation to increase its service demand indefinitely, so the second constraint indicates that since the total budget of ordinary node (follower) i is limited, the maximum electronic medical records (EMRs) data resource demand of each ordinary node does not exceed D max .

[0105] B. The first stage of the Stackelberg game:

[0106] Also due to the effect of externalities, the price of the electronic medical records (EMRs) data resource of the on-duty node j is p ij Will be affected by the price P of other on-duty nodes j -j and general node / data user (DU) requirements The impact of, and each on-duty node j determines its own electronic medical records (EMRs) data resource price p by solving the following profit maximization problem ij :

[0107]

[0108] The subgame problem of the on-duty node (leader) j can be written as the following mathematical programming problem:

[0109]

[0110]

[0111] The traditional Stackelberg game problem has only one leader and multiple followers, such as Figure 4 As shown, the key challenge of the multi-leader and multi-follower game model proposed in this embodiment is the high multidimensionality of the strategy space of each ordinary node (follower) i, which makes it impossible to use the traditional reverse induction method to obtain the Nash equilibrium solution. This is exactly the motivation for us to use the ADMM algorithm to achieve the optimal strategy of the electronic medical record sharing model.

[0112] 3. ADMM algorithm

[0113] Different from the traditional centralized medical information system (HIS), in this medical health public alliance chain, all entities serve as nodes to form the entire system. Figure 2As shown in the figure, medical institutions and some other necessary institutions jointly form a distributed EMRs sharing network. At the same time, the consensus nodes (institutions) jointly maintain a medical health public alliance chain, and the real EMRs are stored in the database managed by each consensus node (institution). Ordinary nodes should first register as legitimate users, then create EMRs, and share their EMRs to the medical health public alliance chain. In order to promote the execution of the EMRs sharing process, a pricing mechanism is introduced to charge for the sharing process, and a distributed EMRs sharing model between ordinary nodes / DUs and DNs is constructed. The multi-leader and multi-follower Stackelberg game theory is used to balance the supply and demand of EMRs data.

[0114] In the multi-leader multi-follower EMRs distributed sharing model based on hybrid alliance chain, both ordinary nodes (followers) i and on-duty nodes (leaders) j are assumed to be rational agents, whose goal is to make their utility u i (d i , D -i , P) and profit Π j (p j , P -j , D) is maximized. We can set the price p ij is regarded as the incentive factor provided by the on-duty node (leader) j to the ordinary node (follower) i, because the on-duty node (leader) j can influence the service demand of the ordinary node (follower) i by controlling the price level. Therefore, we can design the multi-leader and multi-follower EMRs distributed sharing model based on the hybrid alliance chain in 3.2 as an incentive mechanism, so that the profit Π of the on-duty node (leader) j in (13) j (p i , P -j , D) to achieve the optimal result, while considering the utility u of ordinary node (follower) i in (11) i (d i , D -i , P). By adjusting the excitation factor p ij , the on-duty node (leader) j can indirectly enable the ordinary node (follower) i to determine the demand for EMRs data resources d ij , so that the supplier's profit Π j (p j , P -j , D) maximize. Specifically, the incentive mechanism problem can be expressed as the following objective function:

[0115]

[0116] The hierarchical optimization problem of the multi-leader and multi-follower EMRs distributed sharing model based on the hybrid alliance chain includes the equilibrium problem of the upper and lower stages in formula (14), which is a standard equilibrium problem with equilibrium constraints (Equilibrium Problem with Equilibrium Constraints, EPEC). In such an EPEC scenario, it is necessary to coordinate multiple conflicting interests, that is, the utility u of ordinary node (follower) i. i (d i , D -i , P) and the profit Π of the on-duty node (leader) j j (p j , P -j , D), inevitably requires more complex solutions. In order to obtain a solution that can optimize the profit of the on-duty node (leader) j while considering the utility of the ordinary node (follower) i, we use the Alternating Direction Method Multipliers (ADMM) algorithm. ADMM is a parallel optimization tool with decomposition and fast convergence characteristics, which is particularly suitable for large-scale optimization problems.

[0117] 3.1 Preliminary Design of ADMM Mechanism

[0118] In this section, we first briefly introduce the basic implementation of the ADMM algorithm. For ease of explanation, this embodiment considers a simple blockchain network with one on-duty node (leader) j and N ordinary nodes (followers) i. The profit maximization problem of the on-duty node (leader) j is equivalent to the cost minimization problem. Therefore, the cost minimization problem of the on-duty node (leader) j can be expressed as the following optimization problem:

[0119]

[0120] Among them, f i (d i ) is the on-duty node (leader) j selling electronic medical records (EMRs) data resources d to ordinary nodes (followers) i ij Specifically, f i (d i ) is d i Convex function, d i is a real scalar variable, α i and β i is a real scalar constant.

[0121] The on-duty node (leader) j repeatedly updates the EMRs data resource d using the Augmented Lagrange Method (ALM). i Value:

[0122]

[0123] in,

[0124]

[0125] It is a norm that measures the size of a matrix (the dotted box acts as a space). It is calculated by taking the square root of the sum of the squares of all elements in the matrix. ρ is the damping factor, and t represents the iteration step. η is a dual variable, and its update method is:

[0126]

[0127] f i (d i ) is a convex function, ADMM can ensure that f i (d i ) converges to a stable solution set. It should be noted that for non-convex objective functions, the convergence of ADMM can still be guaranteed under certain conditions.

[0128] 3.2 Stackelberg game based on ADMM algorithm

[0129] In this section, we describe the iterative process of the ADMM-based algorithm for optimizing the utility u of an ordinary node (follower) i in a hybrid alliance chain network. i (d i , D -i , P) and the profit Π of the on-duty node (leader) j j (p j , P -j , D). Each iteration consists of the following two-stage optimization problem of the Stackelberg game:

[0130] Utility optimization of ordinary node (follower) i (stage 1): In this stage, each ordinary node (follower) i can observe the price P = {p 1 , p 2 ,......,p M}, and decides to purchase EMRs data resource d from one of the on-duty nodes (leader) j ij , in order to maximize its utility u i (d i , D -i, P), promote the interoperability of medical systems and related institutions, improve data circulation efficiency, alleviate information island problems, and realize EMRs data sharing. Note that the superscript (q) indicates the qth ADMM in the outer loop. th Iterations. Each ordinary node i has a personal utility u i (d i , D -i , P) is maximized to obtain d i This forms the inner loop of ADMM. Specifically, each ordinary node i in the local alliance chain network updates d at each iteration. i The value is as follows:

[0131]

[0132] in,

[0133] In addition, τ = t + 1 if k < i, τ = t if k > i. Here, t is the iteration step of the inner loop. As mentioned before, ρ > 0 is the damping factor, η is the dual variable, and η will be updated as follows:

[0134]

[0135] In each iteration of the outer loop q, ordinary nodes can obtain a set of The values ​​of p that maximize the utility of the common nodes at the end of the inner loop. At the same time, when the value node j can predict such a set of values, these values ​​will be used to update p in the next stage. ij The value of .

[0136] Profit optimization of on-duty node (leader) j (Phase 2): In this stage, on-duty node j in the healthcare public alliance chain network knows the behavior of ordinary node i and can therefore predict d ij Each on-duty node updates p in the strategy space through the ADMM algorithm j value to maximize its profit:

[0137]

[0138] After the value node j completes the price update of the EMRs data, it announces the next iteration, i.e., the (q+1)th iteration, to the ordinary node i.

[0139] This forms the outer loop of the algorithm. The outer loop ends when the following termination condition is met.

[0140]

[0141] Where σ is a predefined threshold. The basic framework of the above ADMM algorithm is shown in Algorithm 1.

[0142]

[0143] Essentially, after the ADMM algorithm based on Stackelberg game examines the utility function of ordinary node i in the local alliance chain and the profit function of on-duty node j in the medical and health public alliance chain, we can draw the following proposition:

[0144] Proposition 2: The utility function i∈N=={1,2,......,N} of each ordinary node i in (10) and the profit function of each on-duty node j∈M=={1,2,......,M} in (12) are strictly concave.

[0145] Proof: First, we find the value of formula (10) with respect to d ij The first and second derivatives of

[0146]

[0147]

[0148] From equation (24), we can conclude that equation (25) is negative. Then, when x ij = 1, we find the value of formula (12) with respect to p ij The first and second derivatives of :

[0149]

[0150]

[0151] Correspondingly, from equations (4), (5), and (26), we can conclude that equation (27) is negative.

[0152] In general, by and It can be seen that both formula (10) and formula (12) are strictly concave functions and converge to a stable solution set.

[0153] 3.3AHC-PBFT consensus algorithm

[0154] Based on the above analysis, the Stackelberg algorithm based on ADMM can essentially obtain the utility function of ordinary node i and the profit function of on-duty node j, thereby promoting the sharing of electronic medical records (EMRs) data in the blockchain network. However, since the research scenario of this embodiment involves multiple heterogeneous networks, the network scale is large and the number of nodes is large. Therefore, this section will focus on the improvement of the PBFT consensus algorithm by agglomerative hierarchical clustering (AHC). The agglomerative hierarchical clustering algorithm (AHC) calculates the similarity between two categories of data points, combines the two most similar data points among all data points, and repeatedly selects this process. Specifically, the agglomerative hierarchical clustering algorithm (AHC) determines the similarity between each category of data points and all data points by calculating the distance between them. The smaller the distance, the higher the similarity, such as Figure 5 .

[0155] In this embodiment, when the client sends a transaction request, the blockchain system clusters all nodes into M clusters through the AHC algorithm, and selects the central node in each cluster as the master node / on-duty node, and the remaining nodes as slave nodes / ordinary nodes, and then performs two rounds of hierarchical PBFT consensus.

[0156] The distance similarity measure between nodes is Ω similarity (m,n), where Since the on-duty node is a special ordinary node,

[0157]

[0158] in,

[0159]

[0160] Angle with the xy plane The size is:

[0161]

[0162] In formula (29), (x m ,y m , z m ) represents the coordinates of node m, (x n ,y n , z n ) represents the coordinates of node n, L is Ω similarityThe projection distance of (m, n) on the xy plane. The mean point of sample features in each cluster is

[0163]

[0164] In formula (31), n′ represents the total number of nodes in the cluster. Represents any node in the cluster Coordinates of . Use formula (29) to calculate the distance similarity measure between each node in the cluster and the sample feature mean point, and find At this time, the node that obtains the minimum value is the central node of the cluster, and it is used as the master node / on-duty node of the PBFT consensus algorithm in the scenario of this embodiment.

[0165]

[0166]

[0167] The reason why the agglomerative hierarchical clustering algorithm (AHC) is selected to optimize and improve the PBFT in this embodiment is that AHC has high clustering accuracy, high anti-interference ability for abnormal data, low noise sensitivity, and high algorithm efficiency, as shown in Table 2. Therefore, this embodiment uses the AHC algorithm to target and cluster a large number of network consensus nodes. Figure 2 A consensus hierarchical model of alliance chain network nodes was built. The lower layer of the hierarchical model consists of clusters. A cluster is a collection of sub-consensus follower nodes, and the number of nodes in each cluster must be greater than three. All clusters conduct internal PBFT consensus in parallel; the upper layer consists of M cluster center nodes that constitute the on-duty consensus master node set. Through the PBFT consensus algorithm, all on-duty nodes reach a final consensus at this layer to achieve data sharing. In other words, improving the PBFT consensus algorithm through AHC is conducive to improving the efficiency of EMRs data sharing and circulation in the medical and health public alliance chain network, promoting the interoperability of medical systems and related institutions, and alleviating the problem of data islands.

[0168] Table 2: Comparison of clustering algorithm performance

[0169]

[0170]

[0171] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A distributed sharing model of electronic medical records based on hybrid alliance chain and Stackelberg game, characterized in that: include: S1, based on blockchain technology, record and manage electronic medical records in the medical and health data management system, establish a mathematical programming model for the interaction between ordinary nodes and on-duty nodes, establish an EMRs data management system based on a hybrid alliance chain, and model the interaction between on-duty nodes and ordinary nodes by introducing Stackelberg game; S2, when the electronic medical record sharing process is executed, a pricing mechanism is established to charge for the sharing process, a secure transaction and sharing model between ordinary nodes and on-duty nodes is constructed, and the multi-leader and multi-follower Stackelberg game theory is used to balance the supply and demand relationship of electronic medical records; S3,implementation of electronic medical record secure transaction and sharing model strategy based on ADMM algorithm; S4, train the PBFT consensus algorithm based on agglomerative hierarchical clustering.

2. The electronic medical record distributed sharing model based on hybrid alliance chain and Stackelberg game according to claim 1, characterized in that: In S1, a medical and health data management scenario based on alliance chain is established, including four heterogeneous blockchain networks: community residents blockchain network, fitness club blockchain network, community hospital blockchain network, and large hospital blockchain network, forming a medical and health public alliance chain network; The medical and health data management scenario based on the alliance chain mainly includes the following roles: certificate authorities, community residents, fitness clubs, community hospitals, large hospitals, on-duty nodes, data owners, data users, and medical coins.

3. The electronic medical record distributed sharing model based on hybrid alliance chain and Stackelberg game according to claim 1, characterized in that: In S1, a distributed sharing model of EMRs based on a hybrid alliance chain and Stackelberg game is established. Ordinary nodes in the blockchain network are regarded as followers in the Stackelberg game model, and on-duty nodes are regarded as leaders in the Stackelberg game model. EMRs resources are traded between leaders and followers in the Stackelberg game model.

4. The electronic medical record distributed sharing model based on hybrid alliance chain and Stackelberg game according to claim 1, characterized in that: In S2, the EMRs data resource transactions between ordinary nodes / data users and on-duty nodes are modeled based on the multi-leader and multi-follower Stark game theory. The on-duty nodes set the price of EMRs data resources with the goal of maximizing profits. Based on the price set by the on-duty node, ordinary nodes determine the demand for EMRs data resources with the goal of maximizing utility.

5. The electronic medical record distributed sharing model based on hybrid alliance chain and Stackelberg game according to claim 4 is characterized by: In the multi-leader and multi-follower EMRs distributed sharing model based on the hybrid alliance chain, ordinary nodes i and on-duty nodes j are assumed to be rational agents, and an incentive mechanism is established.

6. The electronic medical record distributed sharing model based on hybrid alliance chain and Stackelberg game according to claim 1, characterized in that: In S3, Stackelberg game based on ADMM algorithm; Phase 1: Each ordinary node observes the prices published by all on-duty nodes at the beginning of each iteration and decides to purchase EMRs data resources from one of the on-duty nodes; The on-duty node repeatedly updates the value of the EMRs data resource based on the augmented Lagrangian method: Phase 2: The on-duty nodes in the healthcare public alliance chain network know the behavior of ordinary nodes and predict the demand for data resources. Each on-duty node updates the price in the strategy space through the ADMM algorithm.

7. The electronic medical record distributed sharing model based on hybrid alliance chain and Stackelberg game according to claim 1, characterized in that: In S3, based on the AHC-PBFT consensus algorithm, the two most similar data points among all data points are combined and this process is repeated. AHC determines the similarity between the data points of each category by calculating the distance between them and all data points.

8. The electronic medical record distributed sharing model based on hybrid alliance chain and Stackelberg game according to claim 7, characterized in that: In AHC, a part of network consensus nodes is artificially generated randomly. Each common node has EMRs transaction request information. By standardizing indicators and similarity measurement criteria, any number of clusters specified by the user is obtained.

Images