[go: up one dir, main page]

CN109300540B - A privacy-preserving medical service recommendation method in an electronic medical system - Google Patents

A privacy-preserving medical service recommendation method in an electronic medical system Download PDF

Info

Publication number
CN109300540B
CN109300540B CN201811232958.8A CN201811232958A CN109300540B CN 109300540 B CN109300540 B CN 109300540B CN 201811232958 A CN201811232958 A CN 201811232958A CN 109300540 B CN109300540 B CN 109300540B
Authority
CN
China
Prior art keywords
user
doctor
server
feedback
similarity
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Expired - Fee Related
Application number
CN201811232958.8A
Other languages
Chinese (zh)
Other versions
CN109300540A (en
Inventor
徐畅
王家琛
祝烈煌
张川
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Beijing Institute of Technology BIT
Original Assignee
Beijing Institute of Technology BIT
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Beijing Institute of Technology BIT filed Critical Beijing Institute of Technology BIT
Priority to CN201811232958.8A priority Critical patent/CN109300540B/en
Publication of CN109300540A publication Critical patent/CN109300540A/en
Application granted granted Critical
Publication of CN109300540B publication Critical patent/CN109300540B/en
Expired - Fee Related legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H40/00—ICT specially adapted for the management or administration of healthcare resources or facilities; ICT specially adapted for the management or operation of medical equipment or devices
    • G16H40/20—ICT specially adapted for the management or administration of healthcare resources or facilities; ICT specially adapted for the management or operation of medical equipment or devices for the management or administration of healthcare resources or facilities, e.g. managing hospital staff or surgery rooms
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F21/00—Security arrangements for protecting computers, components thereof, programs or data against unauthorised activity
    • G06F21/60—Protecting data
    • G06F21/62—Protecting access to data via a platform, e.g. using keys or access control rules
    • G06F21/6218—Protecting access to data via a platform, e.g. using keys or access control rules to a system of files or objects, e.g. local or distributed file system or database
    • G06F21/6245—Protecting personal data, e.g. for financial or medical purposes

Landscapes

  • Engineering & Computer Science (AREA)
  • Health & Medical Sciences (AREA)
  • General Health & Medical Sciences (AREA)
  • Bioethics (AREA)
  • General Business, Economics & Management (AREA)
  • Theoretical Computer Science (AREA)
  • Business, Economics & Management (AREA)
  • Medical Informatics (AREA)
  • Epidemiology (AREA)
  • Public Health (AREA)
  • Primary Health Care (AREA)
  • Databases & Information Systems (AREA)
  • Computer Hardware Design (AREA)
  • Computer Security & Cryptography (AREA)
  • Software Systems (AREA)
  • Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Biomedical Technology (AREA)
  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
  • Medical Treatment And Welfare Office Work (AREA)

Abstract

本发明涉及一种电子医疗系统中的隐私保护医疗服务推荐方法，属于医疗服务推荐以及隐私保护技术领域。该推荐方法主要是根据用户需求与医生信息的相似度以及医生的信誉分数来进行推荐，具体来说包括相似度计算、医生推荐以及医生信誉分数计算三部分。相似度计算即服务器根据用户需求向量与医生个人信息的向量进行相似度计算；医生推荐即服务器根据用户需求与医生信息的相似度和医生的相似度这两个因素进行医生推荐；医生信誉分数计算即服务器处理用户对医生的反馈评分并更新医生的信誉分数。本发明可实现密文下的相似度和信誉分数计算，实现了用户个人信息在推荐过程中的隐私保护。

Figure 201811232958

The invention relates to a privacy protection medical service recommendation method in an electronic medical system, belonging to the technical field of medical service recommendation and privacy protection. The recommendation method is mainly based on the similarity between the user's needs and the doctor's information and the doctor's reputation score, and specifically includes three parts: similarity calculation, doctor recommendation, and doctor's reputation score calculation. Similarity calculation means that the server performs similarity calculation according to the vector of user demand and doctor's personal information; doctor recommendation means that the server recommends doctors based on the similarity between user demand and doctor's information and the similarity of doctors; doctor's reputation score calculation That is, the server processes the user's feedback score on the doctor and updates the doctor's reputation score. The invention can realize the calculation of similarity and reputation score under ciphertext, and realize the privacy protection of user's personal information in the recommendation process.

Figure 201811232958

Description

Privacy protection medical service recommendation method in electronic medical system
Technical Field
The invention relates to a privacy protection medical service recommendation method in an electronic medical system, and belongs to the technical field of medical service recommendation and privacy protection.
Background
With the continuous development and progress of electronic medical systems and recommendation systems, online electronic medical service recommendations have become an indispensable part of daily life. According to the information of different users and doctors, the online medical recommendation server can find a proper doctor for each patient user. In particular, it is crucial how to efficiently match users to the appropriate doctors and how to ensure privacy of the patient and doctor information in the process that patient users submit their needs to a server, which matches the users' needs to the personal information of the doctors.
Many existing researches are struggling for the performance of a doctor recommendation system, a part of recommendation systems use credibility and a reputation score as recommendation bases, the credibility and the reputation score are a reflection of the service quality of a service provider, and a server recommends the service provider with a high reputation score to a user; some other recommendation systems recommend the user in consideration of user needs or interests, and although these recommendation systems can recommend the user, only a single factor, such as user needs, is considered as a basis for recommendation, which may have an effect on the accuracy of the recommendation. For example, in some articles, a patient matches his or her own needs with doctor's information to find a suitable doctor, but the doctor's personal information is uploaded to a server by the doctor himself or herself, and the doctor's information cannot objectively reflect the doctor's service quality, so that the matching result between the patient and the doctor may be inaccurate, and therefore, a parameter is needed to objectively evaluate the doctor's service to make the recommendation result more accurate.
In addition, with the reputation score as the basis for recommendation, we need to use the feedback of users to the service to calculate the reputation score, because different users have different backgrounds and histories, so the quality of the feedback they provide is different, and worse, some malicious users may give malicious evaluation to reduce the reputation of doctors, and for the above case, the feedback of different users should be given different weights for calculation. In view of the sensitivity of the user information, the user and doctor information should be privacy protected during the recommendation process. In recent research on recommendation systems for privacy protection, methods for privacy protection can be divided into three types, the first is a security-based multi-party computing method, which consumes a large amount of computing resources in the process of protecting privacy; the second method is to use random numbers to disturb sensitive data, so that privacy is protected, but the disturbed data can affect recommendation accuracy to some extent; a third method is to use pseudonyms to protect the user's real ID to obtain privacy protection. But none of the three privacy preserving methods in total are applicable to our medical service recommendation scenario.
In summary, the mechanism and privacy protection method of the existing recommendation system cannot meet the requirements of the medical service recommendation scene.
Disclosure of Invention
The invention aims to solve the difficult problems of recommendation accuracy and privacy in an online medical service recommendation system, and provides a privacy protection medical service recommendation method in an electronic medical system.
The core idea of the privacy protection medical service recommendation method is as follows: the recommendation mechanism is based on similarity and doctor reputation, and improves the accuracy of recommendation by considering a plurality of factors of which similarity and reputation are main; meanwhile, an algorithm of privacy protection is adopted for calculating the similarity and the doctor reputation, and the method specifically comprises the following steps: when the user demand vector is matched with the doctor information vector, adding disturbance to the vectors of the user and the doctor and processing the vectors, and adopting similarity calculation of privacy protection to ensure that the server cannot obtain the data of the user and the doctor in a plaintext; in addition, user feedback scores are aggregated under a ciphertext, a truth value discovery technology is utilized to process a large number of user feedback scores to calculate a truth value, and the credit score of a doctor is calculated by using Dirichlet distribution for the truth values fed back by users in a plurality of time periods.
The privacy protection medical service recommendation method comprises two parts of user requirement matching and doctor reputation score calculation;
the user requirement matching method comprises the following steps:
step 1: a user sends a medical service request to a server;
step 2: the server verifies the user identity and gives feedback;
and step 3: the user sends the own demand attribute vector and the similarity threshold value to the server;
wherein, the demand attribute vector of the user mainly comprises the illness state of the user, the department of the searched doctor and the like;
wherein, the similarity threshold is a threshold which is acceptable for the similarity of the user;
and 4, step 4: the server carries out similarity calculation according to the demand attribute vector in the step 3 and the personal information vector of the doctor stored in the server, and similarity calculation formulas are shown as a formula (1) and a formula (2);
Figure BDA0001837558120000031
Figure BDA0001837558120000032
where dis is the distance between two perturbed vectors, sim is the similarity between two perturbed vectors, aiAs a user demand vector, biIs a doctor information vector, ai' As the user demand vector after adding the disturbance, bi'is the perturbed doctor information vector, s' is the shared secret key between the trust authority and the user, t, r1、r2Is a random number, mod is a remainder taking operator, and is a multiplication operator;
and 5: the server screens out a part of doctors according to the threshold value of the similarity acceptable by the user;
step 6: in a part of screened doctors, the server selects the doctor with the highest credit score according to the credit score of the doctor and recommends the doctor to the user;
the recommendation basis is the credit score and the similarity of doctors;
and 7: the user performs medical service according to the doctor recommended by the server;
and 8: after the medical service is finished, the user feeds back according to the service quality of the doctor and uploads a feedback score to the server;
wherein the feedback score is a parameter of the reputation score calculation;
and step 9: the server collects feedback scores of a plurality of users for doctors and recalculates the reputation scores of the doctors based on the obtained feedback scores;
and calculating the doctor reputation, which comprises the following steps:
step A), after the user receives the medical service, feeding back according to the service quality of a doctor;
step B) processing the feedback scores of a plurality of users, and calculating a true value reflecting the service quality of a doctor;
the real value calculation process of the step B) is mainly divided into two stages of weight updating and true value updating;
wherein, the weight value updating stage dynamically updates the feedback score weight value of each user for the feedback score of each user, and the method comprises the following steps:
step I: initializing an iteration count value to be 1, and setting the maximum iteration number to be 10;
step II: the server distributes truth x to N users*Judging whether the iteration count value is 1, if so, the true value is a random initial value; otherwise, the true value of the iteration is the true value calculated by the last iteration;
step III: each user receives a true value x*And then calculates the distance Dis between the feedback and the truth value based on the formula (1)iThen the user is to DisiEncrypting the data by the following formula (3), and encrypting the encrypted data CijAnd CijUploading to a server;
Figure BDA0001837558120000041
wherein, CijIs the encrypted distance value, Dis, of the user i in the jth iterationiDistance between user's feedback and truth, cijThe encrypted data uploaded by the users are represented by i as the ith user, j as the jth iteration, g and h as encryption parameters, rijIs a random number, and n is an encryption parameter;
step IV: after the server receives the data of N users, the encrypted distance values are aggregated by using the following formula (4) and the encrypted parameters are aggregated by using the following formula (5), and an aggregation result is output:
Figure BDA0001837558120000042
Figure BDA0001837558120000043
step V: the server uses its own secret s2Partially decrypting the aggregation result output in the step IV by using a formula (6), and sending the decrypted data to each user;
Figure BDA0001837558120000051
wherein, Cj' aggregated data decrypted for server part, s2Is a secret key of the server, s1A key for the user;
step VI: when the user receives the data from the server, the user uses its own secret key s1Decrypting according to a formula (7) and a formula (8) and calculating the weight of the user;
Figure BDA0001837558120000052
Figure BDA0001837558120000053
wherein, Cj"user decrypted aggregated data, sumdDis for all usersiSum, wiIs the weight of the user;
so far, from step I to step VI, the weight updating stage is completed, namely the weights of all users are updated;
the true value updating is performed after the weight value updating stage, and specifically comprises the following steps:
step (1): the user disturbs the weight of the user according to a formula (9) and uploads the weight to the server;
Figure BDA0001837558120000054
wherein, Wij,1And Wij,2The weight value of the user added with the disturbance information is obtained;
step (2): the server aggregates the disturbed weights of all users according to the formula (4) and updates the truth value x according to the formula (10)*；
Figure BDA0001837558120000061
And C) based on the output of the step B), the server calculates the credit score of the doctor by using Dirichlet distribution according to the real values obtained by a plurality of time periods.
Advantageous effects
Compared with the conventional medical service recommendation method, the privacy protection medical service recommendation method in the electronic medical system has the following beneficial effects:
1. compared with the traditional medical service recommendation method, the method takes two factors of similarity and reputation score as the basis of recommendation, so that the recommendation result is more accurate and reasonable;
2. during similarity calculation, the server performs similarity calculation between the user requirements and personal information of doctors under the ciphertext, so that the calculation efficiency is improved, and the privacy of the user and doctor data is also guaranteed;
3. for true value calculation of a plurality of user feedback data, the efficiency of calculating true values of the user feedback data is improved.
Drawings
FIG. 1 is a model diagram of a recommendation system upon which a method for privacy preserving medical service recommendation in an electronic medical system of the present invention relies;
FIG. 2 is a graph comparing the time taken for recommendation in the PPMR recommendation and the FSSR recommendation of the present invention;
FIG. 3 is a comparison of the time taken for the actual value calculations in the PPMR and FSSR recommendations of the present invention;
FIG. 4 is a comparison graph of the time taken for updating the weights in the PPMR recommendation and the FSSR recommendation of the present invention;
FIG. 5 is a graph comparing the time taken for the real value update in the PPMR recommendation and the FSSR recommendation of the present invention.
Detailed Description
The following detailed description of the embodiments of the present invention will be made with reference to the accompanying drawings.
Example 1
The embodiment of the invention describes a system model diagram based on a privacy protection medical service recommendation method in an electronic medical system in detail, as shown in fig. 1.
As can be seen from fig. 1, when a patient user wants to apply for medical services, the user can send a medical service request to the server through a smart device such as a mobile phone, a tablet, a computer, etc., and after the server receives the request, the user sends the request of the user for a doctor and an acceptable threshold for similarity to the server.
The server calculates the similarity between the user and each doctor by using the safe similarity matching method provided by the invention according to the requirements of the user and the information of the doctors, then screens out a part of doctors with the similarity meeting the threshold condition according to the threshold value which is provided by the user and acceptable for the similarity, and finally, for the screened part of doctors, the server recommends the doctors with the highest credit score to the user. After the recommendation is completed, the user goes to a medical center to perform medical service, after the medical service is completed, the user gives feedback according to the service quality of doctors and uploads the feedback to the server, and the server comprehensively calculates the credit score of the doctors through the truth value calculation method and the credit score calculation of Dirichlet distribution for the feedback of a plurality of users to the doctors in the same time period. Thus, the entire recommendation process is completed.
Example 2
Compared with the traditional medical service recommendation method such as the medical service recommendation method in the FSSR paper of Cheng Huang et al, the FSSR recommendation method only carries out doctor recommendation according to the factor of similarity. Only one factor of similarity is considered for recommendation, and the user can be matched with the doctor meeting the basic requirement, but the quality of service of the doctor cannot be predicted. According to the recommendation method, doctor recommendation is carried out according to two factors of similarity and doctor reputation scores, a server screens out a batch of doctors meeting user requirements according to the similarity, and then the doctor with the highest reputation score is selected from the batch of doctors to recommend to the user, wherein the higher reputation score indicates that the service quality of the doctor is better. In terms of algorithm efficiency, as shown in fig. 2, when the number of doctors is 500, the time used by the recommendation method of the FSSR is 8.301s, and the time used by the recommendation method in our PPMR scheme is 8.302s, which indicates that the algorithm efficiency is not affected while the recommendation is performed by considering a plurality of factors, and the algorithm efficiency is similar to the FSSR efficiency. In the aspect of privacy protection, the scheme processes attribute vectors of users and information vectors of doctors entering disturbing information, so that calculation of vector similarity is completed under a ciphertext, and a server cannot obtain real information of the users and the doctors during calculation, so that the privacy of the users and the doctors is protected.
Example 3
For true value calculation of multiple user feedback data, the scheme adopts a Modified Paillier encryption algorithm to process the feedback scoring data of the user, and compared with a Threshold Paillier encryption algorithm used in a PPTD scheme proposed by Miao et al to process the user data, the efficiency of user feedback data aggregation and true value calculation is improved. As shown in fig. 3, when the number of users is 500, the time taken for the PPTD scheme to calculate the real value is 652.23s, while the time taken for the our PPMR scheme to calculate the real value is 14.59s, compared to that, the efficiency of the real value calculation is improved by the our scheme. Specifically, the truth discovery technique of privacy protection is divided into two stages of weight update and truth update, as shown in fig. 4 and 5, the time used by our PPMR scheme is lower than that of PPTD scheme in the weight update stage and the truth update stage, i.e. the algorithm efficiency is high.
In summary, the above description is only a preferred embodiment of the present invention, and is not intended to limit the scope of the present invention. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims (4)

1.一种电子医疗系统中的隐私保护医疗服务推荐方法，其特征在于：包括用户需求匹配以及医生信誉分数计算两部分；1. a privacy protection medical service recommendation method in an electronic medical system, it is characterized in that: comprise two parts of user demand matching and doctor credit score calculation; 其中，用户需求匹配，包括如下步骤：Among them, the matching of user requirements includes the following steps: 步骤1：用户向服务器发出医疗服务请求；Step 1: The user sends a medical service request to the server; 步骤2：服务器验证用户身份，并给予反馈；Step 2: The server verifies the user's identity and gives feedback; 步骤3：用户将自己的需求属性向量和相似度阈值发送给服务器；Step 3: The user sends his demand attribute vector and similarity threshold to the server; 步骤4：服务器根据步骤3的需求属性向量和服务器中所存储的医生的个人信息向量进行相似度计算，相似度计算公式如公式(1)、公式(2)所示；Step 4: The server performs similarity calculation according to the demand attribute vector of step 3 and the personal information vector of the doctor stored in the server, and the similarity calculation formula is shown in formula (1) and formula (2);
Figure FDA0003158755180000011
Figure FDA0003158755180000011
Figure FDA0003158755180000012
Figure FDA0003158755180000012
 其中，dis为两个加入扰动后的向量之间的距离，sim为两个加入扰动后的向量之间的相似度，ai为用户需求向量，bi为医生信息向量，a′i为加入扰动后的用户需求向量，b′i为加入扰动后的医生信息向量，s′为信任机构和用户之间的共享密钥，t、r1、r2为随机数，mod为取余数操作符，*为乘法操作符；Among them, dis is the distance between the two perturbed vectors, sim is the similarity between the two perturbed vectors, a i is the user demand vector, b i is the doctor information vector, and a′ i is the added User demand vector after perturbation, b' i is the doctor information vector after perturbation, s' is the shared key between the trust organization and the user, t, r 1 , r 2 are random numbers, mod is the remainder operator , * is the multiplication operator; 步骤5：服务器根据用户可接受的相似度阈值筛选出一部分医生；Step 5: The server screens out some doctors according to the user-acceptable similarity threshold; 步骤6：在筛选出的一部分医生中，服务器根据医生的信誉分数选出信誉分数最高的医生推荐给用户；Step 6: Among the selected doctors, the server selects the doctor with the highest reputation score and recommends it to the user according to the doctor's reputation score; 步骤7：用户根据服务器推荐的医生进行医疗服务；Step 7: The user provides medical services according to the doctor recommended by the server; 步骤8：当完成医疗服务后，用户将根据医生的服务质量进行反馈，上传反馈评分到服务器；Step 8: After completing the medical service, the user will give feedback according to the doctor's service quality, and upload the feedback score to the server; 步骤9：服务器收集多个用户对医生的反馈评分，并基于获得的反馈评分重新计算医生的信誉分数；Step 9: The server collects feedback scores from multiple users on the doctor, and recalculates the doctor's reputation score based on the obtained feedback scores; 医生信誉计算，包括如下步骤：The calculation of doctor's reputation includes the following steps: 步骤A)当用户接受完医疗服务后，根据医生的服务质量进行反馈；Step A) After the user has received the medical service, feedback is given according to the service quality of the doctor; 步骤B)对于多个用户的反馈评分进行处理，计算出一个反映医生服务质量的真实值；Step B) processing the feedback scores of multiple users, and calculating a true value reflecting the doctor's service quality; 其中，真实值使用真值发现算法处理同一时间段的多个用户反馈数据并计算出来的，步骤B)的真实值计算过程，又主要分为权值更新和真值更新两个阶段；Wherein, the true value is calculated by using the true value discovery algorithm to process multiple user feedback data in the same time period, and the true value calculation process in step B) is mainly divided into two stages: weight update and true value update; 其中，权值更新阶段对于每个用户的反馈评分，动态的更新每个用户反馈评分的权值，包括如下步骤：Among them, the weight update stage dynamically updates the weight of each user's feedback score for each user's feedback score, including the following steps: 步骤I：初始化迭代计数值为1，并设置最大迭代次数为10次；Step I: Initialize the iteration count to 1, and set the maximum number of iterations to 10; 步骤II：服务器给N个用户分发真值x*，判断迭代计数值是否为1，若是则真值为随机初始值；否则，本次迭代的真值为上一次迭代所计算出的真值；Step II: The server distributes the true value x * to N users, and judges whether the iteration count value is 1, and if so, the true value is a random initial value; otherwise, the true value of this iteration is the true value calculated by the previous iteration; 步骤III：每个用户接收真值x*，再基于公式(1)计算出自己的反馈与真值之间的距离Disi，之后用户对Disi用如下公式(3)进行加密，将加密后的Cij和cij上传到服务器；Step III: Each user receives the true value x * , and then calculates the distance Dis i between his feedback and the true value based on formula (1), then the user encrypts Dis i with the following formula (3), The C ij and c ij are uploaded to the server;
Figure FDA0003158755180000021
Figure FDA0003158755180000021
 其中，Cij为第j次迭代中用户i的加密后的距离值，Disi为用户的反馈与真值之间的距离，cij为用户上传的加密数据，i为第i个用户，j为第j次迭代，g和h为加密参数，rij为随机数，n为加密参数；Among them, C ij is the encrypted distance value of user i in the jth iteration, Dis i is the distance between the user's feedback and the true value, c ij is the encrypted data uploaded by the user, i is the ith user, j is the jth iteration, g and h are encryption parameters, r ij is a random number, and n is an encryption parameter; 步骤IV：当服务器收到N个用户的数据后，利用如下公式(4)对加密后的距离值进行聚合和公式(5)对加密参数进行聚合，输出聚合结果：Step IV: After the server receives the data of N users, the following formula (4) is used to aggregate the encrypted distance values and formula (5) is used to aggregate the encryption parameters, and the aggregation result is output:
Figure FDA0003158755180000031
Figure FDA0003158755180000031
Figure FDA0003158755180000032
Figure FDA0003158755180000032
 步骤V：服务器使用自己的密钥s2对步骤IV输出的聚合结果使用公式(6)进行部分解密，并将解密后的数据发送给每个用户；Step V: the server uses its own key s 2 to partially decrypt the aggregated result output in step IV using formula (6), and sends the decrypted data to each user;
Figure FDA0003158755180000033
Figure FDA0003158755180000033
 其中，C′j为服务器部分解密后的聚合数据，s2为服务器的密钥，s1为用户的密钥；Wherein, C'j is the aggregated data partially decrypted by the server, s 2 is the server's key, and s 1 is the user's key; 步骤VI：当用户收到服务器的数据后，用户使用自己的密钥s1根据公式(7)和公式(8)解密，并求出自己的权值；Step VI: after the user receives the data from the server, the user decrypts using his own key s 1 according to formula (7) and formula (8), and obtains his own weight;
Figure FDA0003158755180000034
Figure FDA0003158755180000034
Figure FDA0003158755180000035
Figure FDA0003158755180000035
 其中，C″j用户解密后的聚合数据，sumd为所有用户的Disi之和，wi为用户的权值；Wherein, the aggregated data decrypted by users C″ j , sum d is the sum of Dis i of all users, and w i is the weight of users; 至此，从步骤I到步骤VI，完成了权值更新阶段，即更新了所有用户的权值；So far, from step I to step VI, the weight update stage has been completed, that is, the weights of all users have been updated; 真值更新在权值更新阶段之后进行，具体包括如下步骤：The truth value update is performed after the weight update stage, which includes the following steps: 步骤(1)：用户对自己的权值根据公式(9)进行扰动，并上传到服务器；Step (1): the user perturbs his own weight according to formula (9) and uploads it to the server;
Figure FDA0003158755180000041
Figure FDA0003158755180000041
 其中，Wij，1和Wij，2为加入扰动信息后的用户的权值；Among them, W ij,1 and W ij,2 are the weights of users after adding disturbance information; 步骤(2)：服务器根据公式(4)聚合所有用户的扰动后的权值并根据公式(10)更新真值x*；Step (2): the server aggregates the perturbed weights of all users according to formula (4) and updates the true value x * according to formula (10);
Figure FDA0003158755180000042
Figure FDA0003158755180000042
 步骤C)基于步骤B)的输出，服务器根据多个时间段所求出的真实值使用狄利克雷分布来计算出医生的信誉分数。Step C) Based on the output of Step B), the server uses the Dirichlet distribution to calculate the doctor's reputation score according to the real values obtained in multiple time periods. 2.根据权利要求1所述的一种电子医疗系统中的隐私保护医疗服务推荐方法，其特征在于：步骤3中，用户的需求属性向量，包括用户的病情及所寻找的医生的科室；2. The method for recommending privacy protection medical services in an electronic medical system according to claim 1, wherein in step 3, the user's demand attribute vector includes the user's condition and the department of the sought doctor; 其中，相似度阈值为用户对于相似度可接受的阈值。The similarity threshold is a threshold acceptable to the user for the similarity. 3.根据权利要求1所述的一种电子医疗系统中的隐私保护医疗服务推荐方法，其特征在于：步骤6中，推荐的依据是医生的信誉分数以及相似度。3 . The method for recommending a privacy-preserving medical service in an electronic medical system according to claim 1 , wherein in step 6 , the recommendation is based on a doctor's reputation score and similarity. 4 . 4.根据权利要求1所述的一种电子医疗系统中的隐私保护医疗服务推荐方法，其特征在于：步骤8中，反馈评分是信誉分数计算的参数。4 . The method for recommending privacy-preserving medical services in an electronic medical system according to claim 1 , wherein in step 8, the feedback score is a parameter for calculating the reputation score. 5 .
CN201811232958.8A 2018-10-23 2018-10-23 A privacy-preserving medical service recommendation method in an electronic medical system Expired - Fee Related CN109300540B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201811232958.8A CN109300540B (en) 2018-10-23 2018-10-23 A privacy-preserving medical service recommendation method in an electronic medical system

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201811232958.8A CN109300540B (en) 2018-10-23 2018-10-23 A privacy-preserving medical service recommendation method in an electronic medical system

Publications (2)

Publication Number Publication Date
CN109300540A CN109300540A (en) 2019-02-01
CN109300540B true CN109300540B (en) 2021-10-15

Family

ID=65158329

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201811232958.8A Expired - Fee Related CN109300540B (en) 2018-10-23 2018-10-23 A privacy-preserving medical service recommendation method in an electronic medical system

Country Status (1)

Country Link
CN (1) CN109300540B (en)

Families Citing this family (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110335676A (en) * 2019-07-09 2019-10-15 泰康保险集团股份有限公司 Data processing method, device, medium and electronic equipment
CN111585990B (en) * 2020-04-26 2022-04-15 安徽师范大学 A mobile crowd-sensing privacy-preserving truth value discovery method based on security summation
CN115203545A (en) * 2022-07-06 2022-10-18 华南师范大学 Health maintenance service intelligent matching method and system based on deep learning and knowledge graph
CN119207744B (en) * 2024-09-13 2025-11-18 康键信息技术（深圳）有限公司 Traffic distribution methods, devices, computer equipment and storage media

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104796475A (en) * 2015-04-24 2015-07-22 苏州大学 Social recommendation method based on homomorphic encryption
CN105592085A (en) * 2015-12-23 2016-05-18 西安电子科技大学 Privacy protection method specific to location awareness recommendation system
CN106131139A (en) * 2016-06-23 2016-11-16 暨南大学 The encryption of the floating data of a kind of cloud relational database and querying method

Family Cites Families (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US10176488B2 (en) * 2014-02-19 2019-01-08 International Business Machines Corporation Perturbation, monitoring, and adjustment of an incentive amount using statistically valuable individual incentive sensitivity for improving survey participation rate
US9747494B2 (en) * 2015-11-16 2017-08-29 MorphoTrak, LLC Facial matching system
CN107257499B (en) * 2017-07-21 2018-09-18 安徽大学 Privacy protection method in video recommendation system and video recommendation method

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104796475A (en) * 2015-04-24 2015-07-22 苏州大学 Social recommendation method based on homomorphic encryption
CN105592085A (en) * 2015-12-23 2016-05-18 西安电子科技大学 Privacy protection method specific to location awareness recommendation system
CN106131139A (en) * 2016-06-23 2016-11-16 暨南大学 The encryption of the floating data of a kind of cloud relational database and querying method

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
Modeling Background Knowledge for Privacy Preserving Medical Data Publishing;Eric Ke Wang、Binfeng Jia、Nie Ke;《2017 International Conference on Computer Systems, Electronics and Control (ICCSEC)》;20180827;全文第1-6页 *
隐私保护的推荐系统的设计与实现;刘肖;《中国优秀硕士学位论文全文数据库（电子期刊）信息科技辑》;20180415;全文第18-38页 *

Also Published As

Publication number Publication date
CN109300540A (en) 2019-02-01

Similar Documents

Publication Publication Date Title
US11048690B2 (en) Contribution of multiparty data aggregation using distributed ledger technology
JP7471445B2 (en) Privacy-preserving machine learning for content delivery and analytics
US11580417B2 (en) System and method for processing data and managing information
CN109300540B (en) A privacy-preserving medical service recommendation method in an electronic medical system
KR101843340B1 (en) Privacy-preserving collaborative filtering
WO2021120676A1 (en) Model training method for federated learning network, and related device
US20160012238A1 (en) A method and system for privacy-preserving recommendation to rating contributing users based on matrix factorization
US9438421B1 (en) Supporting a fixed transaction rate with a variably-backed logical cryptographic key
US20230274183A1 (en) Processing of machine learning modeling data to improve accuracy of categorization
Neera et al. Private and utility enhanced recommendations with local differential privacy and Gaussian mixture model
Sodagari Trends for mobile IoT crowdsourcing privacy and security in the big data era
Zhou et al. A lightweight matrix factorization for recommendation with local differential privacy in big data
CN111104434B (en) Electronic medical recommendation method based on privacy protection multi-level attribute similarity
US20240135024A1 (en) Method and system for data communication with differentially private set intersection
CN112487456A (en) Federal learning model training method and system, electronic equipment and readable storage medium
WO2024025744A1 (en) Integration of anonymized, member-driven cloud-based groups and content delivery services that collect individual information about content interactions without compromising identities of group members
CN117077816B (en) Training method and system of federal model
Checco et al. Blc: Private matrix factorization recommenders via automatic group learning
CN120430645A (en) A method, device, equipment and medium for analyzing enterprise operating costs based on privacy computing
Wu et al. FCER: A Federated Cloud-Edge Recommendation Framework With Cluster-Based Edge Selection
Silva et al. Privacy-preserving multi-party clustering: An empirical study
US12081541B2 (en) Device-side federated machine learning computer system architecture
Zhu et al. PFED-AGG: A personalized private federated learning aggregation algorithm
Zhao et al. A scalable algorithm for privacy-preserving item-based top-N recommendation
Anelli et al. Addressing Privacy in Recommender Systems with Federated Learning.

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant
CF01 Termination of patent right due to non-payment of annual fee
CF01 Termination of patent right due to non-payment of annual fee

Granted publication date: 20211015