CN111144645B - Medical information service selection method based on hybrid optimization algorithm - Google Patents

Medical information service selection method based on hybrid optimization algorithm Download PDF

Info

Publication number
CN111144645B
CN111144645B CN201911352264.2A CN201911352264A CN111144645B CN 111144645 B CN111144645 B CN 111144645B CN 201911352264 A CN201911352264 A CN 201911352264A CN 111144645 B CN111144645 B CN 111144645B
Authority
CN
China
Prior art keywords
service
attribute
transaction
particle
abstract
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.)
Active
Application number
CN201911352264.2A
Other languages
Chinese (zh)
Other versions
CN111144645A (en
Inventor
樊昭磊
吴军
杨万春
张伯政
李涛
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.)
Zhongyang Health Technology Group Co ltd
Original Assignee
Shandong Msunhealth Technology Group Co Ltd
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 Shandong Msunhealth Technology Group Co Ltd filed Critical Shandong Msunhealth Technology Group Co Ltd
Priority to CN201911352264.2A priority Critical patent/CN111144645B/en
Publication of CN111144645A publication Critical patent/CN111144645A/en
Application granted granted Critical
Publication of CN111144645B publication Critical patent/CN111144645B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06QINFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00Administration; Management
    • G06Q10/04Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06NCOMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00Computing arrangements based on biological models
    • G06N3/004Artificial life, i.e. computing arrangements simulating life
    • G06N3/006Artificial life, i.e. computing arrangements simulating life based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO]
    • GPHYSICS
    • G16INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16HHEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H40/00ICT 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
    • G16H40/20ICT 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

Landscapes

  • Engineering & Computer Science (AREA)
  • Business, Economics & Management (AREA)
  • General Business, Economics & Management (AREA)
  • Health & Medical Sciences (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Biomedical Technology (AREA)
  • Strategic Management (AREA)
  • General Health & Medical Sciences (AREA)
  • Human Resources & Organizations (AREA)
  • Economics (AREA)
  • General Physics & Mathematics (AREA)
  • Evolutionary Computation (AREA)
  • Public Health (AREA)
  • Computational Linguistics (AREA)
  • Data Mining & Analysis (AREA)
  • Artificial Intelligence (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Biophysics (AREA)
  • Development Economics (AREA)
  • Primary Health Care (AREA)
  • Game Theory and Decision Science (AREA)
  • Medical Informatics (AREA)
  • Epidemiology (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Marketing (AREA)
  • Operations Research (AREA)
  • Quality & Reliability (AREA)
  • Tourism & Hospitality (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

A medical information service selection method based on a hybrid optimization algorithm. By the method, the multidimensional attribute of the medical information service is extracted, and a model meeting the user concurrency requirement and multi-granularity is established. Compared with the existing model, the model considers the transaction attribute of the medical information service and designs the probability-based adaptive priority. A multi-granularity recursive decomposition method is adopted to obtain a combination scheme. The method solves the problem of service selection by using a hybrid optimization algorithm, mainly designs a particle encoding mode, and fuses a particle swarm, a cross mutation operator with priority and a simulated annealing method to obtain the optimal service. The cross mutation operator with the priority can avoid solutions which do not meet the transaction attributes, the particle swarm algorithm can realize rapid global search, the simulated annealing method can avoid falling into a local optimal solution, different methods complement each other, and the optimal effect is obtained by making the best of the differences.

Description

Medical information service selection method based on hybrid optimization algorithm
Technical Field
The invention relates to the technical field of medical informatization and the technical field of artificial intelligence, and designs a medical information service selection method based on a hybrid optimization algorithm.
Background
With the development of informatization of the medical industry and the introduction of the idea of all services, service-oriented computing and cloud computing have become mainstream computing modes. The service-oriented computing is a computing method for constructing a distributed software system by using a service as a basic component and by using a combination technology. The service combination is a main means for realizing service resource integration in a service-oriented architecture, and can synthesize related services scattered on the network into a service with large granularity and stronger availability. When selecting a service, a user puts not only functional constraints on the combined service, but also higher requirements on non-functional attributes (such as price, response time, reliability, availability and the like), and the limiting conditions are the basis of service selection. Among the numerous candidate service resources, how to select a high-quality composite service that meets the user's needs is an important goal. The existing service selection method has defects in the aspects of accuracy and time complexity, so that a more effective method is needed to be adopted to optimize candidate services, the calculation time is reduced, and the quality of service selection is ensured.
An important application of intelligent medical treatment is to provide medical services effectively in real time, and the key point is to design a medical information service combination meeting the constraint conditions. The existing medical information service selection model is based on the assumption of single demand, and in practical application, the demands arrive simultaneously in a smaller time interval, namely, the concurrent demand situation. And the traditional medical information service combination is lack of consideration in the granularity level, and the possibility of all granularity combinations cannot be constructed, so that the diversified requirements of users cannot be met.
Disclosure of Invention
The invention provides a medical information service selection method based on a hybrid optimization algorithm, which solves the problem of medical information service combination by analyzing medical information services and utilizing an artificial intelligence technology and provides the medical information service selection method based on the hybrid optimization algorithm.
The invention is realized by the following measures:
1. multidimensional attributes of the medical information service are extracted, including quality of service attributes and transaction attributes. Quality of service (QoS) attributes are composed of a number of non-functional attributes including service price, service response time, service availability and service reliability. The transaction attributes of a service mainly include the following categories: a compensable transaction attribute c, a retriable transaction attribute r, a pivot transaction attribute p, and a compensable transaction attribute cr.
A probabilistic priority is set for transaction attributes of the service. The priority is dynamically changed according to the changing situation of the candidate service.
2. And establishing a model meeting the concurrency requirement and multi-granularity according to the extracted multi-dimensional attributes of the medical information service. The user requirement i corresponds to an Abstract service process set (Abstract service process)set)aspsi. Abstract service flow set aspsiBy quadruplets<Si,Gi,Ii,Pi>Is shown in which S isiRepresenting a collection of abstract services, GiIs a set of granularities, IiIs the structure of an abstract service flow, PiRepresenting a possible abstract service flow. Each task of the abstract service flow may be completed by a different candidate service having different attribute parameter values, forming a different service composition scheme.
Each requirement i corresponds to an abstract service flow set aspsi={aspi1,aspi2,…,aspik}. Asps in abstract service flow setiEach item of (2) represents an abstract service flow, and a concrete service is selected for each abstract service to obtain a corresponding concrete service flow set:
cspsi={cspi1,cspi2,...,cspik}
each specific service flow cspipThe attribute parameters of (2) constitute a vector:
{Q1(cspip),Q2(cspip),Q3(cspip),Q4(cspip)}
wherein each dimension attribute of the vector represents a service process cspipResponse time, price, reliability and availability. Calculating each specific service flow cspipThe QoS value of (1), then the specific service flow set cspsiThe maximum value of the QoS values of the specific service flows is taken, and the calculation formula is as follows:
Figure GDA0003009565410000021
wherein wj(wj> 0) represents a user preference,
Figure GDA0003009565410000022
all candidate composite service flows representing aspsi are attributed in dimension jThe maximum value of (a) above (b),
Figure GDA0003009565410000023
represents the minimum value, Q, of all candidate composite service flows on the j-dimension attributeip,jRepresentation cspipThe j-th dimension of the attribute value. The objective function of the service optimization selection facing the concurrency requirement and the multi-granularity is shown as formula (2), wherein m is the quantity of the concurrency requirements.
Figure GDA0003009565410000024
3. The combination scheme with different granularities is obtained by a hierarchical recursive decomposition method.
3.1. Setting an abstract service set, a granularity highest level t and an abstract service flow structure;
3.2. constructing a to-be-decomposed combined service scheme by the abstract service at the highest level t, and adding the scheme into an abstract service flow list ProcessList;
3.3. searching whether decomposable services exist in the combined scheme to be decomposed, if so, executing the step 3.4, otherwise, turning to the step 3.6;
3.4. decomposing the service of the t layer into the t-1 layer to obtain a plurality of combined service schemes, and adding the schemes into the ProcessList; 3.5. forming a combination scheme to be decomposed by the service of the t-1 layer, returning to the step 3.3 when t is t-1;
3.6. and returning a service combination scheme list ProcessList meeting the user requirements.
4. And solving the model by adopting a hybrid optimization algorithm. The algorithm combines particle swarm, cross mutation operators with priorities and a simulated annealing method.
4.1 coding strategy. And coding by adopting an integer array, wherein the length of the array is the number of services contained in the individual, and the value of each element in the array is the identifier of the concrete service instance corresponding to the abstract service binding.
4.2 in the iterative process of the algorithm, the particles update the positions thereof according to the individual extreme points and the global extreme points. IndividualsExtreme point ppbestThe position of the optimal solution currently found by the particle is represented; global extreme point pgbestIndicating the location of the best solution currently found by the population.
4.3 introducing a crossover and mutation operator with priority into the algorithm to update the population. Let pj(t) is the jth particle of the tth generation, ppbest(t) is the individual optimal solution for the tth generation corresponding to the jth particle, pgbest(t) is the global optimal solution for the t-th generation.
(1)pj(t) and pgbest(t) performing a prioritized crossover operation to generate new particles t1And t2。pj(t) and ppbest(t) performing a prioritized crossover operation to generate new particles t3And t4
(2)pj(t) performing a mutation operation with priority to generate a new particle t5. This operation can avoid premature trapping of particles in the local optimum.
(3) Calculating a new particle t1、t2、t3、t4、t5The maximum value is selected. Suppose f (t)3) When the value of (c) is maximum, f (t) is set3) Comparing with the global optimum and the individual optimum if f (t)3) Is better than the individual optimum, then ppbest(t+1)=t3(ii) a If f (t)3) Is better than the global optimum, then pgbest(t+1)=t3. Simultaneous particles t3P as the next iterationj(t+1)。
4.4 a simulated annealing algorithm is introduced into the algorithm to jump out the local optimum. If the fitness value of the new state is worse than the current optimal value, the algorithm receives the degraded solution with a certain probability, so that the simulated annealing algorithm has the capability of jumping out of the local optimal value.
4.5 hybrid optimization algorithm step:
4.5.1. initializing to generate particles meeting the constraint, setting the maximum iteration number, and initializing the initial temperature T of the simulated annealing0And a termination temperature TfSetting the attenuation coefficient alpha and the Markov chain length Lk
4.5.2. Calculating the value of fitness function f (x) of each particle, and taking the position corresponding to the fitness value as the individual optimal solution PpbestSelecting a global optimal solution P from all current individual optimal solutionsgbest
4.5.3. Setting search position x of simulated annealing algorithm0Is PgbestPerforming a mutation operation on the solution to generate a new solution and using Metropolis criteria to determine whether to accept the new solution, and repeating the process LkThen obtaining a new solution x', and then executing annealing operation to reduce the temperature Tk+1=α*Tk
4.5.4. Each particle is based on the current individual optimal solution PpbestAnd a global optimal solution PgbestTo update the velocity and position and, based on the newly generated particles, to update the individual optimal solution P for each particlepbestAnd a global optimal solution PgbestThe iteration number is increased by 1;
4.5.5. if the maximum number of iterations is reached, outputting the result, otherwise, continuing to perform step 4.5.6;
4.5.6. mixing x' with PgbestIf f (x') is better than f (P)gbest) If yes, randomly deleting one particle in the particle swarm, adding x' into the particle swarm, and returning to the step 4.5.4; if the value of f (x') is greater than f (P)gbest) If poor, let x0=PgbestAnd returning to the step 4.5.3.
The invention has the beneficial effects that: by the method, the multidimensional attribute of the medical information service is extracted, and a model meeting the user concurrency requirement and multi-granularity is established. Compared with the existing model, the model considers the transaction attribute of the medical information service and designs the probability-based adaptive priority. A multi-granularity recursive decomposition method is adopted to obtain a combination scheme. The method solves the problem of service selection by using a hybrid optimization algorithm, mainly designs a particle encoding mode, and fuses a particle swarm, a cross mutation operator with priority and a simulated annealing method to obtain the optimal service.
Drawings
FIG. 1 is a flow chart of hierarchical recursive decomposition steps
FIG. 2 is a flow chart of the steps of a hybrid optimization algorithm
Detailed Description
The invention is further illustrated with reference to the accompanying figures 1-2.
1. Multidimensional attributes of the medical information service are extracted, including quality of service attributes and transaction attributes. Quality of service (QoS) attributes are composed of a number of non-functional attributes including service price, service response time, service availability and service reliability. The transaction attributes of a service mainly include the following categories: a compensable transaction attribute c, a retriable transaction attribute r, a pivot transaction attribute p, and a compensable transaction attribute cr.
A probabilistic priority is set for transaction attributes of the service. The priority is dynamically changed according to the changing situation of the candidate service. (1) If the transaction attribute of service i is c, its alternative candidate service transaction attribute is c or cr. And if the number of the services with the attribute of c is higher than the number of the services with the attribute of cr, the probability of selecting the candidate services with the transaction attribute of c is high. (2) If the transaction attribute of service i is r, its alternative candidate service transaction attribute is r or cr. And if the number of the services with the attribute r is higher than the number of the services with the attribute cr, the probability of selecting the candidate services with the transaction attribute r is high. (3) If the transaction attribute of service i is cr, its alternative candidate service transaction attribute can only be cr. (4) If the transaction attribute of service i is p, its alternative candidate service transaction attribute is p or c or r or cr. And determining the priority of the candidate service according to the number of the services with different transaction attributes.
Such as a referral medical procedure: referral application-hospital recommendation-appointment registration-hospital navigation-disease information reporting. In this flow, there are 5 abstract medical services, each with several candidate concrete services. Such as subscription registration, different service providers, different prices for offered services, response times, service availability and service reliability, and different transaction attributes for services. Meanwhile, due to the non-standard data, the attribute value of the service needs to be normalized and is taken from 0 to 1.
2. User demand i pairShould be an Abstract service process set (Abstract service process set) aspsi. Abstract service flow set aspsiRepresented by quadruplets<Si,Gi,Ii,Pi>In which S isiRepresenting a collection of abstract services, GiIs a set of granularities, IiIs the structure of an abstract service flow, PiRepresenting a possible abstract service flow. Each task of the abstract service flow may be completed by a different candidate service having different attribute parameter values, forming a different service composition scheme.
Each requirement i corresponds to an abstract service flow set aspsi={aspi1,aspi2,…,aspik}. Asps in abstract service flow setiEach item of (2) represents an abstract service flow, and a concrete service is selected for each abstract service to obtain a corresponding concrete service flow set:
cspsi={cspi1,cspi2,...,cspik}
each specific service flow cspipThe attribute parameters of (2) constitute a vector:
{Q1(cspip),Q2(cspip),Q3(cspip),Q4(cspip)}
wherein each dimension attribute of the vector represents a service process cspipResponse time, price, reliability and availability. Calculating each specific service flow cspipThe QoS value of (1), then the specific service flow set cspsiThe maximum value of the QoS values of the specific service flows is taken, and the calculation formula is as follows:
Figure GDA0003009565410000051
wherein wj(wj> 0) represents a user preference,
Figure GDA0003009565410000052
represents the maximum value of all candidate composite service flows of aspsi on the j-dimension attribute,
Figure GDA0003009565410000053
represents the minimum value, Q, of all candidate composite service flows on the j-dimension attributew,jRepresentation cspipThe j-th dimension of the attribute value. The objective function of the service optimization selection facing the concurrency requirement and the multi-granularity is shown as formula (2), wherein m is the quantity of the concurrency requirements.
Figure GDA0003009565410000061
For example, two user needs arrive at the same time, one is medical referral and one is medical follow-up. In the medical referral process, a multi-granularity service (recommendation reservation) can simultaneously realize two functions of hospital recommendation and reservation registration, so that the referral process can be divided into two steps of referral application, hospital recommendation, reservation registration, hospital navigation and disease information reporting and referral application, reservation recommendation, hospital navigation and disease information reporting. In the medical follow-up procedure, there are 4 abstract medical services in the follow-up management-hospital queuing-follow-up questionnaire. And selecting a concrete service for each abstract service, and taking the maximum value of the concrete service flows under the condition of meeting the transaction attribute. And aiming at each demand, after the optimal service is obtained, combining the optimal service, thereby obtaining the maximum value meeting the concurrent demand.
3. The combination scheme with different granularities is obtained by a hierarchical recursive decomposition method.
3.1. Setting an abstract service set, a granularity highest level t and an abstract service flow structure;
3.2. constructing a to-be-decomposed combined service scheme by the abstract service at the highest level t, and adding the scheme into an abstract service flow list ProcessList;
3.3. searching whether decomposable services exist in the combined scheme to be decomposed, if so, executing the step 3.4, otherwise, turning to the step 3.6;
3.4. decomposing the service of the t layer into the t-1 layer to obtain a plurality of combined service schemes, and adding the schemes into the ProcessList; 3.5. forming a combination scheme to be decomposed by the service of the t-1 layer, returning to the step 3.3 when t is t-1;
3.6. and returning a service combination scheme list ProcessList meeting the user requirements.
In medical referrals, for example, there is a multi-granularity service (recommendation reservation service) that can simultaneously implement both functions of recommending a hospital and making a reservation registration. According to the hierarchical recursive decomposition method, two abstract service flows can be obtained, namely referral application-recommendation hospital-appointment registration-hospital navigation-disease information reporting and referral application-recommendation appointment-hospital navigation-disease information reporting.
4. And solving the model by adopting a hybrid optimization algorithm. The algorithm combines particle swarm, cross mutation operators with priorities and a simulated annealing method.
4.1 coding strategy. And coding by adopting an integer array, wherein the length of the array is the number of services contained in the individual, and the value of each element in the array is the identifier of the concrete service instance corresponding to the abstract service binding.
Taking two concurrent requirements of medical referral and medical follow-up as an example, an integer array is constructed, and the length of the array is 10. The first 6 services belong to medical referrals, and the last 4 services belong to medical follow-up visits.
S1 represents a referral application, and the superscript 3 represents that the 3 rd candidate service is selected. S2 represents a recommended hospital, and the superscript 1 represents that the 1 st candidate service is selected. S3 represents a subscription registry, and its superscript 6 represents the selection of the 6 th candidate service. S4 represents hospital navigation, with the 8 th candidate service selected as indicated by the superscript 8. S5 represents the report of disease information, and the superscript 2 represents the selection of the 2 nd candidate service. S2, 3 represents a multi-granularity recommended reservation service which simultaneously satisfies reservation registration and recommended hospital service, and the superscript 2 represents that the 2 nd candidate service is selected. S6 denotes the follow-up management, the superscript 1 denotes that the 1 st candidate service is selected. S7 represents hospital administration, and its superscript 5 represents the selection of the 5 th candidate service. S [8] indicates hospital queuing, and its superscript 7 indicates that the 7 th candidate service was selected. S [9] denotes the follow-up questionnaire, the superscript 4 of which denotes that the 4 th candidate service was selected.
S[1]3 S[2]1 S[3]6 S[4]8 S[5]2 S[2,3]2 S[6]1 S[7]5 S[8]7 S[9]4
4.2 in the iterative process of the algorithm, the particles update the positions thereof according to the individual extreme points and the global extreme points. Individual extreme point ppbestThe position of the optimal solution currently found by the particle is represented; global extreme point pgbestIndicating the location of the best solution currently found by the population.
4.3 introducing a crossover and mutation operator with priority into the algorithm to update the population. Let pj(t) is the jth particle of the tth generation, ppbest(t) is the individual optimal solution for the tth generation corresponding to the jth particle, pgbest(t) is the global optimal solution for the t-th generation.
(1)pj(t) and pgbest(t) performing a prioritized crossover operation to generate new particles t1And t2。pj(t) and ppbest(t) performing a prioritized crossover operation to generate new particles t3And t4. If p ispbest(t) the priority of the selected service transaction is less than or equal to pj(t) transaction priority of the corresponding service, and
Figure GDA0003009565410000071
probability of selecting pjService in (t) to replace ppbestService in (t). Where μ is the initial probability, T is the current iteration number, and T is the maximum iteration number.
E.g. the current particle pjThe transaction attributes of the 10 services of (t) are { p, c, c, cr, p, p, r, cr, r }, respectively. p is a radical ofgbestThe transaction attributes of the 10 services of (t) are { c, c, cr, cr, cr, cr, cr, cr }, respectively. If conventional operation is used, a third service is randomly selected to be crossed, using pj(t) third service replacement pgbest(t) third service, an error occurs. And by adopting a method with priority, the cross operation which does not meet the requirement of transaction attributes can be avoided. At the same time, adopt
Figure GDA0003009565410000072
To select a service. If μ is 0.8, the maximum number of iterations T is 100, and the current number of iterations T is 10, then
Figure GDA0003009565410000073
I.e. with a probability of 0.72 using pj(t) fourth service replacement pgbest(t) a fourth service, which makes the algorithm more random.
(2)pj(t) performing a mutation operation with priority to generate a new particle t5. This operation can avoid premature trapping of particles in the local optimum. Change with priorityIn contrast, if the service transaction attribute to be mutated is p, the mutation is performed according to the probability
Figure GDA0003009565410000081
And randomly selecting candidate services with the attribute greater than or equal to p, thereby ensuring that the mutated combined service meets the transaction attribute.
E.g. pjThe transaction attributes of the 10 services of (t) are { p, c, cr, cr, cr, p, r, cr, r }, respectively. If the traditional mutation operation is adopted, a second service is randomly selected for mutation, and the service with the transaction attribute p is used for replacement, an error occurs. And by adopting a method with priority, mutation operation which does not meet the requirement of transaction attributes can be avoided. By using
Figure GDA0003009565410000082
The mutation can make the algorithm have better randomness.
(3) Calculating a new particle t1、t2、t3、t4、t5The maximum value is selected. Suppose f (t)3) When the value of (c) is maximum, f (t) is set3) Comparing with the global optimum and the individual optimum if f (t)3) Is better than the individual optimum, then ppbest(t+1)=t3(ii) a If f (t)3) Is better than the global optimum, then pgbest(t+1)=t3. Simultaneous particles t3P as the next iterationj(t+1)。
4.4 simulated annealing algorithm is introduced into the algorithm, so as to jump out local optimum. If the fitness value of the new state is worse than the current optimal value, the algorithm receives the degraded solution with a certain probability, so that the simulated annealing algorithm has the capability of jumping out of the local optimal value.
4.5 hybrid optimization Algorithm
The hybrid optimization algorithm combines the capability of quickly searching the optimal solution by the particle swarm and the capability of preventing the particle swarm from falling into the local optimal solution by simulated annealing, and makes up for the shortages to obtain the global optimal solution. Meanwhile, the cross mutation operator with the priority is adopted, so that the problem that the solution does not meet the attribute of the transaction can be avoided.
4.5.1. Initializing to generate particles meeting the constraint, setting the maximum iteration number, and initializing the initial temperature T of the simulated annealing0And a termination temperature TfSetting the attenuation coefficient alpha and the Markov chain length Lk
4.5.2. Calculating the value of fitness function f (x) of each particle, and taking the position corresponding to the fitness value as the individual optimal solution PpbestSelecting a global optimal solution P from all current individual optimal solutionsgbest
4.5.3. Setting search position x of simulated annealing algorithm0Is PgbestPerforming a mutation operation on the solution to generate a new solution and using Metropolis criteria to determine whether to accept the new solution, and repeating the process LkThen obtaining a new solution x', and then executing annealing operation to reduce the temperature Tk+1=α*Tk
4.5.4. Each particle is based on the current individual optimal solution PpbestAnd a global optimal solution PgbestTo update the velocity and position and, based on the newly generated particles, to update the individual optimal solution P for each particlepbestAnd a global optimal solution PgbestThe iteration number is increased by 1;
4.5.5. if the maximum number of iterations is reached, outputting the result, otherwise, continuing to perform step 4.5.6;
4.5.6. mixing x' with PgbestIf f (x') is better than f (P)gbest) If yes, randomly deleting one particle in the particle swarm, adding x' into the particle swarm, and returning to the step 4.5.4; if the value of f (x') is greater than f (P)gbest) If poor, let x0=PgbestAnd returning to the step 4.5.3.

Claims (2)

1. A medical information service selection method based on a hybrid optimization algorithm is characterized by comprising the following steps:
1.1 extracting multidimensional attributes of medical information service, including service quality attributes and transaction attributes; the quality of service (QoS) attribute is composed of some non-functional attributes, including service price, service response time, service availability and service reliability;
the transaction attributes of the service mainly include the following types: a compensable transaction attribute c, a retriable transaction attribute r, a pivot transaction attribute p and a compensable transaction attribute cr;
1.2 user requirement i corresponds to an Abstract service process set (Abstract service process set) aspsiAbstract service flow set aspsiBy quadruplets<Si,Gi,Ii,Pi>Is shown in which S isiRepresenting a collection of abstract services, GiIs a set of granularities, IiIs the structure of an abstract service flow, PiRepresenting a feasible abstract service flow; each task of the abstract service flow is completed by different candidate services, and the candidate services have different attribute parameter values to form different service combination schemes;
each requirement i corresponds to an abstract service flow set aspsi={aspi1,aspi2,…,aspik}; asps in abstract service flow setiEach item of (2) represents an abstract service flow, and a concrete service is selected for each abstract service to obtain a corresponding concrete service flow set:
cspsi={cspi1,cspi2,...,cspik}
each specific service flow cspipThe attribute parameters of (2) constitute a vector:
{Q1(cspip),Q2(cspip),Q3(cspip),Q4(cspip)}
wherein each dimension attribute of the vector represents a service process cspipResponse time, price, reliability and availability of; calculating each specific service flow cspipThe QoS value of (1), then the specific service flow set cspsiThe maximum value of the QoS values of the specific service flows is taken, and the calculation formula is as follows:
Figure FDA0003009565400000011
s.t.T(cspip)∈{p,c,r,cr}
wherein wj(wj> 0) represents a user preference,
Figure FDA0003009565400000012
representation of aspsiThe maximum value of all candidate composite service flows on the j-dimension attribute,
Figure FDA0003009565400000013
represents the minimum value, Q, of all candidate composite service flows on the j-dimension attributeip,jRepresentation cspipThe objective function of concurrent demand and multi-granularity service optimization selection is shown as a formula (2), wherein m is the quantity of concurrent demands;
Figure FDA0003009565400000021
1.3 obtaining different granularities of combination scheme by hierarchical recursive decomposition method
1.3.1. Setting an abstract service set, a granularity highest level t and an abstract service flow structure;
1.3.2. constructing a to-be-decomposed combined service scheme by the abstract service at the highest level t, and adding the scheme into an abstract service flow list ProcessList;
1.3.3. searching whether decomposable services exist in the combined scheme to be decomposed, if so, executing the step 1.3.4, otherwise, turning to the step 1.3.6;
1.3.4. decomposing the service of the t layer into the t-1 layer to obtain a plurality of combined service schemes, and adding the schemes into the ProcessList;
1.3.5. forming a combination scheme to be decomposed by the service of the t-1 layer, returning to the step 1.3.3 when t is t-1;
1.3.6. returning a service combination scheme list ProcessList meeting the user requirements;
1.4 solving the model by adopting a hybrid optimization algorithm
The algorithm combines a particle swarm, a cross mutation operator with priority and a simulated annealing method;
1.4.1 coding strategy
Coding by adopting an integer array, wherein the length of the array is the number of services contained in an individual, and the value of each element in the array is the identifier of a specific service instance corresponding to the abstract service binding;
1.4.2 in the iterative process of the algorithm, updating the position of the particle according to the individual extreme point and the global extreme point; individual extreme point ppbestThe position of the optimal solution currently found by the particle is represented; global extreme point pgbestRepresenting the position of the optimal solution currently found by the population;
1.4.3, introducing a crossover and mutation operator with priority into the algorithm to update the population; let pj(t) is the jth particle of the tth generation, ppbest(t) is the individual optimal solution for the tth generation corresponding to the jth particle, pgbest(t) is the global optimal solution for the t-th generation;
(1) pj (t) and pgbest (t) are subjected to a crossover operation with priority to generate new particles t1 and t 2; pj (t) and ppbest (t) are subjected to a crossover operation with priority to generate new particles t3 and t 4;
if the priority of the service transaction selected in ppbest (t) is less than or equal to pj(t) transaction priority of the corresponding service, and
Figure FDA0003009565400000022
probability of selecting pjService in (t) to replace ppbestThe service in (t); where μ is the initial probability, T is the current iteration number, and T is the maximum iteration number;
(2)pj(t) performing a mutation operation with priority to generate a new particle t5(ii) a This operation can avoid premature trapping of particles in local optima, in the case of prioritized mutation, if the service transaction attribute to be mutated is p, the mutation is performed according to probability
Figure FDA0003009565400000031
Randomly selecting candidate services with the attribute greater than or equal to p to ensure that the mutated combined service meets the transaction attribute;
(3) calculating a new particle t1、t2、t3、t4、t5The maximum value is selected, assuming f (t)3) When the value of (c) is maximum, f (t) is set3) Comparing with the global optimum and the individual optimum if f (t)3) Is better than the individual optimum, then ppbest(t+1)=t3(ii) a If f (t)3) Is better than the global optimum, then pgbest(t+1)=t3While the particle t3P as the next iterationj(t+1);
1.4.4 a simulated annealing algorithm is introduced into the algorithm to jump out of local optima, and if the fitness value of a new state is worse than the current optimum value, the algorithm can accept the degradation solution with a certain probability, so that the simulated annealing algorithm has the capability of jumping out of the local optima;
1.4.5 hybrid optimization Algorithm
The hybrid optimization algorithm combines the capability of quickly searching the optimal solution by the particle swarm and the capability of preventing the particle swarm from falling into the local optimal solution by simulated annealing to obtain the global optimal solution; meanwhile, a cross mutation operator with priority is adopted, so that solutions which do not meet the attributes of the transactions can be avoided;
1.4.5.1. initializing to generate particles meeting the constraint, setting the maximum iteration number, and initializing the initial temperature T of the simulated annealing0And a termination temperature TfSetting the attenuation coefficient alpha and the Markov chain length Lk
1.4.5.2. Calculating the value of fitness function f (x) of each particle, and taking the position corresponding to the fitness value as the individual optimal solution PpbestSelecting a global optimal solution P from all current individual optimal solutionsgbest
1.4.5.3. Setting search position x of simulated annealing algorithm0Is PgbestPerforming a mutation operation on the solution to generate a new solution and using Metropolis criteria to determine whether to accept the new solution, and repeating the process LkAfter the next timeObtaining a new solution x', then performing an annealing operation to reduce the temperature, Tk+1=α*Tk
1.4.5.4. Each particle is based on the current individual optimal solution PpbestAnd a global optimal solution PgbestTo update the velocity and position and, based on the newly generated particles, to update the individual optimal solution P for each particlepbestAnd a global optimal solution PgbestThe iteration number is increased by 1;
1.4.5.5. if the maximum number of iterations is reached, outputting the result, otherwise, continuing to perform step 1.4.5.6;
1.4.5.6. mixing x' with PgbestIf f (x') is better than f (P)gbest) If yes, randomly deleting one particle in the particle swarm, adding x' into the particle swarm, and returning to the step 1.4.5.4; if the value of f (x') is greater than f (P)gbest) If poor, let x0=PgbestReturning to step 1.4.5.3.
2. The hybrid optimization algorithm-based medical information service selection method according to claim 1, wherein: in the step 1.1, probability priority is set for the service transaction attribute; the priority is dynamically changed according to the change condition of the candidate service; (1) if the transaction attribute of the service i is c, the alternative candidate service transaction attribute is c or cr; if the number of the services with the attribute of c is higher than the number of the services with the attribute of cr, selecting the candidate services with the transaction attribute of c to have high probability; (2) if the transaction attribute of the service i is r, the alternative candidate service transaction attribute is r or cr; if the number of the services with the attribute r is higher than the number of the services with the attribute cr, selecting the candidate services with the transaction attribute r with high probability; (3) if the transaction attribute of the service i is cr, the alternative candidate service transaction attribute is cr; (4) if the transaction attribute of the service i is p, the alternative candidate service transaction attribute is p or c or r or cr; determining the priority of the candidate service according to the number of services with different transaction attributes;
due to the non-standard data, the attribute value of the service needs to be normalized and is taken from 0 to 1.
CN201911352264.2A 2019-12-25 2019-12-25 Medical information service selection method based on hybrid optimization algorithm Active CN111144645B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201911352264.2A CN111144645B (en) 2019-12-25 2019-12-25 Medical information service selection method based on hybrid optimization algorithm

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201911352264.2A CN111144645B (en) 2019-12-25 2019-12-25 Medical information service selection method based on hybrid optimization algorithm

Publications (2)

Publication Number Publication Date
CN111144645A CN111144645A (en) 2020-05-12
CN111144645B true CN111144645B (en) 2021-06-01

Family

ID=70519800

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201911352264.2A Active CN111144645B (en) 2019-12-25 2019-12-25 Medical information service selection method based on hybrid optimization algorithm

Country Status (1)

Country Link
CN (1) CN111144645B (en)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112733999B (en) * 2021-01-19 2023-03-21 昆明理工大学 Service mode construction method based on self-error correction mechanism particle swarm optimization algorithm

Family Cites Families (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20010053991A1 (en) * 2000-03-08 2001-12-20 Bonabeau Eric W. Methods and systems for generating business models
CN101739385A (en) * 2009-12-11 2010-06-16 中国科学院软件研究所 Composite service selection method for supporting service quality association
CN102193994B (en) * 2011-04-22 2013-07-24 武汉大学 Method for searching Web services according to non-functional requirements of user

Also Published As

Publication number Publication date
CN111144645A (en) 2020-05-12

Similar Documents

Publication Publication Date Title
Vesselinova et al. Learning combinatorial optimization on graphs: A survey with applications to networking
Li et al. An efficient and reliable approach for quality-of-service-aware service composition
Ng et al. Optimizing construction time and cost using ant colony optimization approach
Balaprakash et al. Estimation-based metaheuristics for the probabilistic traveling salesman problem
Chattopadhyay et al. QoS-aware automatic Web service composition with multiple objectives
CN110033076B (en) Workflow data layout method for cost optimization in mixed cloud environment
CN116501711A (en) Computing power network task scheduling method based on &#39;memory computing separation&#39; architecture
CN114423023B (en) Mobile user-oriented 5G network edge server deployment method
Wang et al. Reputation-enabled federated learning model aggregation in mobile platforms
Gao et al. A particle swarm optimization with lévy flight for service caching and task offloading in edge-cloud computing
CN111047040A (en) Web service combination method based on IFPA algorithm
Wang et al. ISAT: An intelligent Web service selection approach for improving reliability via two-phase decisions
CN116339932A (en) Resource scheduling method, device and server
CN111144645B (en) Medical information service selection method based on hybrid optimization algorithm
Zhang et al. Leveraging QoE heterogenity for large-scale livecaset scheduling
Wang et al. Mgpolicy: Meta graph enhanced off-policy learning for recommendations
Ometov et al. On applicability of imagery-based CNN to computational offloading location selection
Vahedi et al. Heterogeneous task allocation in mobile crowd sensing using a modified approximate policy approach
Liu et al. Towards dynamic reconfiguration of composite services via failure estimation of general and domain quality of services
Mokdad et al. Performance evaluation of a solution for composite service selection problem with multiple consumers
Lozano et al. Network reconstruction from betweenness centrality by artificial bee colony
Shahrokh et al. QoS-based web service composition applying an improved genetic algorithm (IGA) method
Jafarpour et al. QoS-aware Selection of Web Service Compositions using Harmony Search Algorithm.
Xiong et al. Reinforcement learning for finite-horizon restless multi-armed multi-action bandits
Lara-Caballero et al. Multiobjective genetic algorithms for reinforcing equal population in congressional districts

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
CP03 Change of name, title or address

Address after: 12F, Building 1, Aosheng Building, 1166 Xinluo Street, Jinan New District, Shandong 250000

Patentee after: Zhongyang Health Technology Group Co.,Ltd.

Address before: 12 / F, building 1, Aosheng building, 1166 Xinluo street, high tech Zone, Jinan City, Shandong Province

Patentee before: SHANDONG MSUNHEALTH TECHNOLOGY GROUP Co.,Ltd.

CP03 Change of name, title or address