A medical material selection method and system considering individual semantic continuous learning

Abstract

The application discloses a medical material selection method and system considering individual semantic continuous learning, and relates to the technical field of medical material selection.The method comprises the following steps: step one, determining the value of the individualized interval numerical scale at time t; step two, calculating the group consensus level at time t; when the group consensus level is greater than or equal to a predefined consensus threshold, directly entering step four; when the group consensus level is less than the predefined consensus threshold, identifying the decision expert set that needs to implement modification information at time t, modifying the language preference information of each expert in the decision expert set, and entering step three; step three, setting t=t+1, and repeating steps one and two to realize the individualized semantic update at time t and simultaneously calculate the group consensus level; and step four, selecting the final selected brand by using an information aggregation method.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of medical material selection, and particularly relates to a medical material selection method and system considering individual semantic continuous learning. BACKGROUND

[0002] With more and more attention to involving all stakeholders in the medical decision-making process, promoting direct communication, and incorporating their opinions into the production and application of evidence. Although the selection of medical materials focuses on value, this value is multidimensional, and different stakeholders emphasize different aspects, which sometimes leads to conflicts. The consensus reaching process aims to minimize these conflicts, promote coordination between stakeholders, and improve the acceptance of decision experts to the final decision result. Although many studies have explored the consensus model in group decision-making problems based on linguistic information, there are still some deficiencies due to its inherent uncertainty and complexity. These limitations mainly manifest in two aspects: existing studies solve the individual semantic problem between decision experts by designing optimization models based on consistency-driven. However, in the actual decision-making process, people tend to use interval forms to better capture the hesitation characteristics of decision experts. In addition, the cognitive ability of individuals may change over time.

[0003] The Chinese patent CN114819282A is an emergency task planning scheme evaluation method and system based on personalized consensus, which includes obtaining the personal decision preference matrix determined by each evaluation expert after comparing different emergency task planning schemes with each other according to a given number of emergency task planning schemes. The personal decision preference matrix uses a linguistic distribution preference relationship as a decision opinion expression form. By quantifying the language items, not only can the language items in the decision matrix be made operational, but also the personalized individual semantics of the evaluation experts can be obtained. According to the initial group consensus degree, the automatic feedback algorithm is used to adjust the personal preferences of the evaluation experts, and the adjusted interval fuzzy preference relationship of each evaluation expert is obtained. The evaluation information of the planning scheme corresponding to the evaluation expert with low consensus level can also be automatically adjusted.

[0004] However, the patent CN114819282A only considers the static personalized individual semantics of the decision maker, that is, the individual cognition does not change over time, but in the actual decision-making process, the individual views and cognition of the decision maker will continuously learn and dynamically adjust during the interaction with other decision makers and it is impossible to remain static. Therefore, the scheme evaluation based on static personalized individual semantics will have a significant impact on the final evaluation result. SUMMARY

[0005] Based on the technical problems existing in the background art, the present application proposes a medical material selection method and system considering individual semantic continuous learning, which shows greater flexibility and applicability in modeling and extracting group information, and can effectively reduce the influence of abnormal values.

[0006] The medical material selection method considering individual semantic continuous learning provided by the present application comprises:

[0007] Step one, the values of the individual interval numerical scales at the t time are determined by minimizing the consistency index of the interval fuzzy preference relation and the deviation of semantic information between the decision experts. Step two, the group consensus level at the t time is calculated using the weights of each decision expert at the t time and the consensus level, when the group consensus level is greater than or equal to the predefined consensus threshold, directly entering step four, when the group consensus level is less than the predefined consensus threshold, identifying the decision expert set that needs to implement modification information at the t time, and modifying the language preference information of each expert in the decision expert set, and entering step three.

[0008] Step three, another time is set as t+1, and steps one to two are cycled, the individual semantic at the t time is updated based on the updated language preference information, and the group consensus level is calculated until the group consensus level reaches the predefined consensus threshold. Step four, the preference information of each decision expert for each medical material brand is calculated and the preference information of all experts is aggregated, so as to obtain the information granule of each medical material brand, and the medical material brand that makes the middle value of the information granule maximum is selected as the final selected brand.

[0009] Further, before step one, the decision matrix of each decision expert and the individual interval numerical scale of each expert under the language item in the language preference relation are first obtained, and the decision matrix is expressed in the language preference relation. Further, in step one, the values of the individual interval numerical scales at the t time are determined by minimizing the consistency index of the interval fuzzy preference relation and the deviation of semantic information between the decision experts, and the objective function is:

[0010]

[0011]

[0012] wherein, is the total number of decision experts,

[0013] ;

[0014] ​​​​​indexing for decision maker, indexing for decision maker at weighting of time, indexing for decision maker at variable of time, , indexing for decision maker at consistency index of interval fuzzy preference relation at indexing for decision maker at interval fuzzy preference relation of the th medical material brand relative to the th medical material brand, indexing for at group preference degree of the th medical material brand relative to the th medical material brand, denote distance, total number of medical material brands, indexing for medical material brands.

[0015] Further, step one, by minimizing the consistency index of interval fuzzy preference relation and the deviation of semantic information between decision makers, determine the value of personalized interval numerical scale at time, the constraint condition is:

[0016] ;

[0017] wherein, , is left endpoint of interval fuzzy preference relation, is right endpoint of interval fuzzy preference relation, is left endpoint of personalized interval numerical scale of decision maker at time, is right endpoint of personalized interval numerical scale of decision maker at time, is personalized interval numerical scale of decision maker at time for medical material brandCompared to medical supply brands Language preference information, decision experts exist Time Language Item The left and right endpoints of the personalized interval numerical scale, decision experts exist Time Language Item The left and right endpoints of the personalized interval numerical scale, They are the 1st and 2nd respectively. , , and One dividing point, For language item index, This represents the total number of language items, and also the total number of cutoff points. decision experts exist Time Language Item The left and right endpoints of the personalized interval numerical scale, For constraint values, For decision-making experts exist Variables at time, For decision-making experts The variable at time 0.

[0018] Furthermore, in step two, Level of group consensus at any given moment The calculation formula is as follows:

[0019] ;

[0020] in, For decision-making experts exist Weight of time, For decision-making experts exist The level of consensus at any given moment.

[0021] Furthermore, in step two, the language preference information of each expert in the decision expert set is modified, and the corresponding objective function is:

[0022] ;

[0023] in, , For decision-making experts exist Time for the first Medical supply brands the interval fuzzy preference relation of the i-th medical material brand, the left end point of the interval fuzzy preference relation of the i-th medical material brand, the right end point of the interval fuzzy preference relation of the i-th medical material brand, the total number of medical material brands, the index of the i-th medical material brand. Further, in step two, the language preference information of each decision expert in the decision expert set is modified, and the corresponding constraint condition is: ;

[0024] wherein,

[0025] is a predefined consensus threshold, is the temporary consensus level of the decision expert at time t, which is calculated based on the updated language preference information of the decision expert at time t, is the consensus level calculated by the updated language preference information of the decision expert,

[0026] is the language preference information of the decision expert updated at time t for the i-th medical material brand relative to the j-th medical material brand, is the group preference degree of the i-th medical material brand relative to the j-th medical material brand at time t, is a weight parameter for balancing the importance of and denotes a function for converting the interval fuzzy preference information into language information. Further, in step four, the preference information of each decision expert for each medical material brand is calculated, and the preference information of all experts is aggregated to obtain the information granule of each medical material brand, specifically: wherein, , are the preference information of the i-th decision expert for the j-th medical material brand, are the preference information of the i-th decision expert for the j-th medical material brand, are the preference information of the i-th decision expert for the j-th medical material brand, are the preference information of the i-th decision expert for the j-th medical material brand, are the preference information of the i-th decision expert for the j-th medical material brand, are the preference information of the i-th decision expert for the j-th medical material brand, are the preference information of the i-th decision expert for the j-th medical material brand, are the preference information of the i-th decision expert for the j-th medical material brand, are the preference information of the i-th decision expert for the j-th medical material brand, are the preference information of the i-th decision expert for the j-th medical material brand, are the preference information of the i-th decision expert for the j-th medical material brand, are the preference information of the i-th decision expert for the j-th medical material brand, are the preference information of the i-th decision expert for the j-th medical material brand,

[0027] are the preference information of the i-th decision expert for the j-th medical material brand,

[0028] are the preference information of the i-th decision expert for the j-th medical material brand, are the preference information of the i-th decision expert for the j-th medical material brand, are the preference information of the i-th decision expert for the j-th medical material brand, are the preference information of the i-th decision expert for the j-th medical material brand, are the preference information of the i-th decision expert for the j-th medical material brand, ​​a left end point and a right end point of the preference relation of the medical material brand;

[0029] aggregating information of all experts to obtain information granule of each medical material brand , that is,

[0030] ;

[0031] wherein, indicates coverage of the information granule, indicates specificity of the information granule.

[0032] A medical material selection system considering individual semantic continuous learning, comprising a personalized interval numerical scale value determination module, a personalized semantic updating module, a feedback adjustment module and an aggregation selection module;

[0033] The personalized interval numerical scale value determination module is configured to determine values of a personalized interval numerical scale by minimizing a consistency index of an interval fuzzy preference relation and a deviation of semantic information between decision experts, the personalized interval numerical scale being an interval numerical scale of each decision expert under a linguistic term in a linguistic preference relation; at a moment, the personalized interval numerical scale value determination module is configured to determine values of a personalized interval numerical scale at a moment, the personalized interval numerical scale being an interval numerical scale of each decision expert under a linguistic term in a linguistic preference relation;

[0034] The personalized semantic updating module is configured to calculate a group consensus level at a moment by using weights of each decision expert at the moment and a consensus level, and when the group consensus level is greater than or equal to a predefined consensus threshold, directly entering the aggregation selection module, and when the group consensus level is less than the predefined consensus threshold, identifying a set of decision experts that need to implement modification information at the moment, and modifying linguistic preference information of each expert in the set of decision experts to realize that the group consensus level reaches the predefined consensus threshold. The feedback adjustment module is configured to further , and the personalized interval numerical scale value determination module is cycled to the personalized semantic updating module, for updating the personalized semantic corresponding to the decision maker whose modified linguistic preference information until the group consensus is greater than or equal to the predefined consensus threshold.

[0035] The aggregation selection module is configured to calculate preference information of each decision expert on each medical material brand and aggregate preference information of all experts, to obtain an information granule of each medical material brand, and select a medical material brand with a maximum median value of the information granule as a final selected brand.

[0036] The aggregation selection module is configured to calculate preference information of each decision expert on each medical material brand and aggregate preference information of all experts, to obtain an information granule of each medical material brand, and select a medical material brand with a maximum median value of the information granule as a final selected brand.

[0037] Further, in the personalized interval numerical scale value determination module, the objective function is:

[0038] Further, in the personalized interval numerical scale value determination module, the objective function is:

[0038] ;

[0039] in, The total number of decision-making experts, For decision-making experts index, For decision-making experts exist Weight of time, For decision-making experts exist Variables at time, , For decision-making experts exist Interval fuzzy preference relation at time intervals Consistency indicators For decision-making experts exist Time for the first Medical supply brands Compared to the first Medical supply brands Interval fuzzy preference relations, for The first moment Medical supply brands Compared to the first Medical supply brands The degree of group preference Indicates distance, This represents the total number of medical supply brands. These are indexes of medical supply brands;

[0040] In the feedback adjustment module, the language preference information of each expert in the decision expert set is modified, and the corresponding objective function is:

[0041] ;

[0042] in, , for The left endpoint of the corresponding interval fuzzy preference relation, for The right endpoint of the corresponding interval fuzzy preference relation.

[0043] The medical material selection method and system considering individual semantic continuous learning have the advantages that: through collecting historical data generated by multiple rounds of language preference information and consensus reaching processes, individualized expression of decision experts is captured, language information of different decision experts is quantified and updated in real time, and a new method is provided for solving medical material management problems involving language information. In addition, according to the reasonable granularity principle, an information aggregation method is introduced based on information granularity, which intuitively explains the optimal allocation of particles in original digital data, and can effectively summarize and describe the collective opinion at a higher abstraction level. Compared with the previous research, the method shows greater flexibility and applicability in modeling and extracting group information, and can effectively reduce the influence of outliers. BRIEF DESCRIPTION OF DRAWINGS

[0044] Figure 1 It is a structural schematic diagram of the present application. DETAILED DESCRIPTION

[0045] In the following, the technical solutions of the present application will be described in detail through specific embodiments. In the following description, many specific details are set forth in order to provide a thorough understanding of the present application. However, the present application can be implemented in many other ways different from those described herein, and those skilled in the art can make similar improvements without departing from the scope of the present application, so the present application is not limited to the specific implementation disclosed below.

[0046] In the present embodiment, quantifying language items not only makes language items in the decision matrix operable, but also obtains individualized individual semantics of each decision expert. In addition, through individualized semantic update at different time points, combined with the information aggregation method, accurate selection of medical material brands is realized.

[0047] As shown in Figure 1 , the medical material selection method considering individual semantic continuous learning proposed by the present application comprises:

[0048] Step one, determining the numerical value of the individual interval numerical scale corresponding to the language items in the language preference relationship of each decision expert by minimizing the consistency index of the interval fuzzy preference relationship and the deviation of semantic information between decision experts; Step two, calculating the group consensus level at the time point t by using the weight of each decision expert at the time point t and the consensus level, when the group consensus level is greater than or equal to the predefined consensus threshold, directly entering step four, and when the group consensus level is less than the predefined consensus threshold, identifying

[0049] Step three, updating the individualized interval numerical scale of each decision expert at the time point t+1 according to the group consensus level at the time point t, and returning to step one; Step four, calculating the group consensus level at the time point t+1 by using the weight of each decision expert at the time point t+1 and the consensus level, and when the group consensus level is greater than or equal to the predefined consensus threshold, directly entering step six, and when the group consensus level is less than the predefined consensus threshold, identifying ​The decision-making experts set implementing the modification information in real time is modified, and the language preference information of each expert in the decision-making expert set is modified, to realize The personalized semantic update in real time, entering step three;

[0050] Step three, another , the steps one to two are cycled to realize The personalized semantic update in real time, and the group consensus level is calculated simultaneously, until the group consensus level is greater than or equal to the predefined consensus threshold;

[0051] Step four, the preference information of each decision-making expert for each medical material brand is calculated, and the preference information of all experts is aggregated, to obtain the information grain of each medical material brand, and the medical material brand with the maximum median value of the information grain is selected as the final selected brand.

[0052] The embodiment aims to realize the quantification and real-time update of the language information of different decision-making experts by collecting the historical data generated by multiple rounds of language preference information and consensus reaching process, and by capturing the personalized expression of decision-making experts, and provides a new method for solving the medical material management problem involving language information. In addition, the embodiment introduces an information aggregation method based on information granularity according to the reasonable granularity principle, which intuitively explains the optimal allocation of particles in the original digital data, and can effectively summarize and describe the collective opinion at a higher abstraction level. Compared with the previous research, the method shows greater flexibility and applicability in modeling and extracting group information, and can effectively reduce the influence of outliers.

[0053] In one of the embodiments, before step one, the decision matrix of each decision-making expert and the individual interval numerical scale of each expert under the language item in the language preference relationship are first obtained, and the decision matrix is expressed in the language preference relationship, specifically:

[0054] Unlike the evaluation method of conventional brand selection, the selection result of the medical material brand usually has a significant impact on the public in the selection process. Therefore, when making a decision to select a medical material brand, a large number of decision-making experts from different professional backgrounds are usually needed to participate. However, in the traditional selection method, many researchers have studied evaluation models with different numerical preference relationships. It is worth noting that the evaluation and selection of medical material brands have some unique characteristics, and it is difficult to describe the personal opinions of decision-making experts with accurate numerical values.

[0055] Unlike the prior art, the embodiment uses linguistic terms to express the personal preferences of the decision experts during the actual evaluation process. However, the linguistic terms are not operational, and the prior art generally assumes that the decision experts have the same understanding of the linguistic terms used. However, it is generally believed in daily life that the same language means different things to different people. For example, when evaluating the quality of a project, three decision experts all think that the quality of the project is "good". However, if the three experts are asked to adjust the "good" to a numerical value, the first decision expert may give 0.8, the second decision expert may give 0.85, and the third decision expert may give 0.9.

[0056] Therefore, the embodiment is configured to save the above technical defects, and the embodiment specifically comprises the following steps:

[0057] A. Obtain the decision matrix of each decision expert;

[0058] Predefine medical supplies are represented by , is the medical supplies brand, each decision expert provides his / her decision opinion by comparing each medical supplies brand with each other, i.e. the decision matrix, which is expressed by a linguistic preference relation (LPR), is the given LPR of the decision expert , wherein represents the preference degree of the decision expert for the medical supplies brand relative to the medical supplies brand , the pre-defined linguistic term set , the total number of linguistic terms in the linguistic term set is , and , , represents the subscript of the linguistic term of the evaluation information , and represents the subscript of the linguistic term of the evaluation information , the linguistic term is the linguistic term, and the total number of linguistic terms is .

[0059] B. Obtain the individual interval numerical scale of each expert under the linguistic term in the linguistic preference relation;

[0060] by the k-th cutpoint composition the k-th interval , respectively, represent the individual interval numerical scale of the decision expert at the linguistic term , i.e.

[0061] ,

[0062] …

[0063] ;

[0064] wherein is the k-th cutpoint, is the k-th interval, represents the left end point of the individual interval numerical scale of the decision expert at the linguistic term at the time point , represents the right end point of the individual interval numerical scale of the decision expert at the linguistic term at the time point , respectively, represent the left and right end points of the individual interval numerical scale of the decision expert at the linguistic term at the time point , represents the left end point of the individual interval numerical scale of the decision expert at the linguistic term at the time point ,

[0065] represents the right end point of the individual interval numerical scale of the decision expert at the linguistic term at

[0066] the time point ,

[0067] , ,

[0068] ,

[0069] , , represents a constraint value, satisfying , The first , , One cutoff point.

[0070] C. Obtain the interval fuzzy preference relation corresponding to the language preference relation;

[0071] Interval fuzzy preference relations , For decision-making experts exist Time for the first Medical supply brands Compared to the first Medical supply brands Interval fuzzy preference relations, The total number of medical supply brands, of which , , , for The right endpoint of the corresponding interval fuzzy preference relation. For decision-making experts exist Evaluation information represented by language items at all times The left endpoint of the personalized interval numerical scale, For decision-making experts exist Evaluation information represented by language items at all times The right endpoint of the personalized interval numerical scale, For decision-making experts exist Always pay attention to medical supply brands Compared to medical supply brands Language preference information.

[0072] In one embodiment, step one involves determining the consistency index of interval fuzzy preference relations and the deviation of semantic information between decision experts by minimizing the semantic information of these relations. The objective function is defined as follows: (The value is assigned to a personalized interval numerical scale at any given time.)

[0073] ;

[0074] The specific constraints are as follows:

[0075] ;

[0076] in, The total number of decision-making experts, For decision-making experts index, For decision-making experts exist weight of time point, decision maker at time point, , decision maker at time point, interval fuzzy preference relation consistency index, decision maker at time point, interval fuzzy preference relation of the first medical material brand relative to the first medical material brand , is the group preference degree of the first medical material brand relative to the first medical material brand at time point, denotes distance, total number of medical material brands, medical material brand index, decision maker at time point, decision maker at time point 0, decision maker at time point, weight of time point, decision maker at time point, consensus level, medical material brand index.

[0077]

[0078]

[0079] wherein, , , and respectively correspond to the left endpoint of the interval fuzzy preference relation, decision maker at time point, left endpoint of the individual interval numerical scale of the evaluation information represented by the language item , decision maker at time point, left endpoint of the individual interval numerical scale of the evaluation information represented by the language item left endpoint of the personalized interval numerical scale.

[0080] weight , representative decision maker at the similarity of the decision makers at the moment, representative decision maker at the similarity of the decision makers at the moment, wherein,

[0081] , number of representative decision makers. representative decision maker at 0-1 variable of the decision makers at the moment, when representative decision maker at the consensus level of the decision makers at the moment is less than a pre-defined consensus threshold, otherwise . . representing the brand of medical supplies at the moment the degree of group preference relative to , respectively the left and right endpoints of the degree of group preference, .

[0082] representative decision maker at the deviation of the evaluation information of the brand of medical supplies at the moment relative to from the group information;

[0083] ;

[0084] In one embodiment, step two, using the weight of each decision maker at the moment and the consensus level to calculate the group consensus level at the moment , when the group consensus level is greater than or equal to the pre-defined consensus threshold, directly enter step four, when the group consensus level is less than the pre-defined consensus threshold, identify the decision maker set at the moment which needs to implement the modification information, and modify the language preference information of each decision maker in the decision maker set, enter step three, which is specifically:

[0085] the group consensus level at the moment is specifically:

[0086] ​;

[0087] in, Representative decision-making experts exist The level of consensus at any given moment

[0088] .

[0089] Predefined consensus threshold ,if If the representative group reaches a satisfactory level of consensus, proceed to step four; otherwise, identify... The decision-making expert set that needs to modify information at any time, and the language preference information of each expert in the decision-making expert set are modified, proceed to step three.

[0090] Among them, identification A group of decision-making experts who need to modify information at any time Specifically:

[0091] ;

[0092] Then, modify Momentary Decision-Making Experts The language preference information of each decision-making expert is represented as , ; For decision-making experts exist Always pay attention to medical supply brands Compared to medical supply brands Language preference information.

[0093] The following modifications are based on the objective function and constraints. Language preference information of various decision-making experts in China:

[0094] ; ;

[0095] in, This is the consensus threshold. For decision-making experts exist Based on updated language preference information at a time-specific interval numerical scale. The calculated consensus level Representatives of decision-making experts Personalized interval numerical scale The following uses evaluation information represented by linguistic items. The left endpoint of the personalized interval numerical scale, Representatives of decision-making experts Personalized interval numerical scale the evaluation information represented by the linguistic terms the right end point of the personalized interval-valued scale. representative decision experts in the personalized interval-valued scale based on the updated linguistic preference information the corresponding interval-valued fuzzy preference relation the consensus level calculated, a function representing the conversion of the interval-valued fuzzy preference information into linguistic information, the superscript represents the moment, represents the inverse function, is a weight parameter used to balance and importance, let , . Let , return to step one.

[0096] In one of the embodiments, step four, the preference information of each decision expert for each medical material brand is calculated and the preference information of all experts is aggregated to obtain the information granule of each medical material brand. The medical material brand with the maximum median value of the information granule is selected as the final selected brand, specifically:

[0097] In the aggregation process, decision experts with greater weights may manipulate opinions to achieve predetermined goals. Traditional information aggregation methods are usually difficult to effectively summarize and describe collective opinions at a higher level of abstraction, and therefore are not suitable for aggregating decision information to reach consensus among decision experts.

[0098] To solve the above problems, the information aggregation method proposed in the embodiment is specifically:

[0099] (a1) calculating the preference information of each scheme wherein,

[0100] , are respectively the left end point and the right end point of the preference relation of the decision expert for the th medical material brand.

[0101] (a2) aggregating the aggregation information of all experts to obtain the information granule of each medical material brand ， are respectively the left end point and the right end point of the information granule, i.e.

[0102] ;

[0103] wherein, represents the coverage of the information particle.

[0104] ;

[0105] ;

[0106] represents the midpoint of the evaluation information , . represents the specificity of the information particle, wherein represents an exponential function.

[0107] Therefore, the embodiment sets the medical material selection method by considering the individual semantic expression and consensus reaching of all stakeholders in the medical material selection process, aiming to realize the continuous learning of individual semantic information of decision experts, and improve the satisfaction and decision quality of the decision result.

[0108] The above merely describes the preferred embodiments of the present application, but the protection scope of the present application is not limited thereto, and any person skilled in the art can make equivalent replacements or changes to the technical range disclosed in the present application according to the technical solution and the inventive concept of the present application, which should be covered in the protection scope of the present application.

Claims

1. A medical supply selection method considering individual semantic continual learning, characterized in that, Comprise: Step one, determining the interval fuzzy preference relation of each decision expert by minimizing the consistency index of the interval fuzzy preference relation and the deviation of semantic information between decision experts the value of the individualized interval numerical scale at the moment, the individualized interval numerical scale being the interval numerical scale of each decision expert under the linguistic term in the linguistic preference relation; The determination The value of the time individualization interval scale corresponds to the target function: in, The total number of decision-making experts, For decision-making experts index, For decision-making experts exist Weight of time, For decision-making experts exist Variables at time, , For decision-making experts exist Interval fuzzy preference relation at time intervals Consistency indicators For decision-making experts exist Time for the first Medical supply brands Compared to the first Medical supply brands Interval fuzzy preference relations, for The first moment Medical supply brands Compared to the first Medical supply brands The degree of group preference Indicates distance, This represents the total number of medical supply brands. These are indexes of medical supply brands; Step two, using the weight of each decision expert at the moment and the consensus level to calculate the group consensus level at the moment, when the group consensus level is greater than or equal to the predefined consensus threshold, directly entering step four, when the group consensus level is less than the predefined consensus threshold, identifying the set of decision experts that need to implement modification information at the moment, and modifying the language preference information of each expert in the set of decision experts to realize personalized semantic update at the moment, entering step three; Step three, another , cycle steps one to two, update the personalized semantics at the moment based on the updated language preference information, and at the same time calculate the group consensus level until the group consensus level reaches the pre-defined consensus threshold; , cycle steps one to two, update the personalized semantics at the moment based on the updated language preference information, and at the same time calculate the group consensus level until the group consensus level reaches the pre-defined consensus threshold; Step four, calculate each decision expert's preference information for each medical material brand and aggregate all experts' preference information, to obtain each medical material brand's information granule, select the medical material brand that makes the middle value of the information granule maximum as the final selected brand.

2. The medical asset selection method of claim 1, wherein, Before step one, first acquire each decision expert's decision matrix and each expert's individual interval number scale under the language item in the language preference relation, the decision matrix is expressed in the language preference relation.

3. The medical asset selection method of claim 1, wherein, Step one, determine the interval-valued fuzzy preference relation by minimizing the consistency index of interval-valued fuzzy preference relation and the deviation of semantic information between decision experts the value of the interval-valued fuzzy preference relation at time t, subject to the constraint in, , for The left endpoint of the corresponding interval fuzzy preference relation, for The right endpoint of the corresponding interval fuzzy preference relation. For decision-making experts exist time The left endpoint of the personalized interval numerical scale, For decision-making experts exist time The right endpoint of the personalized interval numerical scale, For decision-making experts exist Always pay attention to medical supply brands Compared to medical supply brands Language preference information, decision experts exist Time Language Item The left and right endpoints of the personalized interval numerical scale, decision experts exist Time Language Item The left and right endpoints of the personalized interval numerical scale, They are the 1st and 2nd respectively. , , and One dividing point, For language item index, This represents the total number of language items, and also the total number of cutoff points. decision experts exist Time Language Item The left and right endpoints of the personalized interval numerical scale, For constraint values, For decision-making experts exist Variables at time, For decision-making experts The variable at time 0.

4. The medical supply selection method of claim 1, wherein, In step two, the group consensus level at the moment The calculation formula is as follows: wherein, is a decision specialist at a weight of the moment, is a decision specialist at a consensus level of the moment.

5. The medical asset selection method of claim 1, wherein, In step two, modify each expert's language preference information in the decision expert set, the corresponding objective function is: wherein, , is a decision expert at a time instant for a medical material brand with respect to a medical material brand , and is an interval fuzzy preference relation, is a left end point of the interval fuzzy preference relation, is a right end point of the interval fuzzy preference relation, is a total number of medical material brands, are medical material brand indexes, respectively.

6. The medical asset selection method of claim 5, wherein, In step two, modify each expert's language preference information in the decision expert set, the corresponding constraint condition is: wherein, is a consensus threshold, is a decision expert At the moment, the interval numerical scale is personalized, and the consensus level calculated based on the updated language preference information is calculated, is a decision expert At the moment, the language preference information of the medical material brand is updated relative to the medical material brand , is The group preference degree of the th medical material brand relative to the th medical material brand , is a weight parameter, represents a function of converting interval fuzzy preference information into language information.

7. The medical asset selection method of claim 1, wherein, In step four, calculate each decision expert's preference information for each medical material brand and aggregate all experts' preference information, to obtain each medical material brand's information granule, specifically: computing preference information of each decision expert for each medical material brand wherein, , are respectively left end point and right end point of the preference relation of the decision expert for the first medical material brand aggregate information of all experts, obtain information particles of each medical material brand That is: wherein, represents the coverage of the information particle, represents the specificity of the information particle.

8. A medical supply selection system that considers individual semantic persistent learning, characterized by, Comprise individual interval number scale value determination module, individual semantic update module, feedback adjustment module and aggregation selection module; The personalized interval numerical scale value determination module is configured to determine the value of the personalized interval numerical scale for each decision expert under the linguistic term in the linguistic preference relation by minimizing a consistency index of the interval fuzzy preference relation and deviation of semantic information between decision experts the value of the personalized interval numerical scale at the moment, the personalized interval numerical scale being an interval numerical scale of each decision expert under a linguistic term in a linguistic preference relation; The personalized semantic update module is used to leverage the expertise of various decision-making experts. Time-based weighting and consensus level calculation The system determines the group consensus level at any given time. If the group consensus level is greater than or equal to a predefined consensus threshold, it directly proceeds to the aggregation selection module. If the group consensus level is less than the predefined consensus threshold, it identifies... The decision-making expert set needs to be modified at all times, and the language preference information of each expert in the decision-making expert set needs to be modified in order to achieve... Real-time personalized semantic updates, enter the feedback and adjustment module; The feedback adjustment module is configured to adjust the feedback of the decision maker according to the feedback adjustment rule. The loop personalized interval value scale value determination module and the personalized semantic updating module are configured to update the personalized semantics corresponding to the decision maker with the modified language preference information until the group consensus reaches a predefined consensus threshold. The aggregation selection module is used for calculating each decision expert's preference information for each medical material brand and aggregating all experts' preference information, to obtain each medical material brand's information granule, selecting the medical material brand that makes the middle value of the information granule maximum as the final selected brand; In the personalized interval numerical scale value determination module, the determination The value of the personalized interval numerical scale at the moment, and the corresponding objective function is: wherein, is the total number of decision experts, is the index of decision expert, is the decision expert at the weight of time, is the decision expert at the variable of time, , is the decision expert at the consistency index of interval fuzzy preference relation of decision expert at time, is the interval fuzzy preference relation of decision expert at time for the th medical material brand relative to the th medical material brand, is the group preference degree of the th medical material brand relative to the th medical material brand at time, denotes the distance, is the total number of medical material brands, are the indexes of medical material brands, respectively.​ 9. The medical asset selection system of claim 8, wherein, In the individual semantic update module, modify each expert's language preference information in the decision expert set, the corresponding objective function is: wherein, , is the left end point of the corresponding interval fuzzy preference relation, is the right end point of the corresponding interval fuzzy preference relation.

Images