Proxy negotiation method, device and equipment for doctor-patient shared decision making and storage medium

Abstract

Embodiments of the present application provide a proxy negotiation method, device and equipment for doctor-patient shared decision-making and a storage medium, relating to the technical field of doctor-patient shared decision-making assistance. The proxy negotiation method comprises steps S1 to S4. In S1, preference data and membership functions of two participants in doctor-patient shared decision-making for i negotiation topics are obtained respectively. In S2, based on the preference data and the membership functions, behavior models of the two participants are constructed respectively based on a fuzzy constraint network. In S3, the two behavior models are respectively used to generate and send offers to the other party according to their own preference data. In S4, the two behavior models are respectively used to repeatedly update the offers and send them to the other party until one of the behavior models judges that the satisfaction degree of the offer of the other party or the satisfaction degree of the own offer is greater than a satisfaction threshold. The proxy negotiation method of the present application can simulate the process of doctor-patient shared decision-making, effectively alleviate the negotiation pressure brought by the increase of negotiation topics, and enable humans to be relieved from tedious negotiation affairs.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of medical shared decision-making assistance, in particular to a medical shared decision-making proxy negotiation method and device, equipment and storage medium. BACKGROUND

[0002] With the development of the biological-psychological-social medical model, the continuous improvement of the public legal consciousness and medical participation consciousness, patients pay more attention to the medical process and expect to participate in medical decision-making. At the same time, a number of studies have shown that shared decision-making (SDM) helps to improve patient satisfaction and compliance, and improve the doctor-patient relationship.

[0003] Shared decision-making (SDM) is a medical service model in which medical providers invite patients or their caregivers to participate in decisions about patient care. Although the implementation of SDM has broad prospects, it will increase the number of communications between medical providers and patients and reduce the efficiency of medical care.

[0004] Therefore, the applicant proposes the present application after studying the existing technology. SUMMARY

[0005] The present application provides a medical shared decision-making proxy negotiation method, device, equipment and storage medium to improve the above technical problems.

[0006] In a first aspect,

[0007] The present application provides a medical shared decision-making proxy negotiation method, device, equipment and storage medium to improve the above technical problems.

[0008] S1, acquiring the preference data and membership function of two participants of medical shared decision-making on i negotiation topics, respectively.

[0009] S2, according to the preference data and membership function, respectively constructing the behavior model of the two participants based on the fuzzy constraint network.

[0010] S3, the two behavior models are respectively used to generate and send offers according to their own preference data.

[0011] S4, the two behavior models are respectively used to repeat the following steps until one of the behavior models judges that the other party's offer satisfaction or the own offer satisfaction is greater than the satisfaction threshold.

[0012] S41, according to the offer sent by the other party, calculating the other party's offer satisfaction and updating the other party's preference model based on Bayesian learning.

[0013] S42, update the satisfaction threshold according to the opposite preference model, and determine whether the opposite offer satisfaction is greater than the updated satisfaction threshold. When it is determined that the opposite offer satisfaction is greater than the updated satisfaction threshold, output the opposite offer. Otherwise, continue to execute the subsequent steps.

[0014] S43, obtain the feasible solution set and the expected solution set according to the updated satisfaction threshold.

[0015] S44, generate a new offer according to the feasible solution set and the expected solution set, and calculate the self offer satisfaction of the new offer, and then determine whether the self offer satisfaction is greater than the updated satisfaction threshold. When it is determined that the self offer satisfaction is greater than the updated satisfaction threshold, output the new offer. Otherwise, send the new offer to the opposite.

[0016] The second aspect,

[0017] The embodiment of the application provides a proxy negotiation device for doctor-patient shared decision making, which comprises:

[0018] An initial information acquisition module 1 is configured to acquire preference data and membership functions of two participants in doctor-patient shared decision making for i negotiation topics, respectively.

[0019] A behavior model construction module is configured to construct behavior models of the two participants based on a fuzzy constraint network according to the preference data and the membership functions, respectively.

[0020] An initial offer module is configured to generate offers according to the preference data of the two behavior models and send the offers to the opposite.

[0021] A circulating offer module comprises the following units, which are configured to repeat the following units until one of the behavior models determines that the opposite offer satisfaction or the self offer satisfaction is greater than the satisfaction threshold.

[0022] A first updating unit is configured to calculate the opposite offer satisfaction according to the offer sent by the opposite and update the opposite preference model based on Bayesian learning.

[0023] A second updating unit is configured to update the satisfaction threshold according to the opposite preference model, and determine whether the opposite offer satisfaction is greater than the updated satisfaction threshold. When it is determined that the opposite offer satisfaction is greater than the updated satisfaction threshold, output the opposite offer. Otherwise, continue to execute the subsequent steps.

[0024] A third updating unit is configured to obtain the feasible solution set and the expected solution set according to the updated satisfaction threshold.

[0025] The fourth updating unit is configured to generate a new offer according to the feasible solution set and the expected solution set, calculate a self-offering satisfaction degree of the new offer, and determine whether the self-offering satisfaction degree is greater than the updated satisfaction threshold. When it is determined that the self-offering satisfaction degree is greater than the updated satisfaction threshold, the new offer is output. Otherwise, the new offer is sent to the other party.

[0026] The third aspect,

[0027] The embodiment of the present application provides a proxy negotiation device for doctor-patient shared decision making, which comprises a processor, a memory and a computer program stored in the memory. The computer program can be executed by the processor to implement the proxy negotiation method as described in any one of the first aspect.

[0028] The fourth aspect,

[0029] The embodiment of the present application provides a computer readable storage medium comprising a stored computer program, wherein the computer readable storage medium controls the device where the computer readable storage medium is located to execute the proxy negotiation method as described in any one of the first aspect when the computer program runs.

[0030] By adopting the above technical solution, the present application can achieve the following technical effects:

[0031] Through the proxy negotiation method of the embodiment of the present application, the complete process of doctor-patient shared decision making can be simulated and implemented, the negotiation pressure caused by the increase of negotiation topics can be effectively alleviated, and the method has certain learning ability, so that the human being can be relieved from tedious negotiation affairs.

[0032] In order to make the above objectives, characteristics and advantages of the present application more apparent, the following preferred embodiments are described in detail below, and the accompanying drawings are described as follows. BRIEF DESCRIPTION OF DRAWINGS

[0033] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following will briefly introduce the drawings needed to be used in the embodiments. It should be understood that the following drawings only show some embodiments of the present application, and therefore should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can also be obtained without creative labor on the basis of these drawings.

[0034] Figure 1 is a flowchart of the proxy negotiation method provided by the first embodiment of the present application.

[0035] Figure 2 is a schematic diagram of the negotiation framework of the doctor-patient shared decision making model.

[0036] Figure 3 is a sequence diagram of the negotiation process.

[0037] Figure 4 is a schematic diagram of converting the problem of shared decision making between doctors and patients.

[0038] Figure 5 is a schematic diagram of the relationship of SDM agent negotiation.

[0039] Figure 6 is a schematic diagram of the hypothesis space of possible fuzzy membership functions.

[0040] Figure 7 is a schematic diagram of two fuzzy membership functions approximating a real fuzzy membership function not in the hypothesis space.

[0041] Figure 8 is a fuzzy membership function of two participants in shared decision making between doctors and patients.

[0042] Figure 9 is a schematic diagram of the first round of offers of Case 1.

[0043] Figure 10 is a schematic diagram of the second round of offers of Case 1.

[0044] Figure 11 is the concession of each round of offers under different weights.

[0045] Figure 12 is a schematic diagram of the structure of the agent negotiation device provided by the second embodiment of the present application. DETAILED DESCRIPTION

[0046] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, but not all the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work fall within the scope of protection of the present application.

[0047] In order to better understand the technical solutions of the present application, the embodiments of the present application will be described in detail below with reference to the drawings.

[0048] The terms used in the embodiments of the present application are merely for the purpose of describing the specific embodiments, and are not intended to limit the present application. The singular forms "a", "an" and "the" used in the embodiments of the present application and the appended claims are also intended to include the plural forms, unless the context clearly indicates otherwise.

[0049] It should be understood that the term "and / or" as used herein merely describes an association relationship of associated objects, and can represent three relationships, for example, A and / or B can represent three cases of A existing alone, A and B existing together, and B existing alone. In addition, the character " / " herein generally represents an "or" relationship between the front and rear associated objects.

[0050] Depending on the context, the word "if" as used herein can be interpreted as meaning "when" or "while" or "in response to determining" or "in response to detecting". Similarly, depending on the context, the phrase "if it is determined" or "if (a stated condition or event) is detected" can be interpreted as meaning "when it is determined" or "in response to determining" or "when (a stated condition or event) is detected" or "in response to detecting (a stated condition or event)".

[0051] The "first / second" mentioned in the embodiments is only to distinguish similar objects, and does not represent a specific order of the objects. It can be understood that the "first / second" can be interchanged in a specific order or sequence as appropriate. It should be understood that the objects distinguished by "first / second" can be interchanged as appropriate, so that the embodiments described herein can be implemented in an order other than those illustrated or described herein.

[0052] The application will be described in further detail below with reference to the drawings and specific embodiments:

[0053] Embodiment one:

[0054] Please refer to Figures 1 to 11 The first embodiment of the present application provides a proxy negotiation method for doctor-patient shared decision making, which can be executed by a proxy negotiation device for doctor-patient shared decision making. In particular, it is executed by one or more processors in the proxy negotiation device to implement steps S1 to S4. Step S4 includes steps S41 to S44.

[0055] S1, respectively acquiring preference data and membership functions of two participants in doctor-patient shared decision making on i negotiation topics.

[0056] Specifically, the negotiation topics include cost, effectiveness, side effects, risk, and convenience. The preference data includes a preference value range, a preferred range, a minimum preference value, a maximum preference value, and a weight preference value. For specific examples, please refer to the description in the "actual example" below. It can be understood that in other embodiments, any other problem that can occur in doctor-patient shared decision making can also be included, and the present application does not make specific limitations on this.

[0057] It should be noted that the proxy negotiation device can be a portable notebook computer, a desktop computer, a server, a smart phone, or a tablet computer, or any other electronic device with computing performance.

[0058] S2, constructing behavior models of two participants respectively based on fuzzy constraint networks according to preference data and membership functions. Specifically, the fuzzy constraint networks are N l = (U l , X l , C l ), l e {DA, PA}, wherein, N l denotes the lth fuzzy constraint network, U l denotes the domain of the fuzzy constraint network N l , X l denotes the tuple of all non-repeated objects of the fuzzy constraint network N l , C l denotes the set of fuzzy constraints of the fuzzy constraint network N l , and DA denotes the medical provider and PA denotes the patient.

[0059] In the embodiment, first, the SDMP is modeled into a distributed fuzzy constraint satisfaction problem (DFCSP) by means of the constraints and connections between the doctor and the patient and between the issues.

[0060] Then, based on the characteristics of the SDM distributed fuzzy constraint satisfaction problem, the BDI (Belief-Desire-Intention) model and the Agent technology are used to establish the intelligent agents of the doctor and the patient, DA and PA, and their behavior models; the Bayes learning technology is introduced to construct the opponent model to learn the preference of the other party; and the negotiation process of DA and PA and the negotiation protocol to be followed are determined. Specifically, the fuzzy constraint Agent technology is used to enable DA and PA to represent the doctor and the patient and to automatically negotiate the issues involved in the treatment plan under the conditions of inaccurate preference and behavior.

[0061] Finally, the negotiation content is set according to the actual medical scene, and DA and PA can fully negotiate the negotiation content to reach an agreement and obtain the final negotiation result.

[0062] It can be understood that the modeling of the shared decision making problem (SDMP) between doctors and patients is a distributed fuzzy constraint satisfaction problem (DFCSP). We focus on the bilateral shared decision making (SDM) problem, and the SDM scenario exists when the doctor and the patient meet and choose a special treatment after the SDM. In this case, there are two independent participants, namely the doctor and the patient. Both the doctor and the patient have their own beliefs, desires and intentions, and can communicate with others and self-manage, but are affected by other aspects and restricted by society. Both the doctor and the patient have their own preferences in terms of cost, effect, risk and other issues according to their own beliefs and intentions. Both the doctor and the patient reach an agreement on multiple issues through appropriate negotiation process and SDM process.

[0063] Therefore, as Figure 4 indicated, we can regard the direct participants of the SDM as independent, interrelated and environment-retrained agents. The communication between the doctor and the patient is regarded as the negotiation between the subjects. The overall satisfaction of the doctor / patient with the treatment plan is regarded as the personal utility of the agent. In this way, the DA and the PA can support the medical decision of the doctor and the patient.

[0064] In the actual clinical environment, due to the limitations of time, medical literacy and beliefs, the doctor and the patient cannot share more information. Therefore, we assume that the DA and the PA are in an incomplete information environment, that is, both of them do not have complete information about the opponent, nor do they have inaccurate knowledge about themselves. Therefore, there is uncertainty and inaccurate information. As Figure 5 indicated, we can use trapezoidal fuzzy numbers to represent such inaccurate information and constraints.

[0065] The negotiation model of the SDM can be abstracted as a three-tuple (D, P, Q), where D is the subject representing the doctor (DA), P is the subject representing the patient (PA), and Q is the constraint between the two subjects. In addition, due to the autonomy of the DA and the PA, they generate a relatively satisfactory solution through the negotiation between the DA and the PA. Therefore, the SDM problem is further represented as a distributed fuzzy constraint satisfaction problem (DFCSP) of each intelligent agent.

[0066] In the DFCSP of the SDM, the constraint relationship between intelligent agents determines whether there is a solution that satisfies all the constraints of the DFCSP. Therefore, the goal of each intelligent agent is to find a behavior that satisfies its fuzzy constraints. In addition, the DFCSP can be solved through the negotiation between the DA and the PA, and represented by a distributed fuzzy constraint network (DFCN). The DFCN can be satisfied by assigning fuzzy relationships between agents. The DFCN can be defined as follows:

[0067] Definition 1: A Distributed Fuzzy Constraint Network (DFCN) (U, X, C) of Shared Decision Making (SDM) (D, P, Q) can be defined as a fuzzy constraint network N l = (U l , X l , C l ), agent l e {DA, PA}, i.e., N l e {N DA , N PA}.

[0068] where: U l is the universe of discourse of FCNN l (FCN stands for fuzzy constraint network), X l is the tuple of all non-repeated objects of agent l, C l is the set of fuzzy constraints of agent l, which includes the internal constraints among the objects in X and the external constraints with other agents, N l is connected to other DFCNs through a set of external fuzzy constraints, U is the universe of discourse, is the tuple of all non-repeated objects of agents in DFCN, is the set of all fuzzy constraints in DFCN.

[0069] According to Definition 1, the set of non-repeated objects X l of the lth agent represents his / her beliefs, knowledge of the environment (e.g., deadlines and medical resources), and other attributes (e.g., the agent's desires, intentions, and concerns about the reactions of the opponents). The set of fuzzy constraints C l includes all the constraints of agent l, such as priority constraints (e.g., the priorities of the problems), goal constraints (e.g., the expectations on the cost, effectiveness, and other goals), and constraints with other agents.

[0070] In Definition 1, the solution of the FCN N l proposed by agent l can be viewed as the intention Π l of FCN N l , FCN N l represents the fuzzy set of non-repeated objects X l that satisfy all the fuzzy constraints C l . Given a set of problems Q = Q1, Q2, …, Q i , …, Q n and the feasible solutions S e Π l , the aggregated satisfaction value Ψ l (S) of the solution S of agent l is defined as follows:

[0071]

[0072] where Mi (S) is the i-th membership of S. It can be directly obtained from a group of doctors and patients, and can flexibly and effectively represent their preferences for a certain issue. n is the number of issues to be negotiated by DA and PA, i is the weight factor corresponding to the i-th issue.

[0073] As Figure 3 shown, the negotiation protocol handles the interaction between agents in the DFCSP. It is the rule that all agents must follow. It defines all interactions between agents and determines the order and structure of messages. During the negotiation process, DA and PA can negotiate by sending and receiving various types of messages. Table 1 lists and describes different types of messages and their negotiation results.

[0074] Table 1 Negotiation protocol

[0075]

[0076]

[0077] S3, two behavior models are used to generate offers according to their own preference data and send them to each other.

[0078] Preferably, step S3 is specifically: two behavior models generate offers according to the upper limit and / or lower limit of their own preference data and send them to each other. For specific examples, please refer to the description in the "actual example" below. In other embodiments, the first offer can also be generated with other rules, and the present invention does not make specific limitations on this.

[0079] S4, two behavior models are used to repeat the following steps until one of the behavior models judges that the other party's offer satisfaction or its own offer satisfaction is greater than the satisfaction threshold. Specifically, after obtaining the other party's offer, steps S41 to S44 are executed to constitute the behavior model of doctor-patient shared decision-making.

[0080] S41, according to the offer sent by the other party, calculate the other party's offer satisfaction and update the other party's preference model based on Bayesian learning. Specifically, the other party's preference model is the other party's satisfaction estimation model. Among them, the other party's satisfaction estimation model is:

[0081]

[0082]

[0083]

[0084] In the formula, represents the estimated value of the other party's satisfaction, n represents the number of negotiation issues, represents the estimated value of the other party's weight, m ​​represents the number of assumptions, The probability of the hypothesis expressing the other party's weight preference, Assumptions expressing the other party's weight preferences, represents the estimated value of the other party's fuzzy membership, The probability of the hypothesis about the shape of the other party's membership function, Expresses an assumption about the shape of the other party's membership function.

[0085] Based on the above embodiment, in an optional embodiment of the present invention, step S41 includes steps S411 to S413.

[0086] S411. Calculate the other party's quotation satisfaction Ψ based on the quotation sent by the other party. l (S). Among them, the satisfaction of the other party’s quotation Where M i (S) is the membership degree of the i-th negotiation topic in the quotation S, w i is the weight factor corresponding to the i-th negotiation topic, and n is the number of negotiation topics.

[0087] S412: Based on the offer sent by the other party, the probability of the other party's membership function shape hypothesis in the other party's preference model is updated based on the Bayesian learning rule. The updated model of the probability of the other party's membership function shape hypothesis is: Where, Expressing assumptions Bid b t The probability of represents the probability of the hypothesis of the other party's membership function shape, P represents the probability, Indicates bid b t Part of the expected utility, The assumptions about the shape of the other party's membership function, represents the estimated value of the other party's weight, and m represents the number of hypotheses.

[0088] S413. Based on the quotation sent by the other party, the probability of the other party's weight preference hypothesis in the other party's preference model is updated based on the Bayesian learning rule. The updated model of the probability of the other party's weight preference hypothesis is: Where, Expressing assumptions Bid b t The probability of The probability of the other party's weight preference hypothesis, P represents the probability, Indicates a bid t Part of the expected utility, Assumptions expressing the other party's weight preferences, represents the estimated value of the other party's fuzzy membership, and m represents the number of hypotheses.

[0089] In this embodiment, in order to improve the quality and efficiency of healthcare decision-making and make full use of limited information (such as the opponent's counter-offer), it is necessary to add the opponent's preference model to the behavioral model, and, in each round of bidding, use Bayesian learning technology to update the agent's knowledge about the opponent's preferences (i.e., the opponent's preference model).

[0090] Specifically, this section defines weight assumptions and fuzzy membership function assumptions appropriate for each of the opponent's questions. We then use Bayes' rule to update the probabilities of these assumptions during the negotiation process to learn about the opponent's preferences.

[0091] In this model, we use the overall satisfaction Ψ(b t ) to measure the subject's satisfaction with the negotiation content. Therefore, we use a comprehensive satisfaction function to measure the opponent's satisfaction with the offer. It consists of a set of weights wi for each of the n questions and the corresponding fuzzy membership function f i (x i )definition.

[0092]

[0093] Among them, x i is to bid b at negotiation time t t The value of i in the problem.

[0094] To ensure the aggregate satisfaction function Ψ(b t ) is in the range of [0, 1]. We will fuzzy membership function f i Let be in the range of [0, 1] and normalize the weights w so that their sum equals 1.

[0095] Regarding the construction of the other party preference model:

[0096] In order to learn the opponent's comprehensive satisfaction function Ψ(b t ), we need to learn the opponent’s question weight w i and fuzzy membership function f i (x i ). Therefore, we will assume that w in formula (1) i and f i (x i ).

[0097] First, the issue weight w i Assumptions, where we can first set all possible weight matrices H w The set of , which is then combined with the weight hypothesis using the linear function shown below associated.

[0098]

[0099] where, is the hypothesis h j is the ranking of the weights w i , n is the number of negotiation issues.

[0100] Second, we make assumptions about the opponent's fuzzy membership function. Thus, we can assign a membership to each hypothesis in the hypothesis space and model the fuzzy membership function as a probability distribution. As shown in Figure 6 and Figure 7 we can approximate the shape of the true fuzzy membership function for negotiation issue i by associating various fuzzy membership function assumptions with their corresponding probability

[0101] We assume that the agents use a concession-based time-dependent strategy (TDT). According to this strategy, agents start with the offer that has the highest overall satisfaction and gradually approach their bottom line as the negotiation deadline approaches. We estimate the satisfaction value of the counteroffer Ψ'(b t ) = 1 - 0.05 · t with a monotonically decreasing linear function, and thus compute the conditional probability P(h j | b t ) as formula (3).

[0102]

[0103] Ψ'(b t ) = Ψ'(b t-1 ) - c(t) (4)

[0104] where Ψ(b t | h j ) is the satisfaction value of the counteroffer b j by the negotiation opponent given the hypothesis h t , and Ψ'(b t ) is the opponent's estimated satisfaction value of the next offer. The function c(t) is the negotiation concession strategy assumed by the opponent.

[0105] The expected value of the shape of the fuzzy membership function computed by formula (5).

[0106]

[0107] where, is issue i, and hypothesis j is the satisfaction value of the bid generated by the fuzzy membership function.

[0108] We use ​to denote the hypothesis about the weight of issue i according to hypothesis j, and the value of the related weight, i.e. The expected value of the weight is defined as formula (6).

[0109]

[0110] Finally, the bid b t is calculated based on the comprehensive satisfaction, and the opponent's expected value is calculated as shown in formula (7):

[0111]

[0112] For each issue, the weight on the evaluation function and the probability distribution need to be normalized:

[0113]

[0114] The update about the opponent's preference model:

[0115] We update the hypothesis of issue k using the expected value of the weight of the remaining issues defined by the opponent model and the expected value of the shape of the fuzzy membership function. We need to introduce the partial expected utility of the bid b t , i.e. which is defined as formula (9):

[0116]

[0117] We update the probability of the hypothesis on the shape of the membership function according to the Bayes rule as formula (10).

[0118]

[0119] where is the expected value of the weight of issue k.

[0120] The probability of the hypothesis related to the weight of issue k can be updated using a method similar to formula (11).

[0121]

[0122] About the behavior model:

[0123] As shown in Figure 2 and Figure 3 , the negotiation process of the medical provider agent DA and the patient agent PA is defined in the behavior model, which is the process of DA and PA exchanging offers and counteroffers until both parties agree on all issues, or until someone exits the negotiation due to the limitation of the satisfaction threshold or some external factors.

[0124] The first step of the behavior model is to calculate the satisfaction of the counter-offer and update the counter-party preference model according to the counter-offer. In the embodiment, the agent uses the overall satisfaction (ASV) to evaluate the satisfaction of the counter-offer B to determine whether to reach an agreement or make a concession. The ASV of the agent is calculated according to formula (12).

[0125]

[0126] S42, update the satisfaction threshold according to the counter-party preference model, and determine whether the counter-offer satisfaction is greater than the updated satisfaction threshold. When it is determined that the counter-offer satisfaction is greater than the updated satisfaction threshold, output the counter-offer. Otherwise, continue to execute the subsequent steps.

[0127] On the basis of the above embodiment, in an optional embodiment of the present application, step S42 comprises steps S421 to S427.

[0128] S421, calculate the difference degree σ according to the own offer and the counter-offer. Wherein, the difference degree is σ = 1-(G(A0, B0)-G(A, B)) / G(A0, B0), Wherein, G(A0, B0) represents the distance measure between A0 and B0, A0 represents the first offer of oneself, B0 represents the first offer of the counter-party, G(A, B) represents the distance measure between A and B, A represents the latest offer of oneself, B represents the latest offer of the counter-party, N i represents the number of negotiation topics, L(A i ,B i ) represents the distance between A i and B i , A i represents the possibility distribution of A on the negotiation topics, B i represents the possibility distribution of B on the negotiation topics.

[0129] S422, calculate the counter-party concession value γ according to the counter-party preference model. Wherein, the concession value is Wherein, represents the counter-offer satisfaction of the latest offer of the counter-party, represents the counter-offer satisfaction of the first offer of the counter-party.

[0130] Obtain the expected solution set S * of oneself, and calculate the satisfaction level ρ of the expected solution set. Wherein, ρ = Ψ(S * ), wherein, Ψ(S * ) is the satisfaction of the expected solution set.

[0131] S423、According to the satisfaction level, the compactness δ is calculated. The compactness calculation model is δ = 1 - (p - e), wherein p is the satisfaction level of the expected solution set, and e is the satisfaction threshold of the latest bid.

[0132] S424、The time constraint t is calculated. The time constraint calculation model is t = a + (1 - a) (tnow / tmax) β, wherein tnow represents the current negotiation time, tmax represents the negotiation deadline, a and β are constants, β > 1, and 0 ≤ a ≤ 1.

[0133] S425、According to the difference degree σ, the opponent concession value γ, the satisfaction level p, the compactness δ, and the time constraint, the self-concession value Δε is calculated. The self-concession value is Δε = (μ σ (σ)Λμ γ (γ)Λμ ρ (ρ)Λμ δ (δ)Λμ t (t) ω , wherein μ σ (σ) represents the concession expectation corresponding to the difference degree σ, μ γ (γ) represents the concession expectation corresponding to the opponent concession value γ, μ ρ (ρ) represents the concession expectation corresponding to the satisfaction level p, μ δ (δ) represents the concession expectation corresponding to the compactness δ, and μ t () represents the concession expectation corresponding to the time constraint t.

[0134] S426、According to the self-concession value Δε, the satisfaction threshold is updated. The satisfaction threshold of the first bid is 1. The update model of the satisfaction threshold is e * = e - Δε, wherein e * represents the updated satisfaction threshold, e represents the satisfaction threshold of the last bid, and Δε represents the self-concession value.

[0135] S427、It is judged whether the opponent bid satisfaction is greater than the updated satisfaction threshold. When it is judged that the opponent bid satisfaction is greater than the updated satisfaction threshold, the bid of the opponent is output. Otherwise, the subsequent steps are continued.

[0136] For example, Figure 2 and Figure 3In the second step of the behavioral model, the Agent needs to calculate how much concession it can make in the new round of offer. Specifically, for the Agent, the opponent's response state, its own internal state, and the environment state it is in represent the opponent's will, its own expectation, and the environmental constraints it is subjected to. Therefore, the Agent can decide whether to make a concession and the degree of concession in the next negotiation by evaluating the opponent's response state, its own internal state, and the environment state it is in.

[0137] The opponent's response state O is used to evaluate the difference between the last offer A and the latest counter-offer B, and is defined as follows:

[0138] σ = 1 - (G(A0, B0) - G(A, B)) / G(A0, B0) (13)

[0139] where A0and B0represent the initial offer and counter-offer, and G(A, B) is the distance metric between the offer A and the counter-offer B on the negotiation issue I i ∈ X, and is calculated as follows:

[0140]

[0141] where A i and B i represent the probability distribution of A and B on the negotiation issue I i ∈ X, and N i represents the N i negotiation issues, i.e., the negotiation goals.

[0142] The opponent's concession state D is determined by the concession value γ. The counter-offer B and the opponent's overall satisfaction with the first counter-offer B0determine the size of the concession value γ. The opponent's overall satisfaction with the counter-offer is obtained according to formula (7), and is defined as formula (14).

[0143]

[0144] The Agent's own internal state I involves the satisfaction level ρ related to the latest offer A and the tightness δ of a set of alternative solutions, where:

[0145] ρ = Ψ(S * ) (16)

[0146] δ = 1 - (ρ - ε) (17) where S * represents the Agent's desired solution, Ψ(S * ) represents its satisfaction with the solution S * , and ε represents the overall satisfaction threshold.

[0147] In the SDM negotiation process, the environmental constraint E suffered by the Agent is mainly time constraint. Therefore, we can use a function to represent the time constraint suffered by the Agent, i.e.:

[0148]

[0149] In the above formula, t now represents the current negotiation time, t max represents the deadline of negotiation, t represents the time constraint suffered by the Agent in the negotiation process, and a, b are constants, and b > 1, 0 ≤ a ≤ 1.

[0150] According to the above formula, the opponent response state O, the internal state I of itself, and the environmental state E can be calculated, based on which the concession value is calculated. The calculation formula of the concession value Δε of the Agent in the negotiation is as follows:

[0151] Δε=(μ σ (σ)Λμ γ (γ)Λμ ρ (ρ)Λμ δ (δ)Λμ t (t)) ω (19)

[0152] In the formula, μ σ (σ), μ ρ (ρ), μ δ (δ), μ t () and μ γ (γ) respectively represent the concession expectation corresponding to the difference degree, the satisfaction degree, the closeness degree, the time constraint and the concession value, and ω represents the weight vector related to the size of the concession value in the negotiation process, wherein ω = [w1, w2, w3, w4].

[0153] ω is related to the negotiation strategy. When ω > 1, the Agent adopts a competitive strategy; when ω = 1, the Agent adopts a win-win strategy; and when ω < 1, the Agent adopts a cooperative strategy.

[0154] After the concession value of the new round of offer is calculated, the satisfaction threshold can be updated. Specifically, given the negotiation concession value Δε and the overall satisfaction threshold ε, the new overall satisfaction threshold ε can be updated by formula (20).

[0155] ε * =ε-Δε (20)

[0156] S43、According to the updated satisfaction threshold, the feasible solution set and the expected solution set are obtained.

[0157] On the basis of the above-mentioned embodiments, in one of the embodiments of the present application, the step S43 comprises a step S431 and a step S432.

[0158] S431, according to the updated satisfaction threshold, obtain the feasible solution set P. Wherein, the solving model of the feasible solution is P = Γ (Π, ε * ) = {S | (S ∈ Π) Λ (ε ≥ Ψ (S) ≥ ε * )}, wherein, S represents a feasible solution, Π represents the intention of the fuzzy constraint network, ε represents the satisfaction threshold of the last offer, Ψ (S) represents the satisfaction of the offer S, and ε * represents the updated satisfaction threshold.

[0159] S432, according to the feasible solution set, obtain the expected solution set S * . Wherein, the solving model of the expected solution set is S * = arg (max S∈P H (S, B), wherein, S ∈ P represents a feasible solution, H (S, B) represents the similarity of the feasible solution and the last offer of the opposite party, N i represents the number of negotiation topics, W1 (S i ) represents the preference function on topic i, W2 (S i , B i ) represents the similarity function on topic i, ω1 represents the weight related to preference, and ω2 represents the weight related to similarity.

[0160] In this embodiment, as shown in Figure 2 and Figure 3 , in the third step of the behavior model, a new feasible solution needs to be generated according to the updated concession value and the opponent preference model. Specifically:

[0161] Given the intention Π of the fuzzy constraint network N and the latest behavior state ε * of the agent, the corresponding feasible solution set P can be generated. The feasible solution P is defined as follows:

[0162] P = Γ (Π, ε * ) = {S | (S ∈ Π) Λ (ε ≥ Ψ (S) ≥ ε * )} (21)

[0163] Wherein, Ψ (S) represents the satisfaction of the agent's target set in N.

[0164] Under the premise of giving the offer B and the feasible solution set P, the selection of the expected solution S * follows the following conditions:

[0165] S * = arg (max S∈PH(S, B) (22)

[0166] H(S, B) is a utility function, which is used to evaluate the similarity of feasible solution S e P relative to the offer B, and its definition is as follows:

[0167]

[0168] In the above formula, W1 is the preference function of the Agent on the issue i, and W2 is the similarity function, which measures the difference between the scheme S and the offer B:

[0169]

[0170] In the formula, ω1 and ω2 represent the weights related to preference and similarity, respectively.

[0171] The selection of the values of ω1 and ω2 is related to the negotiation strategy adopted by the Agent. The win-win strategy: ω1≤1.0, ω2≤1.0, ω1=ω2. The cooperative strategy: ω1≤1.0, ω2≤1.0, ω1<ω2. The competitive strategy: ω1≥1.0, ω2≥1.0, ω1>ω2. Using the win-win strategy for negotiation means that the Agent will consider the interests of the opponent Agent while considering its own interests, and expects to obtain a negotiation result that both parties are "satisfied" with. Using the cooperative strategy means that the Agent will consider the interests of the opponent Agent more in the negotiation process in order to reach an agreement as soon as possible. Using the competitive strategy for negotiation means that the Agent pays more attention to its own interests in the negotiation and hopes to maximize the interests obtained.

[0172] S44, generate a new offer according to the feasible solution set and the expected solution set, and calculate the self-offering satisfaction degree of the new offer, and then judge whether the self-offering satisfaction degree is greater than the updated satisfaction threshold. When it is judged that the self-offering satisfaction degree is greater than the updated satisfaction threshold, the new offer is output. Otherwise, the new offer is sent to the other party.

[0173] On the basis of the above embodiment, in an embodiment of the present application, the step S44 comprises steps S441 to S443.

[0174] S441, according to the feasible solution set and the expected solution set, a new asking price for i negotiation issues is obtained. The obtaining model of the new asking price is A * = ∧ (P, S * ), in the formula, A * represents the new asking price, P represents the feasible solution set, S * represents the expected solution set.

[0175] S442, the self-offering satisfaction degree of the new offer is calculated.

[0176] S443: Determine whether the satisfaction level of the bid is greater than the updated satisfaction threshold. If the satisfaction level of the bid is greater than the updated satisfaction threshold, output a new bid. Otherwise, send the new bid to the other party.

[0177] In this embodiment, if Figure 2 and Figure 3 As shown, in the fourth step of the behavioral model, a new quotation needs to be generated based on the updated feasible solution set and the expected solution set. Specifically:

[0178] Given feasible solution set P and expected solution set S*, about topic I i A set of asking prices for ∈X The generation of can be defined as follows:

[0179] A * =∧(P,S * ) (25) In the above formula, Topic I i ∈X's asking price and set A * Chinese elements Correspondingly, is value The marginal probability distribution in space X can be defined as:

[0180]

[0181] in, yes In space Cylindrical expansion, X i is the object of topic i, N X is the total number of objects.

[0182] It should be noted that in the behavioral model, DA and PA will continuously exchange offers and counteroffers until they reach an agreement or no new offers / counteroffers are generated. Given a feasible solution set P and a counteroffer B, the negotiation will terminate in two states: reaching an agreement or failing to reach an agreement.

[0183] The following conditions must be met to reach a consensus:

[0184] Ψ(S * )≥ε * (27) Negotiation failure meets the following conditions:

[0185]

[0186] Specifically, if the Agent's overall satisfaction value for the opponent's asking price B or the asking price S to be sent to the opponent in the next round *If the satisfaction value of the DA is greater than the overall satisfaction threshold of the new round, it indicates that the agent agrees to accept the offer / counteroffer, that is, an agreement is reached between the DA and the PA, which also means that the fuzzy constraint satisfaction problem is solved. Failure to reach an agreement between the DA and the PA indicates that there is no agreement between the DA and the PA, which may be due to the fact that the satisfaction threshold of the new round is less than 0, or the feasible solution set is empty .

[0187] The agent negotiation method of the embodiment of the application models the SDM problem into a distributed fuzzy constraint satisfaction problem based on agents according to the real doctor-patient joint decision-making scenario, thereby providing a theoretical basis for solving the doctor-patient joint decision-making problem. The opponent model is constructed using the Bayesian learning technique, and the limited information (for example, the counteroffer of the opponent) is fully utilized to enable the agent to continuously improve the understanding of the preference information of the opponent during the negotiation process. The influence of emotions and biases on decision-making is reduced, and the quality and efficiency of healthcare decision-making are improved. The negotiation behavior of the DA and the PA and the negotiation protocol followed by the DA and the PA are defined in combination with the opponent preference model, thereby providing a negotiation model for solving the doctor-patient joint decision-making problem (SDMP). The negotiation model based on Bayesian and fuzzy constraint agents (BLFCAN) is proposed to realize doctor-patient joint decision-making and alleviate or even eliminate the factors such as the unequal status of doctors and patients, the lack of doctor-patient communication skills, the asymmetry of doctor-patient information, and the limited time for doctor visits in the implementation process of doctor-patient joint decision-making. An agreement is reached on the issues of concern to both the doctor and the patient through the continuous interaction between the DA and the PA.

[0188] It can be understood that, in order to implement the agent negotiation method of the embodiment of the application:

[0189] First, according to the characteristics of the SDM, the SDMP is converted and modeled into a DFCSP, and the complete agent automatic negotiation content of the SDM is established, as shown in FIG. 1. Figure 4 , Figure 5 We model the SDMP as a DFCSP with fuzzy constraints between agents and other agents, and between issues and other issues. The goal of each agent is to establish a behavior model with fuzzy constraints, and the mutual constraint relationship between each agent determines whether there is a solution that satisfies all the constraint conditions of the DFCSP. The DA and the PA can solve this DFCSP through negotiation.

[0190] Then, the opponent preference model based on Bayesian learning is constructed. We assume the weight and the fuzzy membership function (i.e., the preference function) of the opponent, and set the same initial probability for each assumed value, as shown in FIG. 2. Figure 6As the negotiation process proceeds, the probability of each hypothesis is updated using Bayes' rule based on the opponent's offers. Combining each hypothesis and its probability, the true preference function and weights of the opponent are approximated more and more closely, as Figure 7 .

[0191] Then, the BLFCAN negotiation model of SDM is constructed, and the doctor and patient are modeled as DA and PA, and the main framework of negotiation decision is provided for them, as Figure 2 The DA and PA interact with the opponent through the steps of offer evaluation, opponent preference model, concession value calculation, feasible solution generation, offer generation and negotiation termination judgment. In this process, the DA or PA first evaluates the satisfaction of the received offer, and then evaluates the response state of the opponent agent, the internal state of itself, the concession amplitude of the opponent agent and the state of the environment, decides whether to make concessions in the next round of negotiation and the degree of concession, that is, the size of the concession value. Then, based on the size of the concession value, a new behavior state is determined, and a set of feasible solutions are generated, and a solution more in line with the preferences of both parties (i.e. satisfaction) is selected to generate an offer / offer to send to the opponent agent. When this offer / offer cannot be accepted by the opponent agent, the agent will make an offer based on the negotiation strategy, and consider solutions with the same satisfaction level, or provide solutions with lower satisfaction, and send them to the opponent. Before the termination condition (the negotiation reaches an agreement or fails) is met, the above negotiation process is repeatedly repeated.

[0192] After determining the negotiation process of DA and PA, the negotiation protocol to be followed by DA and PA in the negotiation process is defined, which is used to handle the interaction between agents in the negotiation process, and it is essentially the rules that all agents must follow, which defines all interactions between agents and determines the sequence and structure of messages between agents. The specific negotiation steps and negotiation protocol are shown in Figure 3 . DA and PA reach an agreement on the values of each issue according to the negotiation protocol and negotiation steps, and generate the values of the treatment plan. The plan has a certain support and reference for shared decision-making between doctors and patients.

[0193] In order to facilitate the understanding of the present application, the following actual example is used to illustrate the application of the present embodiment.

[0194] Actual example: The experiment of the method is performed by taking the doctor-patient shared decision-making of children's asthma as an example, and the experiment is divided into two parts.

[0195] The first part provides an example to explain the proposed FCAN negotiation model and recommendation model, and introduces the negotiation process and recommendation process of the method. Assuming that a 9-year-old child has severe asthma, which can reach level 4, the doctor and the patient (patient's family) need to set the corresponding preferences when seeking medical treatment, and DA and PA will negotiate on their behalf. The preferences of the doctor and the patient for the treatment plan and the relevant information of the issues involved are shown in Tables 2 and 3, and the fuzzy membership function of their preferences is shown in Figure 8 According to the provided doctor-patient joint decision-making model based on Bayesian learning and fuzzy constraint Agent, DA and PA can be constructed to represent the doctor and the patient for negotiation. DA and PA can negotiate according to the steps and protocols provided, and finally obtain a negotiation result. Figure 3

[0196] Table 2 DA negotiation issue preference setting

[0197]

[0198] Table 3 PA negotiation issue preference setting

[0199]

[0200] The negotiation process is as follows:

[0201] Assuming that DA starts the negotiation. The DA's starting negotiation will produce the following results.

[0202] In the first round:

[0203] Without negotiation history, DA / PA provides their expected offer to the other party, [cost: 4.5, effectiveness: 7.0, side effects: 0.1, risk: 0.1, convenience: 7.0]. As Figure 9 shown.

[0204] In the second round:

[0205] DA first evaluates PA's counteroffer by formula (12). The DA's satisfaction threshold for the feasible solution in the first round is 1. According to formulas (12)-(19), the new satisfaction threshold in the second round is 0.975. DA / PA updates the new feasible solution set and the expected solution of all issues according to formulas (21)-(24), as Figure 10 shown. The DA's new feasible solution set update process in Cost is shown in Figure 11 Then, DA generates a new offer for the feasible solution set and the expected solution according to formulas (25) and (26). Finally, DA compares according to formula (27). PA's counteroffer does not meet their new satisfaction threshold (0.975). Therefore, DA will continue to negotiate and send a new offer to PA. ​

[0206] The calculation method of the next round of negotiation is the same as the second round. The results of the negotiation process are summarized in Table 4, Table 5. Until the 7th round, the satisfaction value of the DA evaluation of the PA offer is greater than the new threshold 0.373 by 0.561, and an agreement is reached. The negotiation result is now [cost: 4.3, effectiveness: 9.0, side effects: 0.05, risk: 0.08, convenience: 9.0].

[0207] Table 4 Negotiation process of DA

[0208]

[0209] Table 5 Negotiation process of PA

[0210]

[0211] The second part of the experiment is to compare the three negotiation strategies involved in the method proposed in the application: cooperation, win-win, and competition under the condition of increasing negotiation topics.

[0212] We first adjust the parameter settings to change the negotiation strategies of DA and PA by changing ω1 and ω2 (formula (23)). In the experiment, the strategy parameters are set as (1) ω1 = 0.2, ω2 = 1.0, which is a cooperative strategy; (2) ω1 = ω2 = 1.0, which is a win-win strategy; (3) ω1 = 1.8, ω2 = 1.0, which is a competitive strategy.

[0213] Next, we study the convergence speed of negotiation by changing the concession value of the negotiation strategy parameter ω in formula (19). Figure 11 The number of negotiation rounds and the size of the concession value when the negotiation strategy is a cooperative strategy (ω = 0.6, 0.8), a win-win strategy (ω = 1.0), and a competitive strategy (ω = 1.2, 1.4) are shown. When ω = 0.6 or 0.8, the concession value of each round is greater than that of other strategies, resulting in fewer negotiation rounds than other strategies. When the ω value gradually increases to 1.2 or 1.4, i.e., when the model adopts a competitive strategy, the concession value of each round is significantly smaller, and the number of negotiation rounds increases.

[0214] The agent negotiation method of the embodiment of the application can simulate the complete process of doctor-patient joint decision-making, effectively alleviate the negotiation pressure brought by the increase of negotiation topics, and has a certain learning ability, so that humans are relieved from tedious negotiation affairs.

[0215] Embodiment two,

[0216] As Figure 12 shown, the embodiment of the application provides an agent negotiation device for doctor-patient shared decision-making, which comprises:

[0217] An initial information acquisition module 1 is configured to acquire preference data and membership functions of two participants in a shared decision-making process with respect to i negotiation topics, respectively.

[0218] A behavior model construction module 2 is configured to construct a behavior model of each participant based on a fuzzy constraint network according to the preference data and the membership functions.

[0219] An initial offer module 3 is configured to cause each behavior model to generate an offer according to its own preference data and send the offer to the other behavior model.

[0220] A repeated offer module 4 includes the following units, and each behavior model is configured to repeat the following units until the other behavior model judges that the satisfaction degree of the offer of the other behavior model or the satisfaction degree of the offer of the behavior model itself is greater than a satisfaction threshold.

[0221] A first updating unit 41 is configured to calculate the satisfaction degree of the offer of the other behavior model and update the preference model of the other behavior model based on the offer sent by the other behavior model.

[0222] A second updating unit 42 is configured to update the satisfaction threshold based on the preference model of the other behavior model and judge whether the satisfaction degree of the offer of the other behavior model is greater than the updated satisfaction threshold. When it is judged that the satisfaction degree of the offer of the other behavior model is greater than the updated satisfaction threshold, the offer of the other behavior model is output. Otherwise, the following steps are continued.

[0223] A third updating unit 43 is configured to acquire a feasible solution set and an expected solution set based on the updated satisfaction threshold.

[0224] A fourth updating unit 44 is configured to generate a new offer based on the feasible solution set and the expected solution set, calculate the satisfaction degree of the new offer of the behavior model, and judge whether the satisfaction degree of the new offer of the behavior model is greater than the updated satisfaction threshold. When it is judged that the satisfaction degree of the new offer of the behavior model is greater than the updated satisfaction threshold, the new offer is output. Otherwise, the new offer is sent to the other behavior model.

[0225] Specifically, the agent negotiation device of the embodiment of the present application can simulate the complete process of shared decision-making of doctors and patients, can effectively alleviate the negotiation pressure caused by the increase of negotiation topics, and has a certain learning ability, so that humans are relieved from tedious negotiation affairs.

[0226] Embodiment three,

[0227] The embodiment of the present application provides an agent negotiation device for shared decision-making of doctors and patients, which comprises a processor, a memory, and a computer program stored in the memory. The computer program can be executed by the processor to implement the agent negotiation method as described in any one of the embodiments one.

[0228] Embodiment four,

[0229] The embodiment of the present application provides a computer readable storage medium, the computer readable storage medium comprises a stored computer program, wherein the computer program controls a device where the computer readable storage medium is located to execute the proxy negotiation method as any one of the embodiments one to twenty.

[0230] The above merely provides the preferred embodiment of the present application, but should not be used to limit the present application. For those skilled in the art, various modifications and changes can be made to the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A proxy negotiation method for patient-doctor shared decision making, characterized by, The method comprises the following steps: respectively obtaining preference data and membership functions of two participants of a doctor-patient shared decision making on a negotiation issue respectively obtaining preference data and membership functions of two participants of a doctor-patient shared decision making on a negotiation issue According to the preference data and the membership function, behavior models of the two participants are respectively constructed based on a fuzzy constraint network; The two behavior models are respectively used to generate quotes according to their own preference data and send them to the other party; The two behavior models are respectively used to repeat the following steps until one of the behavior models judges that the other party's quote satisfaction or the own party's quote satisfaction is greater than a satisfaction threshold value; According to the other party's sent quote, the other party's quote satisfaction is calculated and the other party's preference model is updated based on Bayesian learning; The satisfaction threshold value is updated according to the other party's preference model, and it is judged whether the other party's quote satisfaction is greater than the updated satisfaction threshold value; when it is judged that the other party's quote satisfaction is greater than the updated satisfaction threshold value, the other party's quote is output; otherwise, the subsequent steps are continued to be executed; According to the updated satisfaction threshold value, a feasible solution set and an expected solution set are obtained; According to the feasible solution set and the expected solution set, a new quote is generated, the own party's quote satisfaction of the new quote is calculated, and then it is judged whether the own party's quote satisfaction is greater than the updated satisfaction threshold value; when it is judged that the own party's quote satisfaction is greater than the updated satisfaction threshold value, the new quote is output; otherwise, the new quote is sent to the other party; The partner preference model is a partner satisfaction estimation model; wherein the partner satisfaction estimation model is: wherein, an estimated value of partner satisfaction, a number of negotiation topics, an estimated value of partner weight, a number of hypotheses, a probability of a hypothesis of partner weight preference, a hypothesis of partner weight preference, an estimated value of partner fuzzy membership, a probability of a hypothesis of partner membership function shape, a hypothesis of partner membership function shape; According to the other party's sent quote, the other party's quote satisfaction is calculated and the other party's preference model is updated based on Bayesian learning, which specifically comprises: Calculate the other party's quotation satisfaction based on the quotation sent by the other party ; Among them, satisfaction with the other party's quotation , where It is a quotation Middle The degree of affiliation of the negotiation topic, For the The weight factor corresponding to each negotiation topic, is the number of negotiation topics; Based on the quote sent by the other party, the probability of the hypothesis of the other party's membership function shape in the other party's preference model is updated based on the Bayesian learning rule. The updated model of the hypothesis of the probability of the other party's membership function shape is: , where Expressing assumptions Bid in case of The probability of represents the probability of the hypothesis of the other party's membership function shape, P represents the probability, Indicate a bid Part of the expected utility, The assumptions about the shape of the other party's membership function, represents the estimated value of the other party's weight, represents the number of hypotheses; According to the offer sent by the other party, the probability of the hypothesis of the other party's weight preference in the other party's preference model is updated based on the Bayesian learning rule; wherein the update model of the probability of the hypothesis of the other party's weight preference is: , wherein, represents the probability of the hypothesis that the other party offers , represents the probability of the hypothesis of the other party's weight preference, represents the probability, represents the part of the expected utility of the offer , represents the hypothesis of the other party's weight preference, represents the estimated value of the other party's fuzzy membership degree, represents the number of hypotheses. The satisfaction threshold value is updated according to the other party's preference model, which specifically comprises: According to the own offer and the offer of the other party, calculate the difference degree ; wherein the difference degree is , , in the formula, denotes and between the distance measure, denotes the first offer of oneself, denotes the first offer of the other party, denotes and between the distance measure, denotes the last offer of oneself, denotes the last offer of the other party, denotes the number of negotiation topics, denotes and distance, denotes the possibility distribution on the negotiation topic, denotes the possibility distribution on the negotiation topic; According to the opposite preference model, the opposite concession value is calculated ; wherein the concession value is ; wherein represents the opposite offer satisfaction degree of the latest offer of the opposite party, represents the opposite offer satisfaction degree of the first offer of the opposite party; obtaining a desired solution set of the other party and calculating a satisfaction level of the desired solution set ; wherein , wherein is the satisfaction level of the desired solution set According to the satisfaction level, a compactness is calculated ; wherein the compactness calculation model is wherein, is the satisfaction level of the expected solution set, is the satisfaction threshold of the latest bid. Computing time constraints ; wherein the computing model of time constraints is , where denotes the current negotiation time, denotes the negotiation deadline, and are constants, , ; According to the difference degree , the opponent concession value , the satisfaction level , the closeness and the time constraint, a self concession value is calculated; wherein the self concession value is , wherein, represents the difference degree corresponding concession expectation, represents the opponent concession value corresponding concession expectation, represents the satisfaction level corresponding concession expectation, represents the closeness corresponding concession expectation, ) represents the time constraint corresponding concession expectation; According to the self concession value , update the satisfaction threshold; wherein the satisfaction threshold of the first offer is 1; the update model of the satisfaction threshold is , wherein, represents the updated satisfaction threshold, represents the satisfaction threshold of the last offer, represents the self concession value.

2. The agent negotiation method for shared decision making between doctor and patient according to claim 1, characterized in that, The negotiation topic includes cost, effectiveness, side effect, risk, and convenience; The preference data includes a preference value range, a favorite range, a minimum preference value, a maximum preference value, and a weight preference value; The fuzzy constraint network is , where Indicates the A fuzzy constraint network, Representing fuzzy constraint networks The domain of Representing fuzzy constraint networks tuples of all distinct objects in Representing fuzzy constraint networks The set of fuzzy constraints, Indicates medical providers, Indicates patient; The two behavior models respectively generate quotes according to their own preference data and send them to the other party, which specifically comprises: The two behavior models respectively generate quotes according to the upper limit and / or the lower limit of their own preference data and send them to the other party.

3. The agent negotiation method for patient shared decision making according to claim 1, wherein, According to the updated satisfaction threshold value, a feasible solution set and an expected solution set are obtained, which specifically comprises: According to the updated satisfaction threshold, obtain a feasible solution set ; wherein, the solving model of the feasible solution is , wherein, represents the feasible solution, represents the intention of the fuzzy constraint network, represents the satisfaction threshold of the last offer, represents the satisfaction of the asking price S, represents the updated satisfaction threshold; According to the feasible solution set, an expected solution set is obtained ; wherein, a solution model of the expected solution set is , , in the formula, represents a feasible solution, represents the similarity of the feasible solution and the last offer, represents the number of negotiation topics, represents a preference function on the topic , represents a similarity function on the topic , represents a weight related to the preference, represents a weight related to the similarity.

4. The agent negotiation method for patient shared decision making according to claim 1, wherein, According to the feasible solution set and the expected solution set, a new quote is generated, the own party's quote satisfaction of the new quote is calculated, and then it is judged whether the own party's quote satisfaction is greater than the updated satisfaction threshold value; when it is judged that the own party's quote satisfaction is greater than the updated satisfaction threshold value, the new quote is output; otherwise, the new quote is sent to the other party. According to the feasible solution set and the desired solution set, a new asking price about the negotiation topic is obtained; wherein, the obtaining model of the new asking price is , wherein, , wherein, represents the new asking price, represents the feasible solution set, represents the desired solution set.

5. A proxy negotiation device for shared decision making between a patient and a doctor, adapted to perform the proxy negotiation method for shared decision making between a patient and a doctor according to any one of claims 1 to 4, characterized in that, The method comprises the following steps: The initial information acquisition module is used to obtain the information of the two participants in the doctor-patient shared decision-making process. Preference data and membership functions for negotiation topics; The behavior model component module is used to construct behavior models of the two participants based on a fuzzy constraint network according to the preference data and the membership function; The initial quote module is used for the two behavior models to generate quotes according to their own preference data and send them to the other party; The circulating quote module comprises the following units, which are used for the two behavior models to repeat the following units until one of the behavior models judges that the other party's quote satisfaction or the own party's quote satisfaction is greater than a satisfaction threshold value; The first updating unit is used to calculate the other party's quote satisfaction according to the other party's sent quote and update the other party's preference model based on Bayesian learning; The second updating unit is used to update the satisfaction threshold value according to the other party's preference model and judge whether the other party's quote satisfaction is greater than the updated satisfaction threshold value; when it is judged that the other party's quote satisfaction is greater than the updated satisfaction threshold value, the other party's quote is output; otherwise, the subsequent steps are continued to be executed; The third updating unit is used to obtain a feasible solution set and an expected solution set according to the updated satisfaction threshold value; A fourth updating unit is configured to generate a new offer according to the feasible solution set and the desired solution set, calculate a self-offer satisfaction degree of the new offer, and then determine whether the self-offer satisfaction degree is greater than an updated satisfaction threshold; when it is determined that the self-offer satisfaction degree is greater than the updated satisfaction threshold, output the new offer; otherwise, send the new offer to the other party.

6. A proxy negotiation device for shared decision making between a patient and a physician, characterized by The computer program can be executed by the processor to implement the agent negotiation method for shared decision-making between a doctor and a patient according to any one of claims 1 to 4.

7. A computer-readable storage medium, characterized in that, The computer readable storage medium comprises a stored computer program, wherein the computer readable storage medium controls a device where the computer readable storage medium is located to execute the agent negotiation method for shared decision-making between a doctor and a patient according to any one of claims 1 to 4 when the computer program is running.

Images