An artificial intelligence assisted medication service development method and system

Abstract

The present application relates to a kind of artificial intelligence assisted medication service development method and system, the method comprises: creating the agent that produces medication service information;Build the multi-dimensional relationship matrix of the association between agents;One or more agents are instantiated based on electronic medical record to constitute user agent set;Determine one or more agent processing paths based on user agent set;Determine a medication service record based on each processing path;Further decision to obtain medication service auxiliary information.The present application is based on the construction of lightweight agent, by constructing the association between agents, the exponential complexity is reduced to linear complexity, improve the feasibility and medication service efficiency, promote the development of medication service.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of artificial intelligence, and particularly relates to a method and system for developing medication service assisted by artificial intelligence. BACKGROUND

[0002] In the treatment of stroke patients, the role of clinical pharmacists is indispensable. They ensure that patients correctly understand and follow medical advice through precise drug selection and reasonable dose adjustment, which is crucial for improving treatment effectiveness and reducing the risk of stroke recurrence. The differences in patients' understanding of medical information highlight the urgent need for efficient and personalized medication education.

[0003] In recent years, generative artificial intelligence (GenAI) technology has been increasingly applied in the medical field, particularly in the preparation of patient education materials. GenAI models have shown great potential in efficiently integrating complex medical information and generating scientifically accurate and easy-to-understand educational content. Moreover, they can customize personalized education programs based on patients' specific circumstances, effectively improving the relevance and acceptability of education. In addition, GenAI's expertise in language conversion enables it to translate professional terms into languages that patients can easily understand, greatly improving the readability and practicality of educational materials. In the field of medicine and artificial intelligence, the provision and application of medication service information are crucial. In the subfield of medication education for stroke patients, GenAI has a broad application prospect and is expected to become a disruptive force that enhances the effectiveness of pharmacist services and expands the coverage of pharmaceutical services. After patients are discharged, medication service needs to be carried out for them, and a medication service list needs to be issued for medication assistance, which can greatly save doctors' time, improve medication service efficiency, and promote the development of medication service. In existing technology, when artificial intelligence technology is used for medication service, it is usually hoped that the intelligent agent constructed by artificial intelligence will give a consistent solution for the electronic medical record. In fact, on the one hand, during the vertical application of artificial intelligence technology in the medical field, training data is a scarce resource, which is difficult to label and has low consistency. Therefore, it is difficult to effectively train complex intelligent agents. On the other hand, the condition of inpatients is often complex, and the symptoms and manifestations are also complex and diverse. As the number of symptoms increases, the training amount and computational load of intelligent agents also increase exponentially. Therefore, it is difficult to complete the training of such intelligent agents and put them into actual use in practical applications. Based on the above problems, the present application constructs a lightweight intelligent agent, reduces the exponential complexity to linear complexity by constructing the association between intelligent agents, improves the feasibility and medication service efficiency, and promotes the development of medication service. SUMMARY

[0004] To address the aforementioned problems in the prior art, this invention proposes an artificial intelligence-assisted method and system for providing medication services, the method comprising:

[0005] Step S1: Construct an agent whose input is an electronic medical record and which generates medication service information in a service scenario; create a two-dimensional coordinate (x, y) for each agent as an identifier; where: the x-axis is the drug type and the y-axis is the patient feature classification;

[0006] Step S2: Construct a multidimensional relation matrix representing the relationships between agents; the multidimensional relation matrix [ra i，j The relationship ra in ] i，j Used to indicate the relationship between agent i and agent j; i and j are both used to refer to two-dimensional coordinates (x, y);

[0007] Step S3: Input the electronic medical record, instantiate one or more intelligent agents based on the electronic medical record to form a user intelligent agent set; determine one or more intelligent agent processing paths based on the user intelligent agent set; determine a medication service record based on each processing path;

[0008] Step S4: Make a decision on one or more medication service records obtained from one or more intelligent processing paths to obtain medication service auxiliary information; present the medication service auxiliary information to the user.

[0009] Furthermore, the ranges of patient characteristic values ​​for the y-axis may overlap or not overlap.

[0010] Furthermore, when the y-coordinate divisions are non-overlapping, the same patient feature may fall into the patient feature classification range corresponding to two or more agent identifiers.

[0011] Furthermore, the input to the intelligent agent includes the patient's electronic medical records, medical data, medication information, or current prescriptions.

[0012] Furthermore, the intelligent agent includes one or more of the following: rule-driven intelligent agent, data-driven intelligent agent, knowledge graph intelligent agent, artificial intelligence intelligent agent, and drug interaction detection intelligent agent.

[0013] Furthermore, the instantiation of one or more intelligent agents based on electronic medical records specifically involves: obtaining characterization indicators corresponding to various drug types from the electronic medical records; when the characterization indicators in the electronic medical records conform to the scope of action of the drug type, determining the intelligent agent identifier based on the conforming drug type and patient characteristics; instantiating the intelligent agent corresponding to the intelligent agent identifier and adding it to the user intelligent agent set.

[0014] The determining one or more agent processing paths specifically comprises: determining weight values of the agents, and selecting an agent with a weight value greater than a preset value as a starting agent; and determining an agent processing path from each starting agent, so that all agents in the processing path have different x coordinate values.

[0015] The determining an agent processing path from each starting agent specifically comprises: taking the user agent set as a current agent set; taking the starting agent as a current agent and initializing the agent processing path as containing only the current agent; determining, based on the multi-dimensional relationship matrix, an agent having the largest correlation with the current agent in the current agent set and different x coordinate values from all agents in the processing path as a next agent of the current agent; updating the current agent to the next agent; deleting the next agent from the current agent set, and repeating the process until the current agent set is empty; and connecting all next agents determined from the starting agent to form the determined agent processing path.

[0016] An artificial intelligence-assisted medication service development system for implementing the above artificial intelligence-assisted medication service development method.

[0017] An artificial intelligence-assisted medication service development device for implementing the above artificial intelligence-assisted medication service development method.

[0018] An artificial intelligence-assisted medication service development platform for implementing the above artificial intelligence-assisted medication service development method.

[0019] An artificial intelligence-assisted medication service development server for implementing the above artificial intelligence-assisted medication service development method.

[0020] The beneficial effects of the present application include:

[0021] (1) Constructing lightweight agents, comprehensively using these agents by constructing the correlation between the agents, thereby reducing the exponential complexity to linear complexity, improving the feasibility and medication service efficiency, and promoting the development of medication service.

[0022] (2) Through two-dimensional coordinates, the agent can be quickly positioned, and different agents with different y-axis values can provide drug service information for the same drug type, which relaxes the threshold for creating, training and generating agents, and provides drug service information diversity; based on path creation and node traversal, the relationship problem is converted into a data problem, so that multiple agents need to cooperate, which is convenient for decision compatibility and generation and processing of medication schemes. BRIEF DESCRIPTION OF DRAWINGS

[0023] The accompanying drawings, which are included to provide a further understanding of the application and constitute a part of this application, illustrate embodiments of the application and together with the description serve to explain the application, but do not limit the application.

[0024] Figure 1 A schematic diagram of an artificial intelligence assisted medication service development method is provided.

DETAILED DESCRIPTION

[0025] The application will be described in detail below with reference to the accompanying drawings and specific embodiments, wherein the schematic embodiments and descriptions are only used to explain the application, but do not limit the application.

[0026] The application provides an artificial intelligence assisted medication service development method and system, as shown in the accompanying drawings. Figure 1 The method comprises the following steps:

[0027] Step S1: constructing an agent whose input is an electronic medical record and generating medication service information in a service scenario; creating a two-dimensional coordinate (x, y) for each agent as an identifier; wherein: the x-axis is the drug type, and the y-axis is the patient characteristic classification, for example: the drug type includes antibiotics, cardiovascular drugs, and mental drugs; the y-axis is classified according to patient characteristic areas: age stratification (children / elderly), liver and kidney function classification, gender, genotype, and complications; service scenario classification includes dose generation, dose adjustment (drug and its measurement), incompatibility, medication time optimization, prescription review, dose conversion, and treatment planning; when the service scenario changes, the corresponding agent function or rule is switched to meet the required service scenario;

[0028] Preferably: different coordinate values are set for each drug type and patient characteristic classification to identify agents with different functions; further: agents with similar functions are set to have similar coordinate values; further: the drug classification coordinate value is a positive integer, and considering that the patient characteristic classification can overlap, the y-axis coordinate value can be set to a non-integer continuous value to identify the proximity between agents; through such a setting method, the agent can be quickly positioned;

[0029] Preferably: the patient feature value range of the y coordinate is overlapping or non-overlapping, that is, the patient feature classification (patient feature range) corresponding to the y coordinate value is overlapping or non-overlapping; the same patient feature classification (value, value range) corresponds to different y coordinates; therefore, the same patient can be provided with drug service information by different agents with different y coordinates for the same drug type; in this way, the threshold for creating and generating agents is relaxed, and the diversity of drug service information is provided; in this case, data from one institution, one clinic, and one department can support the training of agents;

[0030] Alternatively: the input of the agent includes the medical data of the patient, drug information, current prescription, etc., and the output is one or more of the drug dosage and recommendation, adjusted dose, drug recommendation, warning, etc.;

[0031] Preferably: the x coordinate and the y coordinate can be set to one-dimensional or multi-dimensional elements according to specific needs; for example: the patient feature classification can be set in multiple dimensions to finely describe the user; the two-dimensional coordinates can be used to locate the agent for generating drug service information, that is, the two-dimensional coordinates are unique;

[0032] Preferably: the agent is set or trained using historical data that meets the agent identification limit; the agent is maintained, reduced in training, or deleted based on the proportion of the number of times the agent is instantiated and the proportion of the number of times the drug service information generated by the agent is finally used, and the agent is optimized to optimize the software and hardware storage space; Since the patient feature classification in the agent can be overlapping, the data used for training different agents can obviously be the same;

[0033] Preferably: the agent includes one or more of a rule-driven agent, a data-driven agent, a knowledge graph agent, an artificial intelligence agent, and a drug interaction detection agent; the rule-driven agent generates drug service information based on explicit knowledge of clinical guidelines / pharmacopoeia; for example: Drools rule engine, FHIR clinical pathway model, SWRL semantic reasoning; the data-driven agent performs implicit knowledge mining based on a machine learning model; for example: XGBoost / LightGBM (structured data), Transformer time series model, DeepSurv survival analysis, chemotherapy dose prediction agent; the knowledge graph agent is based on the association reasoning of the medical knowledge network; for example: Neo4j graph database, TransE / KG-BERT embedding engine, SPARQL query engine; the hybrid enhanced agent makes collaborative decisions based on multiple types of agents;

[0034] Preferably, the intelligent agent is an artificial intelligence model-based intelligent agent; in constructing the intelligent agent, a base model is selected based on an intelligent agent creation platform, the base model is further selected through a single-agent mode, appropriate plug-ins are installed on the base model, and the plug-ins are parameterized; the plug-ins are parameterized; specifically, parameters such as learning rate, maximum input length, response generation temperature, etc. are adjusted to optimize the fluency and accuracy of the dialogue; for example, the learning rate is set between 0.001 and 0.01 to avoid instability in training caused by excessively high learning rate; the maximum input length is set according to the hardware and software resources to avoid invalid settings that increase resource consumption; the response generation temperature is set to a temperature value (0 to 1) to control the randomness of the generated text, with a low value being more conservative and a high value being more creative, for example, the initial setting is 0.7;

[0035] Step S2: constructing a multi-dimensional relationship matrix representing the association relationship between intelligent agents; the association relationship rai,j in the multi-dimensional relationship matrix [ra i，j ] can be understood that i and j are used to represent two-dimensional coordinates (x, y) identification; i，j is used to indicate the association relationship between intelligent agent i and intelligent agent j; it can be understood that i and j are used to represent two-dimensional coordinates (x, y) identification;

[0036] Preferably, when the x coordinate values of two intelligent agents are the same, the association relationship is set to 0;

[0037] Alternatively, set i = hash(x, y); hash() is a two-dimensional hash code, which realizes the rapid positioning of intelligent agents through hash coding;

[0038] Preferably, the elements in the multi-dimensional relationship matrix are used to indicate the first or multiple of the drug use relationship, the drug interaction relationship, the patient characteristic similarity relationship, the medical order mode relationship, etc. between two intelligent agents; these association relationships can be indicated in the form of multiple groups or in the form of composite values; the relationship is represented by a numerical value, which is larger when the relationship is close, and smaller when the relationship is not close; further, the numerical value is normalized to a value between 0 and 1; the two-dimensional intelligent agent matrix is responsible for intelligent agent function diversity allocation and positioning, and the relationship matrix handles the association and knowledge transfer between intelligent agents; when processing a complex case, relevant intelligent agents are found from the two-dimensional intelligent agent matrix, and other associated intelligent agents are linked through the three-dimensional relationship matrix to provide service information;

[0039] Step S3: inputting an electronic medical record, instantiating one or more intelligent agents based on the electronic medical record to form a user intelligent agent set; determining one or more intelligent agent processing paths; determining one or more medication service records based on each processing path;

[0040] The electronic medical record is instantiated one or more agents, specifically: from the electronic medical record and various drug types corresponding to the characterization index, when the characterization index in the electronic medical record meets the drug type action range, determine the agent identifier based on the consistent drug type and patient characteristics; instantiate the intelligent agent corresponding to the intelligent agent identifier and join the user intelligent agent set; in the user intelligent agent set, there may be one or more intelligent agents for the same drug type; for example: when the age layer of intelligent agent A is 10-11 years old, and intelligent agent B is 9.5-11 years old, when the patient is 10 years old, intelligent agent A and intelligent agent B patient feature classification will be simultaneously classified;

[0041] The one or more agent processing paths are determined, specifically: determining the weight value of the intelligent agent, selecting the intelligent agent with a weight value greater than a preset value as the starting intelligent agent; determining an intelligent agent processing path from each starting intelligent agent, so that the x-coordinate values of all intelligent agents in the processing path are different;

[0042] Preferably, the weight value of the intelligent agent is set to a preset value; here, the expert experience value can be set;

[0043] Alternatively, the weight value of the intelligent agent is determined, specifically: for each intelligent agent, determine all characterization indexes corresponding to the drug type; determine the weight value of each intelligent agent based on the deviation of the characterization index and the normal index and the number of corresponding characterization indexes; for example: the weight value w of the intelligent agent (x, y) under the current electronic medical record is determined by the following formulas (1)-(2) x，y ; Wherein: d b is the deviation of the characterization index b and the normal index; w b is the weight value of the characterization index under the drug type; tdn b is the effective threshold of the characterization index b; N is the number of characterization indexes corresponding to the drug type; wb x，y is the basic weight of the drug type x;

[0044]

[0045] Preferably, the weight value corresponding to the deviation of each intelligent agent to different characterization indexes is set in advance;

[0046] The method comprises the following steps: determining an agent processing path from each starting agent; specifically, taking the user agent set as the current agent set, taking the starting agent as the current agent and initializing the agent processing path to contain only the current agent; determining, based on the multi-dimensional relationship matrix, an agent in the current agent set that has the largest correlation with the current agent (or randomly selecting one of the top several agents with the largest correlation, or randomly selecting one of the agents with a correlation greater than a threshold) and has a different x-coordinate value from all agents in the processing path, as the next agent of the current agent (that is, constructing an edge between the current agent and the next agent in the path); updating the current agent to the next agent; removing the next agent from the current agent set, and repeating the process until the current agent set is empty; connecting all the next agents determined from the starting agent to form a determined agent processing path;

[0047] Of course, when selecting from the current agent set, a relatively exclusive selection can be made according to the y-coordinate value. Within the selectable range, the selection with the y-coordinate value opposite to that of the current user agent set is selected, or the y-coordinate value that makes the user feature of the current user fall into the center position of the range is selected;

[0048] Alternatively, the weight values of the agents are clustered, and all agents in the cluster with the largest cluster mean are taken as starting agents. By clustering, typical drug types can be found at the same time. When there are multiple typical drug types, the stronger the correlation of their simultaneous action is, the less significant the distinction is.

[0049] The method comprises the following steps: determining an agent processing path from each starting agent; specifically, taking the user agent set as the current agent set, taking the starting agent as the current agent and initializing the agent processing path to contain only the current agent; determining, based on the multi-dimensional relationship matrix, an agent in the current agent set that has the largest correlation with the current agent (or randomly selecting one of the top several agents with the largest correlation, or randomly selecting one of the agents with a correlation greater than a threshold) and has a different x-coordinate value from all agents in the processing path, as the next agent of the current agent (that is, constructing an edge between the current agent and the next agent in the path); updating the current agent to the next agent; removing the next agent from the current agent set, and repeating the process until the current agent set is empty; connecting all the next agents determined from the starting agent to form a determined agent processing path;

[0050] Step S3A1: initializing the current node as the starting agent; initializing the existing medication service information as empty;

[0051] Step S3A2: input the electronic medical record to the agent corresponding to the current node to update the medication service information; specifically, input the electronic medical record to the agent corresponding to the current node to obtain the medication service information corresponding to the agent, judge whether the medication service information meets the existing medication service information (there is no contradiction between the existing medication service information), if yes, combine the medication service information with the existing medication information to update the medication service information; otherwise, input the electronic medical record to the agent corresponding to the current node to obtain the replaced medication service information corresponding to the agent; repeat the step until the existing medication service information is met; since the processing method for the same disease type is various, obviously, the agent can obtain multiple different medication service information based on the same electronic medical record; during the medication service information updating process, the agent needs to passively replace the medication service information because its priority is lower than that of the precedent node (precedent agent);

[0052] Alternatively, the input of the electronic medical record to the agent corresponding to the current node to update the medication service information; specifically, set the service scene (set to metering adjustment) to input the electronic medical record to the corresponding agent to obtain the medication service information corresponding to the agent under the restriction condition of the existing medication service information, combine the medication service information with the existing medication information to update the medication service information; this alternative method increases the complexity of the agent to a certain extent, and additional service scene setting is required, thereby additional agent logic is required;

[0053] Step S3A3: judge whether the processing path is traversed, if no, set the current node as the next agent of the current node, and return to step S3A2; if yes, record the current medication service information as the medication service record of the processing path;

[0054] Step S4: make a decision on the one or more medication service records obtained by the one or more intelligent processing paths to obtain medication service auxiliary information; present the medication service auxiliary information to the user; specifically, select one medication record from the multiple medication service records as the medication service auxiliary information;

[0055] Alternatively, sequentially select, combine or eliminate contradictions for the medication service information of the same medication type x in the one or more medication service records to obtain the medication service auxiliary information for the medication service auxiliary information containing all medication types x; further, present the medication service auxiliary information to the user after manual decision;

[0056] Alternatively, the decision on the one or more medication service records obtained by the one or more intelligent processing paths to obtain the medication service auxiliary information; specifically, represent the medication service record corresponding to the pth processing path as Re p= (re p，k ), wherein: re p，k is the medication service information given by the kth agent in the processing path p; all medication service information re p，k involved in the medication type x is summed up to obtain the medication service information rep,k involved in the medication type x; the medication service information rep,k with the highest score sc p，k in the P medication service information in p is taken as the medication service information for the medication type x; all medication types x are the number of all medication types involved in the user agent set; of course, since the medication types are not continuous, the corresponding X is also not continuous;

[0057] Preferably, the score sc p，k of the medication service information rep,k involved in the medication type x is calculated based on the following formula (3) (4) or (3) (5): p，k ; the medication service information with the highest score is taken as the medication service information for the medication type x;

[0058] sc p，k = ∑ (p1＝1～P,k 1 ＝1～K ) ∧((p1,k1)≠(p，k)) tem.sc p1,k1 (3);

[0059]

[0060] The medication service assistance information is presented to the user; specifically, a one-key medication service assistance information presentation is provided, a one-key button is presented when the patient is discharged, and the medication service assistance information is presented to the user in a modifiable manner when clicked;

[0061] Based on the same inventive concept, the present application also provides an artificial intelligence assisted medication service development system, which is used to complete the above-mentioned artificial intelligence assisted medication service development method;

[0062] A computer program (also referred to as a program, software, software application, script, or code) can be written in any form of programming language, including an assembly or an interpreted language, a declarative or a procedural language, and it can be deployed in any form, including as a stand-alone program or as a module, component, subroutine, object, or other unit suitable for use in a computing environment. The computer program can but does not have to correspond to a file in a file system. A program can be stored in a portion of a file that holds other programs or data (e.g., one or more scripts stored in a markup language document), in a single file dedicated to the program in question, or in multiple coordinated files (e.g., files that store one or more modules, sub programs, or portions of code). The computer program can be deployed to execute on one computer or on multiple computers that are located at one site or distributed across multiple sites and interconnected by a communication network.

[0063] Those skilled in the art will appreciate that embodiments of the present application can be readily used as a method, apparatus such as a system, or computer program product. Accordingly, the present application can take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application can take the form of a computer program product on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROMs, optical storage devices, etc.) embodying computer readable program code.

[0064] The present application is described in reference to the drawings using a flowchart and / or a block diagram of methods, apparatus (systems) and computer program products according to embodiments of the application. It will be understood that each block of the flowchart and / or block diagram, and combinations of blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general purpose computer, special purpose computer, embedded processing system or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart and / or block diagram block or blocks. Figure 1 one or more functions specified in one or more of the flowchart or block diagram block or blocks. Figure 1 means for carrying out each of the one or more functions specified in the flowchart or block diagram block or blocks.

[0065] These computer program instructions can also be stored in a computer- readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instructions which implement the flowchart or block diagram block or blocks. Figure 1 one or more functions specified in one or more of the flowchart or block diagram block or blocks. Figure 1 means for carrying out each of the one or more functions specified in the flowchart or block diagram block or blocks.

[0066] These computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the flowchart or block diagram block or blocks. Figure 1 one or more functions specified in one or more of the flowchart or block diagram block or blocks. Figure 1 means for carrying out each of the one or more functions specified in the flowchart or block diagram block or blocks.

[0067] Finally, it should be noted that the above-described embodiments are merely intended for describing and illustrating, not limiting, the technical solution of the present application. Although the present application has been described in detail with reference to the above embodiments, those skilled in the art should understand that the specific embodiments of the present application can be modified or replaced by equivalents without departing from the spirit and scope of the present application, and any modification or equivalent replacement without departing from the spirit and scope of the present application should be covered in the protection scope of the claims of the present application.

Claims

1. An artificial intelligence-assisted medication service development method, characterized by, The method comprises: Step S1: Constructing the agent whose input is electronic medical record and which generates medication service information in the service scenario; creating two-dimensional coordinates for each agent As an identifier; wherein: The axis is the drug type, The axis is the patient characteristic classification; the patient characteristic numerical value range of the y coordinate is overlapping or non-overlapping; when the y coordinate division is non-overlapping, the same patient characteristic falls into the patient characteristic classification range corresponding to two or more agent identifiers; Step S2: Construct a multidimensional relation matrix representing the relationships between agents; the multidimensional relation matrix Relationships in Used to instruct intelligent agents and intelligent agents The relationship between them; i and j are both used to refer to two-dimensional coordinates. Logo; Step S3: inputting an electronic medical record, instantiating one or more agents based on the electronic medical record to constitute a user agent set; determining one or more agent processing paths based on the user agent set; determining a medication service record based on each processing path; Step S4: making a decision on the one or more medication service records obtained by the one or more intelligent processing paths to obtain medication service auxiliary information; and presenting the medication service auxiliary information to the user; The instantiation of the one or more agents based on the electronic medical record comprises: obtaining, from the electronic medical record, a representation index corresponding to each drug type; when the representation index in the electronic medical record meets the action range of the drug type, determining an agent identifier based on the drug type and the patient characteristics that meet the requirement; instantiating the agent corresponding to the agent identifier and adding the agent to the user agent set; The determination of the one or more agent processing paths comprises: determining a weight value of the agent, and selecting an agent with a weight value greater than a preset value as a starting agent; and determining an agent processing path from each starting agent, so that the x-coordinate values of all agents in the processing path are different; The determination of the agent processing path from each starting agent comprises: taking the user agent set as a current agent set; taking the starting agent as a current agent and initializing the agent processing path to contain only the current agent; determining, based on a multi-dimensional relationship matrix, an agent that has the largest association relationship with the current agent in the current agent set and has an x-coordinate value different from those of all agents in the processing path, as a next agent of the current agent; updating the current agent to the next agent; deleting the next agent from the current agent set; and repeating the process until the current agent set is empty; and connecting all next agents determined from the starting agent to constitute the determined agent processing path.

2. The AI-assisted medication service development method according to claim 1, characterized in that, The agent comprises one or more of a rule-driven agent, a data-driven agent, a knowledge graph agent, an artificial intelligence agent, and a drug interaction detection agent.

3. An artificial intelligence-assisted medication service development system, characterized by, The artificial intelligence-assisted medication service system is used to implement the artificial intelligence-assisted medication service method in any one of claims 1-2.

4. An artificial intelligence-assisted medication service development device, characterized by The artificial intelligence-assisted medication service device is used to implement the artificial intelligence-assisted medication service method in any one of claims 1-2.

5. An artificial intelligence assisted medication service development platform, characterized in that, The artificial intelligence-assisted medication service platform is used to implement the artificial intelligence-assisted medication service method in any one of claims 1-2.

6. An artificial intelligence-assisted medication service development server, characterized by, The artificial intelligence-assisted medication service server is used to implement the artificial intelligence-assisted medication service method in any one of claims 1-2.

Images