Decision-making method and device for multiple medication of old people based on large language model

Abstract

The invention provides a decision-making method and device for multiple medication of old people based on a large language model, and relates to the technical field of data processing, user information of a target user is input into a key information extraction model, and key medical information of the target user is obtained; performing vectorization processing on the key medical information through an open source vector model to obtain key medical vector information; in a vector database, acquiring clinical pathways of multiple medication and complication matched with the key medical vector information, and converting the clinical pathways into triple information and corresponding description information through a knowledge graph large language model; wherein the triple information comprises nodes, relationships and attributes; and on the basis of an EviMed base model, in combination with the triple information and the description information, summarizing and generating a clinical pathway text, and generating a clinical pathway reasoning result of multiple medication and complication management corresponding to the user information.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular to a decision-making method and device for elderly polypharmacy based on a large language model. Background Art

[0002] Elderly patients often suffer from multiple chronic diseases. Studies have shown that more than 60% of elderly patients are involved in polypharmacy, which may significantly increase the risk of drug interactions and side effects, which may aggravate health problems and lead to additional medical interventions. Polypharmacy may increase the risk of adverse drug effects, and effective polypharmacy management is essential.

[0003] Therefore, how to effectively assist in compound medication decision-making has become an urgent problem to be solved in the industry. Summary of the invention

[0004] The present invention provides a decision-making method and device for multiple medications for the elderly based on a large language model, which are used to solve the problem of how to effectively assist in compound medication decision-making in the prior art.

[0005] The present invention provides a decision-making method for elderly polypharmacy based on a large language model, comprising: Inputting user information of a target user into a key information extraction model to obtain key medical information of the target user; After vectorizing the key medical information through an open source vector model, key medical vector information is obtained; In the vector database, clinical pathways of multiple medications and comorbidities matching the key medical vector information are obtained, and the clinical pathways are converted into triple information and corresponding description information through the knowledge graph language model; wherein the triple information includes: nodes, relationships, and attributes; Based on the EviMed base model, the clinical pathway text is summarized and generated in combination with the triple information and the description information, and the clinical pathway reasoning results of multiple medication and comorbidity management corresponding to the user information are generated.

[0006] According to a decision method for elderly polypharmacy based on a large language model provided by the present invention, before the step of inputting the user information of the target user into a key information extraction model to obtain the key medical information of the target user, the method further includes: Acquire multiple user sample information with manual markings, wherein each of the user sample information includes: user basic information, symptom and diagnosis information, and medication information; Based on the multiple user sample information carrying manual labels, a key information extraction model including PubMedBERT and BioMistral models is trained, and when a first preset training condition is met, the training is stopped to obtain a trained key information extraction model; Among them, PubMedBERT is the main line key information recognition model, and the BioMistral model is a medical generation model, which is responsible for calculating the entity coverage of the input text.

[0007] According to a decision method for elderly polypharmacy based on a large language model provided by the present invention, after the key medical information is vectorized by an open source vector model, before the step of obtaining the key medical vector information, the method further includes: Inputting user information into a key information extraction model to obtain key medical information of the user information; The user information and the key medical information are vectorized using the open source vector model BGE model to construct a vector database.

[0008] According to a decision method for elderly polypharmacy based on a large language model provided by the present invention, based on the EviMedbase model, the clinical pathway text is summarized and generated in combination with the triple information and the description information, and before the step of generating the clinical pathway reasoning results of polypharmacy and comorbidity management corresponding to the user information, the method also includes: Verifying and comparing the triplet information with key medical information corresponding to the clinical pathway; In case of successful verification, the triplet information is saved; or in case of failed verification, the triplet information is deleted.

[0009] The present invention also provides a decision-making device for elderly polypharmacy based on a large language model, comprising: An input module, used to input user information of a target user into a key information extraction model to obtain key medical information of the target user; A processing module, used for vectorizing the key medical information through an open source vector model to obtain key medical vector information; A conversion module is used to obtain clinical pathways of multiple medications and comorbidities matching the key medical vector information in a vector database, and convert the clinical pathways into triple information and corresponding description information through a knowledge graph large language model; wherein the triple information includes: nodes, relationships, and attributes; A generation module is used to summarize and generate clinical pathway text based on the EviMed base model in combination with the triple information and description information, and to generate clinical pathway reasoning results for multiple medication and comorbidity management corresponding to user information.

[0010] The system is also used to: Acquire multiple user sample information with manual markings, wherein each of the user sample information includes: user basic information, symptom and diagnosis information, and medication information; Based on the multiple user sample information carrying manual labels, a key information extraction model including PubMedBERT and BioMistral models is trained, and when a first preset training condition is met, the training is stopped to obtain a trained key information extraction model; Among them, PubMedBERT is the main line key information recognition model, and the BioMistral model is a medical generation model, which is responsible for calculating the entity coverage of the input text.

[0011] The system is also used to: Inputting user information into a key information extraction model to obtain key medical information of the user information; The user information and the key medical information are vectorized using the open source vector model BGE model to construct a vector database.

[0012] The system is also used to: Verifying and comparing the triplet information with key medical information corresponding to the clinical pathway; In case of successful verification, the triplet information is saved; or in case of failed verification, the triplet information is deleted.

[0013] The present invention also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the decision-making method for multiple medications in the elderly based on a large language model as described above is implemented.

[0014] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements any of the above-described decision-making methods for multiple medications in the elderly based on a large language model.

[0015] The present invention also provides a computer program product, including a computer program, which, when executed by a processor, implements any of the above-mentioned decision-making methods for multiple medications in the elderly based on a large language model.

[0016] The decision-making method and device for polypharmacy in the elderly based on a large language model provided by the present invention utilizes large model technology, and the system conducts an in-depth search of the local medical knowledge base to obtain medical knowledge and previous cases related to the input information. The system uses advanced algorithms and reasoning capabilities to analyze the patient's specific situation, and combines the information in the medical knowledge base to generate personalized polypharmacy pathways and comorbidity management plans. Large model technology can identify and prevent potential medication errors, such as drug interactions, inappropriate dosages, etc., and reduce the risks of polypharmacy in elderly patients. The system assists doctors in considering individual differences in patients, including factors such as genes, lifestyles, and disease severity, and provides personalized treatment recommendations. The system has the ability to process data in real time, can respond quickly to changes in patient conditions, and provide doctors with the latest treatment information and recommendations. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0018] Figure 1 A flowchart of a decision-making method for elderly polypharmacy based on a large language model provided in an embodiment of the present application; Figure 2 The overall technical flow chart provided for the embodiments of the present application; Figure 3 A schematic diagram of the structure of a decision-making device for elderly polypharmacy based on a large language model provided in an embodiment of the present application; Figure 4 It is a structural schematic diagram of the electronic device provided by the present invention. DETAILED DESCRIPTION

[0019] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0020] Figure 1 A flowchart of a decision-making method for elderly polypharmacy based on a large language model provided in an embodiment of the present application is shown in FIG. Figure 1 As shown, including: Step 110, inputting the user information of the target user into a key information extraction model to obtain the key medical information of the target user; Step 120, vectorizing the key medical information through an open source vector model to obtain key medical vector information; Step 130, obtaining clinical pathways of multiple medications and comorbidities matching the key medical vector information in the vector database, and converting the clinical pathways into triple information and corresponding description information through the knowledge graph large language model; wherein the triple information includes: nodes, relationships, and attributes; Step 140, based on the EviMed base model, combined with the triple information and the description information, a summary of the clinical pathway text is generated to generate a clinical pathway reasoning result of multiple medication and comorbidity management corresponding to the user information.

[0021] In the embodiment of the present application, the personal information and medical records of the target user are input into the key information extraction model. The key information extraction model (such as the fusion model of PubMedBERT and BioMistral) is used to analyze the user information and extract key medical information, such as disease diagnosis, symptoms, medication history, etc.

[0022] The extracted key medical information is vectorized through an open source vector model (such as the BGE model) and converted into a mathematical vector form to facilitate efficient information retrieval and matching.

[0023] In the constructed vector database, matching clinical pathways are retrieved based on key medical vector information. This step usually involves calculating the similarity between vectors, such as using cosine similarity. The retrieved clinical pathway information is input into a knowledge graph large language model (such as the EviMed model) to convert it into structured triple information, including nodes (entities), relationships, and attributes.

[0024] The triple information clearly describes the relationship between entities, such as (patient, suffering from, disease), (drug, used to treat, disease), etc. Using the EviMed base model, the triple information and the corresponding description information are combined to generate a detailed clinical pathway text. This step may involve natural language processing technology to ensure the readability and logic of the text.

[0025] According to the text generated by the EviMed base model, the clinical pathway reasoning results for polypharmacy and comorbidity management corresponding to the user information are obtained.

[0026] More specifically, professionals can review the generated clinical pathway and make necessary adjustments based on the patient's specific circumstances and the physician's professional judgment.

[0027] In the embodiment of the present application, using big model technology, the system conducts an in-depth search of the local medical knowledge base to obtain medical knowledge and previous cases related to the input information. The system uses advanced algorithms and reasoning capabilities to analyze the patient's specific situation, and combines the information in the medical knowledge base to generate personalized multiple medication pathways and comorbidity management plans. Big model technology can identify and prevent potential medication errors, such as drug interactions, improper dosage, etc., and reduce the risks of multiple medications in elderly patients. The system assists doctors in considering individual differences in patients, including factors such as genes, lifestyle, and disease severity, and provides personalized treatment recommendations. The system has the ability to process data in real time, can respond quickly to changes in patient conditions, and provide doctors with the latest treatment information and recommendations.

[0028] Optionally, before the step of inputting the user information of the target user into the key information extraction model to obtain the key medical information of the target user, the method further includes: Acquire multiple user sample information with manual markings, wherein each of the user sample information includes: user basic information, symptom and diagnosis information, and medication information; Based on the multiple user sample information carrying manual labels, a key information extraction model including PubMedBERT and BioMistral models is trained, and when a first preset training condition is met, the training is stopped to obtain a trained key information extraction model; Among them, PubMedBERT is the main line key information recognition model, and the BioMistral model is a medical generation model, which is responsible for calculating the entity coverage of the input text.

[0029] Specifically, based on the existing local knowledge base and real clinical data, data related to polypharmacy and comorbidities of elderly patients are extracted to build a clinical medical record database of polypharmacy and comorbidities of elderly patients, which contains outpatient information text such as patient age, clinical diagnosis, comorbidity diagnosis, medication records, clinical treatment plans, regular review and tracking, etc. The acquisition of data in this database has been authorized by users in a compliant manner.

[0030] Based on the above dataset, 10,000 data were randomly selected and manually labeled and verified by clinical experts to construct a training dataset for the key information extraction model. The model aims to extract entity information such as diseases, symptoms, diagnoses, complications, medications, and efficacy from the dataset text.

[0031] Prepare a key information extraction model that includes PubMedBERT and BioMistral models to train and learn how to extract key medical information from text.

[0032] The collected manually labeled user sample information is used to train the key information extraction model. PubMedBERT is used as the mainline key information recognition model, focusing on identifying key medical entities in the text; BioMistral is used as a medical generation model to calculate the entity coverage of the input text.

[0033] After the data is fed into the model, it will enter two branch models. PubMedBERT is the main key information recognition model, and the BioMistral model is a medical generation model responsible for calculating the entity coverage of the input text. Here, the probability calculation of the token corresponding to the entity extracted by its output layer is normalized to a weight of 0-1 to obtain the weight of the entity coverage. The PubMedBERT loss function uses softmax. By weighting the PubMedBERT loss function and the entity coverage calculated by the BioMistral model, the final loss formula in the supervised learning process of the PubMedBERT model is obtained: The sample set is recorded as: The basic loss is: ; The final loss formula is: ; in, is the weight coefficient, which can be adjusted according to the training effect. is the current text entity coverage weight.

[0034] Optionally, after vectorizing the key medical information through an open source vector model and before obtaining the key medical vector information, the method further includes: Inputting user information into a key information extraction model to obtain key medical information of the user information; The user information and the key medical information are vectorized using the open source vector model BGE model to construct a vector database.

[0035] In the embodiment of the present application, a key information extraction model is used to extract key medical information from the original text, such as diseases, symptoms, diagnoses, comorbidities, medications, etc. These key information are input into the BGE model together with the original text for vectorization processing to generate a numerical vector representation. A vector database is constructed to store all vectorized medical information, laying the foundation for subsequent retrieval and matching.

[0036] When a user enters specific disease or drug information, the system uses the vector database for retrieval. The cosine distance is used to calculate the similarity between the vector representation of the user's input information and the vector in the database to find the most relevant medical information. Multiple pieces of data most relevant to the user's input are retrieved and recalled, and these data will be used to generate the clinical pathway.

[0037] Use a large language model, such as fine-tuned Llama2, to process the retrieved data. Llama2 is fine-tuned using existing datasets and manual labeling to enable it to understand and summarize medical information. The knowledge data retrieved from the vector library is input into the fine-tuned Llama2 model to generate a clinical pathway for polypharmacy and comorbidities in elderly patients.

[0038] Optionally, based on the EviMed base model, in combination with the triple information and the description information, a summary of the clinical pathway text is generated, and before the step of generating a clinical pathway reasoning result of multiple medication and comorbidity management corresponding to the user information, the step further includes: Verifying and comparing the triplet information with key medical information corresponding to the clinical pathway; In case of successful verification, the triplet information is saved; or in case of failed verification, the triplet information is deleted.

[0039] In the embodiment of the present application, through the above steps, a preliminary clinical pathway for polypharmacy and comorbidities in elderly patients has been generated.

[0040] These initially generated clinical pathway texts are used as input to the EviMed large language model. The EviMed model is fine-tuned to enable it to better understand and process specific clinical pathway information. As a 60B parameter generative large language model, EviMed has powerful language understanding and generation capabilities.

[0041] The fine-tuned EviMed model converts text information into a structured triple form, including nodes (entities), relationships, and attributes. This conversion helps to clarify the relationships and attributes between different entities. At the same time, the model also generates description information corresponding to the triples, providing richer context and details.

[0042] Then, possible deviations are corrected by checking and comparing the key medical information corresponding to the clinical pathway. Through checking and comparing, it is ensured that the triple information finally included in the clinical pathway reasoning is a strongly related entity, that is, they are closely related to the patient's specific condition and treatment needs.

[0043] Construct a knowledge graph and integrate triple information and description information to form a structured knowledge base to support clinical decision-making. Use the verified and optimized triple information and description information to generate the final clinical pathway reasoning results, which will be more accurate and personalized.

[0044] Figure 2 The overall technical flow chart provided for the embodiment of this application is as follows: Figure 2 As shown, based on the clinical information database and user search conditions related to polypharmacy and comorbidities, a key information extraction model is generated, and then a vector database is obtained. Based on the search results, the clinical pathway decision expert model is input to obtain the primary clinical pathway text information, and then the knowledge graph large language model is input to obtain accurate triple information and text description information, and a summary of the clinical pathway results for polypharmacy and comorbidities in elderly patients is generated.

[0045] The decision-making device for multiple medications in the elderly based on a large language model provided by the present invention is described below. The decision-making device for multiple medications in the elderly based on a large language model described below and the decision-making method for multiple medications in the elderly based on a large language model described above can be referenced to each other.

[0046] Figure 3 A schematic diagram of the structure of a decision-making device for elderly polypharmacy based on a large language model provided in an embodiment of the present application is shown in FIG. Figure 3 As shown, including: The input module 310 is used to input the user information of the target user into the key information extraction model to obtain the key medical information of the target user; The processing module 320 is used to vectorize the key medical information through an open source vector model to obtain key medical vector information; The conversion module 330 is used to obtain the clinical pathways of multiple medications and comorbidities matching the key medical vector information in the vector database, and convert the clinical pathways into triple information and corresponding description information through the knowledge graph language model; wherein the triple information includes: nodes, relationships, and attributes; The generation module 340 is used to generate a summary of the clinical pathway text based on the EviMed base model in combination with the triple information and the description information, and to generate clinical pathway reasoning results for multiple medication and comorbidity management corresponding to the user information.

[0047] The system is also used to: Acquire multiple user sample information with manual markings, wherein each of the user sample information includes: user basic information, symptom and diagnosis information, and medication information; Based on the multiple user sample information carrying manual labels, a key information extraction model including PubMedBERT and BioMistral models is trained, and when a first preset training condition is met, the training is stopped to obtain a trained key information extraction model; Among them, PubMedBERT is the main line key information recognition model, and the BioMistral model is a medical generation model, which is responsible for calculating the entity coverage of the input text.

[0048] Optionally, the system is further used for: Inputting user information into a key information extraction model to obtain key medical information of the user information; The user information and the key medical information are vectorized using the open source vector model BGE model to construct a vector database.

[0049] Optionally, the system is further used for: Verifying and comparing the triplet information with key medical information corresponding to the clinical pathway; In case of successful verification, the triplet information is saved; or in case of failed verification, the triplet information is deleted.

[0050] In the embodiment of the present application, using big model technology, the system conducts an in-depth search of the local medical knowledge base to obtain medical knowledge and previous cases related to the input information. The system uses advanced algorithms and reasoning capabilities to analyze the patient's specific situation, and combines the information in the medical knowledge base to generate personalized multiple medication pathways and comorbidity management plans. Big model technology can identify and prevent potential medication errors, such as drug interactions, improper dosage, etc., and reduce the risks of multiple medications in elderly patients. The system assists doctors in considering individual differences in patients, including factors such as genes, lifestyle, and disease severity, and provides personalized treatment recommendations. The system has the ability to process data in real time, can respond quickly to changes in patient conditions, and provide doctors with the latest treatment information and recommendations.

[0051] Figure 4 is a schematic diagram of the structure of the electronic device provided by the present invention, such as Figure 4As shown, the electronic device may include: a processor 410, a communication interface 420, a memory 430 and a communication bus 440, wherein the processor 410, the communication interface 420 and the memory 430 communicate with each other through the communication bus 440. The processor 410 may call the logic instructions in the memory 430 to execute a decision method for elderly polypharmacy based on a large language model, the method comprising: inputting user information of a target user into a key information extraction model to obtain key medical information of the target user; After vectorizing the key medical information through an open source vector model, key medical vector information is obtained; In the vector database, clinical pathways of multiple medications and comorbidities matching the key medical vector information are obtained, and the clinical pathways are converted into triple information and corresponding description information through the knowledge graph language model; wherein the triple information includes: nodes, relationships, and attributes; Based on the EviMed base model, the clinical pathway text is summarized and generated in combination with the triple information and the description information, and the clinical pathway reasoning results of multiple medication and comorbidity management corresponding to the user information are generated.

[0052] In addition, the logic instructions in the above-mentioned memory 430 can be implemented in the form of a software functional unit and can be stored in a computer-readable storage medium when it is sold or used as an independent product. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc. Various media that can store program codes.

[0053] On the other hand, the present invention further provides a computer program product, the computer program product includes a computer program, the computer program can be stored on a non-transitory computer-readable storage medium, when the computer program is executed by a processor, the computer can execute the decision-making method for elderly polypharmacy based on a large language model provided by the above methods, the method comprising: inputting user information of a target user into a key information extraction model to obtain key medical information of the target user; After vectorizing the key medical information through an open source vector model, key medical vector information is obtained; In the vector database, clinical pathways of multiple medications and comorbidities matching the key medical vector information are obtained, and the clinical pathways are converted into triple information and corresponding description information through the knowledge graph language model; wherein the triple information includes: nodes, relationships, and attributes; Based on the EviMed base model, the clinical pathway text is summarized and generated in combination with the triple information and the description information, and the clinical pathway reasoning results of multiple medication and comorbidity management corresponding to the user information are generated.

[0054] In another aspect, the present invention further provides a non-transitory computer-readable storage medium having a computer program stored thereon, which is implemented when executed by a processor to execute the decision-making method for elderly polypharmacy based on a large language model provided by the above methods, the method comprising: inputting user information of a target user into a key information extraction model to obtain key medical information of the target user; After vectorizing the key medical information through an open source vector model, key medical vector information is obtained; In the vector database, clinical pathways of multiple medications and comorbidities matching the key medical vector information are obtained, and the clinical pathways are converted into triple information and corresponding description information through the knowledge graph language model; wherein the triple information includes: nodes, relationships, and attributes; Based on the EviMed base model, the clinical pathway text is summarized and generated in combination with the triple information and the description information, and the clinical pathway reasoning results of multiple medication and comorbidity management corresponding to the user information are generated.

[0055] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Those of ordinary skill in the art may understand and implement it without creative effort.

[0056] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0057] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A decision-making method for elderly polypharmacy based on a large language model, characterized in that: include: Inputting user information of a target user into a key information extraction model to obtain key medical information of the target user; After vectorizing the key medical information through an open source vector model, key medical vector information is obtained; In the vector database, clinical pathways of multiple medications and comorbidities matching the key medical vector information are obtained, and the clinical pathways are converted into triple information and corresponding description information through the knowledge graph language model; wherein the triple information includes: nodes, relationships, and attributes; Based on the EviMed base model, the clinical pathway text is summarized and generated in combination with the triple information and the description information, and the clinical pathway reasoning results of multiple medication and comorbidity management corresponding to the user information are generated.

2. The decision-making method for elderly polypharmacy based on a large language model according to claim 1, characterized in that: Before the step of inputting the user information of the target user into the key information extraction model to obtain the key medical information of the target user, the method further includes: Acquire multiple user sample information with manual markings, wherein each of the user sample information includes: user basic information, symptom and diagnosis information, and medication information; Based on the multiple user sample information carrying manual labels, a key information extraction model including PubMedBERT and BioMistral models is trained, and when a first preset training condition is met, the training is stopped to obtain a trained key information extraction model; Among them, PubMedBERT is the main line key information recognition model, and the BioMistral model is a medical generation model, which is responsible for calculating the entity coverage of the input text.

3. The decision-making method for elderly polypharmacy based on a large language model according to claim 2, characterized in that: After the key medical information is vectorized by an open source vector model, before the step of obtaining the key medical vector information, the step further includes: Inputting user information into a key information extraction model to obtain key medical information of the user information; The user information and the key medical information are vectorized using the open source vector model BGE model to construct a vector database.

4. The decision-making method for elderly polypharmacy based on a large language model according to claim 1, characterized in that: Based on the EviMed base model, combining the triple information and the description information, generating a summary of the clinical pathway text, and generating a clinical pathway reasoning result of multiple medication and comorbidity management corresponding to the user information, the step also includes: Verifying and comparing the triplet information with key medical information corresponding to the clinical pathway; In case of successful verification, the triplet information is saved; or in case of failed verification, the triplet information is deleted.

5. A decision-making device for elderly patients with multiple medications based on a large language model, characterized in that: include: An input module, used to input user information of a target user into a key information extraction model to obtain key medical information of the target user; A processing module, used for vectorizing the key medical information through an open source vector model to obtain key medical vector information; A conversion module is used to obtain clinical pathways of multiple medications and comorbidities matching the key medical vector information in a vector database, and convert the clinical pathways into triple information and corresponding description information through a knowledge graph large language model; wherein the triple information includes: nodes, relationships, and attributes; A generation module is used to summarize and generate clinical pathway text based on the EviMed base model in combination with the triple information and description information, and to generate clinical pathway reasoning results for multiple medication and comorbidity management corresponding to user information.

6. The decision-making device for elderly polypharmacy based on a large language model according to claim 5, characterized in that: The system is also used to: Acquire multiple user sample information with manual markings, wherein each of the user sample information includes: user basic information, symptom and diagnosis information, and medication information; Based on the multiple user sample information carrying manual labels, a key information extraction model including PubMedBERT and BioMistral models is trained, and when a first preset training condition is met, the training is stopped to obtain a trained key information extraction model; Among them, PubMedBERT is the main line key information recognition model, and the BioMistral model is a medical generation model, which is responsible for calculating the entity coverage of the input text.

7. The decision-making device for elderly polypharmacy based on a large language model according to claim 5, characterized in that: The system is also used to: Inputting user information into a key information extraction model to obtain key medical information of the user information; The user information and the key medical information are vectorized using the open source vector model BGE model to construct a vector database.

8. The decision-making device for elderly multiple medication based on a large language model according to claim 5, characterized in that: The system is also used to: Verifying and comparing the triplet information with key medical information corresponding to the clinical pathway; In case of successful verification, the triplet information is saved; or in case of failed verification, the triplet information is deleted.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, the decision-making method for multiple medications in the elderly based on a large language model as described in any one of claims 1 to 4 is implemented.

10. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, it implements the decision-making method for polypharmacy in the elderly based on a large language model as described in any one of claims 1 to 4.

Images