Drg group recommendation method, device and equipment based on large model and medium

Abstract

The application provides a DRG grouping recommendation method, device, equipment and medium based on a large model, wherein the method comprises the following steps: based on a medical model, extracting a disease from a medical record of a patient to be recommended to obtain a diagnosed disease, and selecting a main diagnosis based on the diagnosed disease, a diagnosis disease list and the medical record to obtain a main diagnosis disease; and performing DRG grouping recommendation based on the main diagnosis disease to obtain a DRG grouping recommendation result of the patient to be recommended, which overcomes the defects of poor clinical adaptability, easy omission of important information, and insufficient accuracy and reliability of the current DRG grouping, starts from the perspective of diagnosis and treatment, excludes diagnoses based on insufficient medical models, thereby optimizing and improving the diagnosis disease list, and on this basis, the main diagnosis is selected and the DRG grouping is recommended according to the medical record, which can realize accurate and reliable DRG grouping, has strong explainability and adaptability in the clinical use process, and can meet the actual application requirements.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data processing, and in particular to a DRG group recommendation method and device based on a large model, equipment and medium. BACKGROUND

[0002] With the continuous progress of medical technology, the DRG (Diagnosis-Related Group) medical insurance payment system has been widely used. The accuracy of DRG group is of great significance to the compliance of medical insurance payment, the efficiency of hospital operation and the medical experience of patients. However, DRG group is facing many challenges in actual operation.

[0003] The first problem of DRG group is the mass and complexity of medical record information, which makes it difficult for traditional solutions to obtain important information and make correct DRG group recommendations. For example, the scheme based on artificial rules is difficult to adapt to the complex and variable clinical actual situation, and is prone to errors. The scheme based on deep learning directly predicts DRG group has certain prediction ability, but its explainability is poor, and it is difficult to meet the standard requirements of medical insurance payment and clinical management. The scheme based on DRG grouper often lacks comprehensive support of medical record information, and is easy to miss important clinical information of patients, resulting in absurd errors. SUMMARY

[0004] The present application provides a DRG group recommendation method and device based on a large model to solve the problems of poor clinical adaptability and explainability of existing DRG group solutions, which are prone to absurd errors in actual use, and lack of accuracy and reliability.

[0005] The present application provides a DRG group recommendation method based on a large model, comprising:

[0006] determining the medical record of the patient to be recommended, and the diagnosis disease list corresponding to the medical record;

[0007] based on the medical model, extracting diseases from the medical record to obtain confirmed diseases, and selecting the main diagnosis based on the confirmed diseases, the diagnosis disease list and the medical record to obtain the main diagnosis disease of the patient to be recommended;

[0008] based on the main diagnosis disease, performing disease diagnosis related group (DRG) group recommendation to obtain the DRG group recommendation result of the patient to be recommended.

[0009] According to the DRG group recommendation method based on a large model provided by the present application, the main diagnosis disease of the patient to be recommended is obtained by selecting the main diagnosis based on the confirmed disease, the diagnosis disease list and the medical record, comprising:

[0010] determine a target diagnosis disease list based on the confirmed disease and the diagnosis disease list;

[0011] generate a summary based on the target diagnosis disease list and the medical record to obtain a diagnosis summary corresponding to each diagnosis disease in the target diagnosis disease list;

[0012] select a main diagnosis based on the each diagnosis disease and the corresponding diagnosis summary to obtain a main diagnosis disease of the patient to be recommended.

[0013] According to the DRG group recommendation method based on a large model provided by the application, the summary is generated based on the target diagnosis disease list and the medical record to obtain a diagnosis summary corresponding to each diagnosis disease in the target diagnosis disease list, which comprises:

[0014] determine diagnosis knowledge corresponding to each diagnosis disease, and perform vectorization processing on the diagnosis knowledge and the medical record to obtain diagnosis knowledge vectors and medical record segment vectors;

[0015] perform medical record retrieval based on the diagnosis knowledge vectors and the medical record segment vectors to obtain medical record diagnosis segments corresponding to each diagnosis disease in the medical record;

[0016] generate a summary based on the medical record diagnosis segments corresponding to each diagnosis disease to obtain a diagnosis summary corresponding to each diagnosis disease in the target diagnosis disease list.

[0017] According to the DRG group recommendation method based on a large model provided by the application, the diagnosis summary comprises a diagnosis disease summary and a diagnosis treatment summary;

[0018] The main diagnosis of the patient to be recommended is selected based on the each diagnosis disease and the corresponding diagnosis summary, which comprises:

[0019] perform diagnosis effectiveness verification and treatment effectiveness verification based on the each diagnosis disease, the diagnosis disease summary corresponding to each diagnosis disease, and the diagnosis treatment summary to obtain a verification result;

[0020] determine a candidate diagnosis disease from the each diagnosis disease based on the verification result;

[0021] select a main diagnosis based on the candidate diagnosis disease and the corresponding diagnosis summary to obtain a main diagnosis disease of the patient to be recommended.

[0022] According to the DRG group recommendation method based on a large model provided by the application, the verification result comprises a diagnosis effectiveness verification result and a treatment effectiveness verification result;

[0023] The candidate diagnostic disease is determined from the diagnostic diseases based on the check result, and the candidate diagnostic disease is determined from the diagnostic diseases based on the check result.

[0024] In a case where the diagnostic validity check result and the treatment validity check result corresponding to any diagnostic disease are both check passed, the any diagnostic disease is taken as a candidate diagnostic disease.

[0025] According to the DRG group recommendation method based on a large model provided by the application, the diagnostic validity check and the treatment validity check are performed based on the diagnostic diseases and the diagnostic disease abstracts and the diagnosis and treatment abstracts corresponding to the diagnostic diseases, and the check result is obtained, and then the method further comprises the following steps.

[0026] The diagnostic diseases with the diagnostic validity check result passed in the check result are selected as the candidate diagnostic diseases from the diagnostic diseases.

[0027] The missed diagnosis recommendation is performed based on the candidate diagnostic diseases and the diagnostic disease list.

[0028] According to the DRG group recommendation method based on a large model provided by the application, the DRG group recommendation is performed based on the main diagnostic disease, and the DRG group recommendation result of the patient to be recommended is obtained, and the method comprises the following steps.

[0029] The DRG group recommendation is performed based on the main diagnostic disease, and the initial group recommendation result of the patient to be recommended is determined.

[0030] The initial group recommendation result is updated and optimized based on the confirmed diagnostic disease and other diagnostic diseases in the diagnostic disease list, and the DRG group recommendation result of the patient to be recommended is obtained.

[0031] The application further provides a DRG group recommendation device based on a large model, comprising:

[0032] The medical record determination unit is configured to determine the medical record of the patient to be recommended and a diagnostic disease list corresponding to the medical record.

[0033] The main diagnosis selection unit is configured to perform disease extraction on the medical record based on a medical model to obtain a confirmed diagnostic disease, and perform main diagnostic selection based on the confirmed diagnostic disease, the diagnostic disease list and the medical record to obtain the main diagnostic disease of the patient to be recommended.

[0034] The group recommendation unit is configured to perform DRG group recommendation based on the main diagnostic disease to obtain the DRG group recommendation result of the patient to be recommended.

[0035] The application further provides an electronic device, including a memory, a processor, and a computer program stored in the memory and running on the processor, and the processor implements the DRG grouping recommendation method based on a large model according to any one of the above when executing the computer program.

[0036] The application further provides a non-transitory computer-readable storage medium, which stores a computer program, and the computer program is executed by a processor to implement the DRG grouping recommendation method based on a large model according to any one of the above.

[0037] The DRG grouping recommendation method, device, equipment and medium based on a large model provided by the application can overcome the defects of poor clinical adaptability, easy omission of important information, obvious errors, insufficient accuracy and reliability of the current DRG grouping, and can realize accurate and reliable DRG grouping from the perspective of diagnosis and treatment, optimize and improve the diagnosis disease list by using the medical model to exclude diagnoses based on insufficient evidence, and can realize accurate and reliable DRG grouping based on the medical record for main diagnosis selection and DRG grouping recommendation, and has strong explainability and adaptability in the clinical use process, and can meet the actual application requirements. BRIEF DESCRIPTION OF DRAWINGS

[0038] In order to more clearly illustrate the technical solutions in the application or prior art, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings in the following description are some embodiments of the application, and those skilled in the art can obtain other drawings according to these drawings without creative labor.

[0039] Figure 1 is a flowchart of the DRG grouping recommendation method based on a large model provided by the application;

[0040] Figure 2 is a flowchart of the summary generation process provided by the application;

[0041] Figure 3 is a flowchart of the missed diagnosis recommendation provided by the application;

[0042] Figure 4 is a flowchart of the DRG grouping recommendation provided by the application;

[0043] Figure 5 is a structural diagram of the DRG grouping recommendation device based on a large model provided by the application;

[0044] Figure 6 FIG. 1 is a structural schematic diagram of an electronic device provided by the present application. DETAILED DESCRIPTION

[0045] In order to make the objects, technical solutions and advantages of the present application clearer, the technical solutions in the present application will be described clearly and completely below in combination with the drawings in the present application. Obviously, the described embodiments are some of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0046] At present, most of the existing DRG group recommendation schemes have limitations. For example, the scheme based on artificial rules is often difficult to adapt to complex and variable clinical actual situations, and is prone to errors; the scheme based on deep learning directly predicting the DRG group has certain prediction ability, but its explainability is poor, and it is difficult to meet the standard requirements of medical insurance payment and clinical management; and the scheme of the DRG grouper often lacks comprehensive medical record information support, resulting in missing important clinical information, making the errors obvious, and further affecting the medical insurance payment and the medical experience of patients.

[0047] To this end, the present application provides a DRG group recommendation method based on a large model, which aims to optimize and improve the diagnosis disease list from the perspective of diagnosis and treatment, and on this basis, combined with the medical record and the specific selection principle, to select the main diagnosis, so as to realize accurate DRG group recommendation, overcome the defects that the current DRG group recommendation scheme has poor clinical adaptability, is prone to miss important information, leads to obvious errors, and has insufficient accuracy and reliability, has strong explainability and clinical adaptability, and can meet the actual application requirements.

[0048] Figure 1 FIG. 1 is a structural schematic diagram of an electronic device provided by the present application. Figure 1 As shown in FIG. 1, the method comprises the following steps.

[0049] Step 110, determining the medical record of the patient to be recommended, and the diagnosis disease list corresponding to the medical record;

[0050] Step 120, based on the medical model, extracting diseases from the medical record to obtain the diagnosed diseases, and based on the diagnosed diseases, the diagnosis disease list and the medical record, selecting the main diagnosis to obtain the main diagnosis disease of the patient to be recommended;

[0051] Step 130, based on the main diagnosis disease, performing disease diagnosis related group DRG group recommendation to obtain the DRG group recommendation result of the patient to be recommended.

[0052] Specifically, considering the many problems existing in the current DRG group recommendation scheme in terms of clinical adaptability, standardization and explainability, and the adverse effects caused thereby, i.e. leading to DRG group recommendation errors, thereby affecting the normal payment of medical insurance, the medical experience of patients, and the normal operation of the medical financial system, etc., in the embodiment of the present application, it is proposed that the list can be optimized through a medical model, and by excluding diagnoses based on insufficient information, the diagnosis disease list can be improved, so that all diagnoses in the list are based on sufficient diagnoses, and on this basis, the main diagnosis is selected and DRG grouping is performed, to realize accurate DRG group recommendation, thereby avoiding the problems of current main diagnosis selection difficulty, easy to make mistakes, and DRG grouping errors, and at the same time, having strong explainability and excellent clinical adaptability.

[0053] It can be understood that in actual application process, before DRG group recommendation, the medical record information of the patient to be grouped needs to be determined, i.e. the patient to be recommended needs to be determined, and its medical record and the diagnosis disease list corresponding to the medical record are obtained.

[0054] It should be noted that the patient to be recommended here can be one or more, each patient to be recommended can have one or more medical records, and each medical record corresponds to a diagnosis disease list, which can be the diagnosis given by the doctor during the current and previous visits that the patient to be recommended may have, has a certain disease, i.e. the set of diagnosis diseases.

[0055] And after obtaining the medical record of the patient to be recommended and the diagnosis disease list corresponding to the medical record, the present embodiment can analyze it for subsequent DRG group recommendation. However, considering that DRG group needs to consider the overall information in the medical record, and even additional patient individual information, and the medical record information alone often reaches tens of thousands of words or even hundreds of thousands of words, the information is extremely complex, and various means available at present, such as artificial rules, learning prediction, etc. cannot handle this complex medical record information well, ultimately leading to the fact that the final group recommendation result is not completely accurate and reliable due to the limitation of the amount and quality of the input information during DRG group recommendation.

[0056] Based on this, in the embodiments of the present application, the powerful natural language processing and understanding ability of the large model can be used to sort and analyze the massive and complex medical record information to extract key information and help DRG grouping. Specifically, this can be based on the strong contextual understanding ability of the large language model (LLM, referred to as large model) in the field of natural language processing, as well as the problem analysis and processing ability, which is manifested in its strong environmental perception ability, autonomous understanding, decision making and execution ability. In order to better and more efficiently group DRGs, the present application proposes to build a medical model based on the large model, and process the medical records based on this to extract key information, thereby improving the problems existing in DRG grouping.

[0057] In detail, the medical model can be used to process the medical records to extract the determined diseases, so as to obtain the diagnosed diseases corresponding to the recommended patients. The diagnosed diseases can include all diseases diagnosed before this time period, i.e. the current diagnosed disease and the previously determined disease. Moreover, the extracted diagnosed diseases must be directly recorded in the medical records of the recommended patients, and cannot be inferred or predicted from the information recorded in the medical records.

[0058] In short, when processing the medical records, the medical model needs to analyze the whole information of the medical records, and perform semantic analysis according to the context in each chapter to determine which diseases mentioned in the medical records are diagnosed and which are to be determined. For example, diseases corresponding to words such as "suspected", "possible" and "not excluded" are usually not diagnosed and are still in the stage of waiting for determination, so they cannot be used as diagnosed diseases. In the embodiments of the present application, the medical model is used to directly extract the directly recorded and clearly diagnosed diseases from the medical records as the basis for subsequent DRG grouping.

[0059] Here, when extracting diseases through the medical model, the generated prompt text can be used to prompt the medical model to extract diseases, and then the generated prompt text can be input into the medical model to make the medical model extract diseases under the prompt of the prompt text and output corresponding results. It should be noted that the generated prompt text usually contains medical records, and in addition to this, it can also include requirements for the disease extraction process, as well as requirements for the content, format, layout, etc. of the output reply. For example, it can include the limitation condition for the disease extraction process "the extracted disease must be clearly extracted in the medical record, and the extracted disease must be consistent with the description in the medical record, and there can be no character changes". For example, it can include the requirement for the output content and format "the output format is [{“disease”:XX,“whether previously diagnosed”:previously diagnosed / this time diagnosed / uncertain}]".

[0060] The medical model is a large model suitable for the medical and medical fields based on a large language model. The medical model can be trained in advance by a large amount of medical and medical field data, so that the medical model can learn specific knowledge and rules in the medical and medical fields, thereby better understanding and sorting the input medical and medical field information in the actual application process, and giving more reasonable and accurate results.

[0061] Further, after obtaining the diagnosed disease of the patient to be recommended extracted by the medical model, in the embodiment of the present application, the main diagnosis can be selected according to the diagnosed disease and the diagnosis disease list corresponding to the medical record obtained in advance, so as to determine the main disease from the diseases suffered by the patient to be recommended. That is, based on the extracted diagnosed disease and the diagnosis disease list, the main diagnosis is selected according to the medical record to obtain the main diagnosis disease of the patient to be recommended.

[0062] Specifically, the prompt text for main diagnosis selection can be generated according to the diagnosed disease, the diagnosis disease list and the medical record, and then the prompt text can be input into the medical model to make the medical model select the main diagnosis based on the prompt text and output the selected main diagnosis disease. Here, the generation of the prompt text can be similar to the generation of the prompt text in the disease extraction process, but the limitation conditions of the model processing process are different. The prompt text generated here can contain requirements for main diagnosis selection, such as "1. Main diagnosis definition: the disease (or health condition) that causes the patient to be hospitalized for treatment and determines the main reason for the patient's current hospitalization. 2. Main diagnosis should be: (1) the most medical resources consumed during this hospitalization. (2) the greatest harm to the patient's health. (3) the longest impact on the length of hospital stay." and requirements for the content, format, layout, etc. of the output reply.

[0063] Moreover, it is worth noting that different types of diseases have different limitation conditions / specifications when selecting the main diagnosis. For example, when the disease includes a tumor, the specification in the prompt text for main diagnosis selection should be different from other cases. The specific main diagnosis selection specification / limitation condition can be set according to the actual disease condition of the patient to be recommended.

[0064] Here, the main diagnosis selection of the patient to be recommended by the medical model not only lays a foundation for subsequent DRG grouping, improves the efficiency and accuracy of DRG grouping recommendation, but also helps to develop reasonable payment standards and optimize medical insurance payment, thereby achieving effective control of medical expenses, and further helping to improve hospital management level, promote medical quality improvement, and improve patient medical security and medical experience.

[0065] After that, the DRG grouping can be performed according to the selected primary diagnosis disease to obtain the DRG grouping recommendation result of the patient to be recommended. That is, the disease-related group grouping recommendation can be performed on the patient to be recommended according to the primary diagnosis disease, so as to obtain the disease-related group suitable for the patient to be recommended, and the disease-related group can be taken as the DRG grouping recommendation result of the patient to be recommended. Here, the process of the DRG grouping is actually to determine the condition of the patient to be recommended, the clinical characteristics of the patient to be recommended, and the like according to the primary diagnosis disease, and to determine the medical resource consumption required by the patient to be recommended according to the condition and the clinical characteristics of the patient to be recommended, so as to group the patient to be recommended on this basis. The purpose of the grouping is to group the patients with similar clinical characteristics and resource consumption into the same group, so as to facilitate the management in the unit of group.

[0066] It should be noted that in the embodiment of the present application, the DRG grouping considers the overall information of the medical record of the patient to be recommended, such as the primary diagnosis disease, the clinical characteristics, the surgical treatment, the disease severity, and the complications, and the like. According to the principle that the patients with similar clinical processes and resource consumption are grouped into the same group, the patient to be recommended is grouped, which not only can be grouped into the appropriate disease diagnosis related group, and the accurate and reliable DRG grouping is realized, but also can provide important data support for the quality evaluation of medical services and the payment of medical insurance fees.

[0067] In the embodiment of the present application, from the perspective of diagnosis and treatment, the diagnosis disease list is optimized and improved through the medical model, and the diagnosis based on insufficient basis can be excluded, so as to facilitate the selection of the primary diagnosis, and finally the DRG grouping recommendation can be performed through the primary diagnosis disease, the accurate DRG grouping is realized, the traditional scheme is more complete, the accuracy is higher, and the adaptability is stronger, the problems existing in the current DRG grouping can be solved, and the controllability and interpretability are very strong. In addition, in the embodiment of the present application, the medical model is introduced for disease extraction and primary diagnosis selection, the problems of difficulty in processing massive medical record information, and difficulty in complying with the principle / limiting condition / specification of primary diagnosis selection existing in the current DRG grouping can be well avoided, the accurate information is sorted out at each link, so as to provide accurate and reliable data support for the DRG grouping.

[0068] The DRG grouping recommendation method based on a large model provided by the application extracts the diseases of the medical record of the patient to be recommended through a medical model to obtain a diagnosed disease, selects a main diagnosis according to the diagnosed disease and the medical record and a diagnosis disease list to obtain a main diagnosis disease, and performs DRG grouping recommendation according to the main diagnosis disease to obtain a DRG grouping recommendation result of the patient to be recommended, thereby overcoming the defects of poor clinical adaptability, easy omission of important information, obvious errors, insufficient accuracy and reliability of the DRG grouping recommendation scheme in the prior art, starting from the perspective of diagnosis and treatment, excluding diagnoses based on insufficient medical models, thereby optimizing and improving the diagnosis disease list, and on this basis, selecting a main diagnosis and performing DRG grouping recommendation according to the medical record, which can realize accurate and reliable DRG grouping, and has strong explainability and adaptability in clinical use, and can meet the actual application requirements.

[0069] Based on the above embodiment, in step 120, the main diagnosis of the patient to be recommended is selected based on the diagnosed disease, the diagnosis disease list and the medical record, including:

[0070] Based on the diagnosed disease and the diagnosis disease list, a target diagnosis disease list is determined;

[0071] Based on the target diagnosis disease list and the medical record, a diagnosis summary corresponding to each diagnosis disease in the target diagnosis disease list is generated;

[0072] Based on each diagnosis disease and the corresponding diagnosis summary, the main diagnosis of the patient to be recommended is selected.

[0073] Specifically, in step 120, the process of selecting the main diagnosis through the medical model can include the following steps:

[0074] After obtaining the diagnosed disease of the patient to be recommended through disease extraction, in the embodiment of the application, to further ensure the accuracy of the extracted diagnosed disease, diagnosis screening can be performed to exclude non-disease interference and duplication with the diagnosis diseases in the existing diagnosis disease list.

[0075] In detail, the diagnosed disease can be screened first to select the diagnosis of the disease and exclude non-disease interference, and then all the diagnosed diseases obtained after disease screening can be compared with all the diagnosis diseases in the diagnosis disease list to merge the same diseases and retain different diseases to avoid duplication, thereby constructing a target diagnosis disease list.

[0076] It should be noted that, since medical record writing / entry is relatively rigorous, the diagnosis given by the doctor is professional, so there is usually no case where the same disease has different professional names in medicine, and even if there is, based on the strong knowledge reserve and understanding ability of the medical model, it can be annotated, so it will not affect the disease screening and comparison and merging here.

[0077] After that, the target diagnosis disease list and the medical record can be used to select the main diagnosis to determine the main diagnosis disease of the patient to be recommended. However, considering that in the process of main diagnosis selection, the medical record information is massive and complex, usually including admission record, course record, discharge record, medical order and other complex chapters, the number of characters in a medical record usually reaches ten thousand or even hundreds of thousands, even if the medical model can process such long text and perform complex task reasoning, it will consume more computing resources and take longer time, and the effect is difficult to guarantee. In view of this, in the embodiment of the application, the information in the medical record can be simplified to extract the key information, which represents the corresponding medical record information, to participate in the main diagnosis selection, so as to avoid the above problems and improve the processing efficiency and effect.

[0078] Specifically, here the target diagnosis disease list and the medical record can be used to generate an abstract to obtain the diagnosis abstract corresponding to each diagnosis disease in the target diagnosis disease list, and then the generated diagnosis abstracts are used to select the main diagnosis to determine the main diagnosis disease of the patient to be recommended from each diagnosis disease. That is, the key information corresponding to each diagnosis disease in the target diagnosis disease list can be extracted from the medical record to obtain the diagnosis abstract corresponding to each diagnosis disease according to the key information; further, the diagnosis abstract can be used to replace the medical record to select the main diagnosis, that is, each diagnosis disease and its corresponding diagnosis abstract are input into the medical model to make the medical model determine the main diagnosis disease of the patient to be recommended from each diagnosis disease according to the diagnosis abstract.

[0079] In the embodiment of the application, the medical model is introduced to process the current main diagnosis selection principle which is difficult to follow, and further to solve the problems existing in long text processing. It is proposed that the diagnosis abstract is used to replace the full medical record to ensure that the medical model can grasp the full medical record information when making a judgment, so as to reduce the computing load while ensuring the processing efficiency and effect. By inputting the diagnosis abstract and the main diagnosis selection principle into the medical model, the main diagnosis disease can be directly output, so that the whole DRG entry recommendation process is more controllable and more interpretable.

[0080] Based on the above embodiment, the target diagnosis disease list and the medical record are used to generate an abstract to obtain the diagnosis abstract corresponding to each diagnosis disease in the target diagnosis disease list, including:

[0081] determining diagnosis knowledge corresponding to each diagnosis disease;

[0082] vectorizing the diagnosis knowledge and the medical record to obtain diagnosis knowledge vectors and medical record segment vectors;

[0083] performing medical record retrieval based on the diagnosis knowledge vectors and the medical record segment vectors to obtain medical record diagnosis segments corresponding to each diagnosis disease in the medical record;

[0084] generating summaries based on the medical record diagnosis segments corresponding to each diagnosis disease to obtain diagnosis summaries corresponding to each diagnosis disease in the target diagnosis disease list.

[0085] Specifically, the process of generating summaries according to the target diagnosis disease list and the medical record to obtain diagnosis summaries corresponding to each diagnosis disease can specifically include:

[0086] Since the diagnosis diseases contained in the target diagnosis disease list only have one disease name, the information is too thin, and it is difficult to extract accurate and complete key information from the medical record based on the key information extraction. Therefore, in the embodiment of the present application, in order to ensure the accuracy and reliability of the entire summary generation process, the diagnosis diseases in the target diagnosis disease list can be first enriched and perfected to obtain rich and diverse disease diagnosis related knowledge, that is, to obtain diagnosis knowledge corresponding to each diagnosis disease. Here, it can be specifically obtained by network searching, medical database retrieval, knowledge graph association and other technologies, to find disease diagnosis related knowledge corresponding to each diagnosis disease, that is, diagnosis knowledge related to each diagnosis disease.

[0087] Figure 2 is a flowchart of the summary generation process provided by the present application, as shown in Figure 2 In the embodiment of the present application, the diagnosis knowledge corresponding to each diagnosis disease is preferably obtained through a pre-constructed disease diagnosis knowledge graph. Further, after obtaining each diagnosis knowledge, the diagnosis knowledge and the medical record can be vectorized to convert them into vectors, thereby obtaining diagnosis knowledge vectors corresponding to each diagnosis disease and medical record segment vectors corresponding to the medical record.

[0088] It should be noted that, for the vectorization processing of the medical record, due to the massive and complex information in the medical record, when the vectorization processing is performed, the medical record can be first divided into chapters, so as to divide the entire medical record into multiple chapters, and then the medical record information in each chapter can be vectorized, that is, the sentences or paragraphs in each chapter are vectorized, so as to obtain the medical record vector of each chapter. Since the entire medical record is divided into multiple chapters, each chapter can be understood as a medical record segment, and thus the medical record vector of each chapter is also called the medical record segment vector corresponding to the medical record. Here, the vectorization processing of the medical record can be completed in advance and does not need to be performed together with the vectorization processing of the diagnosis knowledge.

[0089] After obtaining the diagnosis knowledge vector corresponding to each diagnosis disease and the medical record segment vector corresponding to the medical record, the medical record retrieval can be performed based on this in the embodiment of the application, so as to obtain the key information corresponding to each diagnosis disease in the medical record, that is, the medical record is retrieved by means of vector matching, specifically, each diagnosis knowledge vector is matched with all medical record segment vectors, and if the matching is successful, it means that the two are related, and the corresponding chapter, sentence, paragraph and the like in the chapter are the key information corresponding to the diagnosis disease in the medical record, that is, the medical record diagnosis segment corresponding to the diagnosis disease.

[0090] After that, the diagnosis summary corresponding to each diagnosis disease can be generated based on the medical record diagnosis segment corresponding to each diagnosis disease, specifically, each medical record diagnosis segment extracted is processed to simplify the information, so as to obtain the core essence of the medical record diagnosis, that is, the content of each medical record diagnosis segment corresponding to the diagnosis disease is compressed by the medical model to realize the simplification and essence extraction of the information, so as to obtain the compressed medical record information corresponding to each diagnosis disease, and then the diagnosis summary corresponding to each diagnosis disease can be generated based on the compressed medical record information to generate the diagnosis summary corresponding to each diagnosis disease by the medical model.

[0091] It is worth noting that, in order to avoid the problem of knowledge illusion and make the DRG grouping have a basis, when the summary is generated, the generated summary needs to be the original text sentence in the medical record and cannot be the summary obtained by the model after the information is extracted. That is, the prompt text in the summary generation process can include the limitation condition of the summary generation process, for example, the extracted segment after the summary output must be extracted from the input segment, and the given original text must be used; the segment after the summary output is directly output, and the reason is not explained.

[0092] Based on the above embodiment, the diagnosis summary includes a diagnosis disease summary and a diagnosis treatment summary;

[0093] Based on each diagnosis disease and the diagnosis summary corresponding thereto, the main diagnosis selection is performed to obtain the main diagnosis disease of the patient to be recommended, including:

[0094] Based on each diagnosed disease, and the diagnosis disease summary and diagnosis treatment summary corresponding to each diagnosed disease, diagnosis effectiveness verification and treatment effectiveness verification are performed to obtain a verification result;

[0095] Based on the verification result, a candidate diagnosed disease is determined from each diagnosed disease;

[0096] Based on the candidate diagnosed disease and the diagnosis summary corresponding thereto, a main diagnosis is selected to obtain a main diagnosed disease of the patient to be recommended.

[0097] Specifically, to ensure the effectiveness of the diagnosis and facilitate subsequent DRG grouping, when generating the summary, content limitation can be performed, that is, the generated diagnosis summary includes two parts, namely the content related to disease diagnosis, that is, the diagnosis disease summary, and the content related to disease treatment, that is, the diagnosis treatment summary.

[0098] Based on this, in the embodiment of the present application, when the main diagnosis is selected, the effectiveness verification can be performed according to the diagnosed disease and the diagnosis summary corresponding thereto. Specifically, the diagnosis effectiveness verification can be performed according to each diagnosed disease and the diagnosis disease summary corresponding thereto to verify whether the corresponding diagnosis is effective, whether the patient to be recommended has the diagnosed disease, and the treatment effectiveness verification can be performed according to each diagnosed disease and the diagnosis treatment summary corresponding thereto to verify whether the patient has received targeted treatment for the corresponding diagnosed disease, thereby obtaining a verification result. The verification result can be an overall result, or can include the results of diagnosis effectiveness verification and treatment effectiveness verification, which are not limited in the embodiment of the present application.

[0099] Further, the candidate diagnosed disease can be determined based on the verification result on the basis of the target diagnosed disease list. That is, the candidate diagnosed disease is selected from each diagnosed disease according to the verification result. Specifically, in the case where the verification passes, that is, the diagnosis effectiveness verification of the corresponding diagnosed disease passes, and the treatment effectiveness verification result also passes, it can be understood that the patient to be recommended indeed has the corresponding diagnosed disease and has received targeted treatment, and then the diagnosed disease can be taken as the candidate diagnosed disease. Otherwise, the corresponding diagnosed disease is directly excluded from the target diagnosed disease list. After all the diagnosed diseases are screened, the remaining diagnosed diseases in the target diagnosed disease list are the candidate diagnosed diseases.

[0100] After that, the main diagnosis can be selected according to the candidate diagnosed disease and the diagnosis summary corresponding thereto to obtain the main diagnosed disease of the patient to be recommended. That is, all the candidate diagnosed diseases and the diagnosis summary corresponding thereto are input into the medical model to enable the medical model to determine the main diagnosed disease from each candidate diagnosed disease according to the diagnosis summary.

[0101] In the embodiments of the present application, the validity check is performed according to the diagnosed diseases, the diagnosed disease abstracts in the diagnosis abstracts, and the diagnosed treatment abstracts, and the candidate diagnosed diseases are determined and the main diagnosis is selected according to the check results, so that the selected main diagnosed diseases can all receive sufficient treatment and meet the recommended basic conditions, and the absurd errors are avoided.

[0102] Based on the above embodiments, the check results include the diagnosis validity check results and the treatment validity check results.

[0103] Based on the check results, the candidate diagnosed diseases are determined from the diagnosed diseases, including:

[0104] In the case that the diagnosis validity check result and the treatment validity check result corresponding to any diagnosed disease are both passed, the diagnosed disease is taken as the candidate diagnosed disease.

[0105] Specifically, the check results can include the results obtained by the diagnosis validity check, i.e., the diagnosis validity check results, and the results obtained by the treatment validity check, i.e., the treatment validity check results. Based on this, when determining the candidate diagnosed diseases, if the diagnosis validity check result and the treatment validity check result corresponding to any diagnosed disease are both passed, the diagnosed disease is taken as the candidate diagnosed disease, i.e., after the validity check, it is verified that the patient to be recommended actually has the diagnosed disease and has received targeted treatment, and at this time, the diagnosed disease can be directly taken as the candidate diagnosed disease.

[0106] Correspondingly, if at least one of the diagnosis validity check result and the treatment validity check result corresponding to any diagnosed disease is not passed, the diagnosed disease needs to be excluded from the target diagnosed disease list and cannot be taken as the candidate diagnosed disease.

[0107] Based on the above embodiments, the diagnosis validity check and the treatment validity check are performed based on the diagnosed diseases, the diagnosed disease abstracts corresponding to the diagnosed diseases, and the diagnosed treatment abstracts, and the check results are obtained, and then the following steps are further included:

[0108] From the diagnosed diseases, the diagnosed diseases whose diagnosis validity check results in the check results are passed are selected as the candidate diagnosed diseases.

[0109] Based on the candidate diagnosed diseases and the diagnosed disease list, the missed diagnosis recommendation is performed.

[0110] Specifically, after the diagnosis validity check is performed according to the diagnosed diseases and the diagnosed abstracts corresponding thereto, the missed diagnosis recommendation can also be performed according to the check results. Figure 3 is the flowchart of the missed diagnosis recommendation provided by the present application, as Figure 3As shown, after the diagnostic validity check of each diagnosed disease, if the diagnostic validity check result is that the check passes, the corresponding diagnosed disease can be selected as a to-be-selected diagnosed disease, and compared with the diagnosis disease list corresponding to the medical record to determine whether the to-be-selected diagnosed disease actually suffered by the patient is missed by the doctor or not a missed disease. On the contrary, if the diagnostic validity check result is that the check fails, the corresponding diagnosed disease cannot be selected as a to-be-selected diagnosed disease.

[0111] Then, the determined each to-be-selected diagnosed disease is matched with the diagnosis disease list corresponding to the medical record, and if the matching is successful, it indicates that the corresponding to-be-selected diagnosed disease has been diagnosed by the doctor and is included in the diagnosis disease list, and on the contrary, if the matching fails, it indicates that the doctor has not diagnosed it, which is a missed disease.

[0112] In the embodiment of the present application, the strategy of using diagnosis abstract to verify the diagnostic validity can ensure the accuracy and reasonableness of the recommended missed diagnosis recommendation, and can provide a basis for subsequent doctor diagnosis evaluation and diagnosis process review and verification.

[0113] Based on the above embodiment, step 140 includes:

[0114] Based on the main diagnosed disease, the disease diagnosis related group (DRG) grouping recommendation is performed, and the initial grouping recommendation result of the to-be-recommended patient is determined;

[0115] Based on the diagnosed disease and other diagnosed diseases in the diagnosis disease list, the initial grouping recommendation result is updated and optimized to obtain the DRG grouping recommendation result of the to-be-recommended patient.

[0116] Specifically, the process of performing disease diagnosis related group (DRG) grouping recommendation according to the main diagnosed disease to obtain the DRG grouping recommendation result of the to-be-recommended patient includes:

[0117] Figure 4 The flowchart of the DRG grouping recommendation provided by the present application is shown in FIG. 1. Figure 4 As shown, after disease extraction, abstract generation and main diagnosis selection, when performing DRG grouping, the to-be-recommended patient can be first grouped according to the main diagnosed disease to obtain a preliminary grouping result, i.e. the preliminary grouping recommendation result of the to-be-recommended patient. Here, the main diagnosed disease of the to-be-recommended patient is determined to determine the patient's condition, clinical characteristics, etc., and the medical resource consumption of the to-be-recommended patient is determined accordingly, so that the to-be-recommended patient is roughly grouped on this basis. The purpose of this step of grouping is to determine the main direction so as to facilitate subsequent accurate adjustment, thereby obtaining the preliminary grouping recommendation result.

[0118] Then, the initial group recommendation result can be updated and optimized according to the confirmed disease and other diagnosed diseases in the disease diagnosis list, and the DRG group recommendation result of the patient to be recommended is obtained, that is, whether there is a complication, a serious complication in the target diagnosis disease list except the main diagnosis disease is judged, and the preliminary group recommendation result is adjusted according to this to realize the optimization of grouping, so that the final DRG group recommendation result is obtained.

[0119] In the embodiment of the application, the DRG grouping considers the whole information of the medical record of the patient to be recommended, such as the main diagnosis disease, the clinical characteristics, the surgical treatment, the disease severity and the comorbidity, and the complication, and the patient to be recommended is grouped according to the principle that the patients with similar clinical processes and resource consumption are grouped into the same group, which not only can be grouped into the appropriate disease diagnosis related group to realize the accurate and reliable DRG grouping, but also can provide important data support for the quality evaluation of medical services and the payment of medical insurance fees.

[0120] The DRG group recommendation device based on a large model provided by the application is described below, and the DRG group recommendation device based on a large model described below can be correspondingly referred to the DRG group recommendation method based on a large model described above.

[0121] Figure 5 The structure diagram of the DRG group recommendation device based on a large model provided by the application is shown in FIG. 1, which comprises: Figure 5

[0122] The medical record determination unit 510 is configured to determine the medical record of the patient to be recommended and the diagnosis disease list corresponding to the medical record.

[0123] The main diagnosis selection unit 520 is configured to extract the disease from the medical record based on the medical model to obtain the confirmed disease, and select the main diagnosis based on the confirmed disease, the diagnosis disease list and the medical record to obtain the main diagnosis disease of the patient to be recommended.

[0124] The group recommendation unit 530 is configured to perform DRG group recommendation based on the main diagnosis disease to obtain the DRG group recommendation result of the patient to be recommended.

[0125] ​The application provides a DRG grouping recommendation device based on a large model, which extracts diseases from the medical record of a patient to be recommended through a medical model, obtains a diagnosed disease, selects a main diagnosis according to the diagnosed disease, the medical record and a diagnosis disease list, obtains a main diagnosis disease, and performs DRG grouping recommendation according to the main diagnosis disease to obtain a DRG grouping recommendation result of the patient to be recommended. The device overcomes the defects of poor clinical adaptability, easy omission of important information, obvious errors, insufficient accuracy and reliability of the DRG grouping recommendation scheme in the traditional scheme, starts from the perspective of diagnosis and treatment, excludes diagnoses based on insufficient information by using a medical model, thereby optimizing and improving the diagnosis disease list, and on this basis, the main diagnosis is selected and the DRG grouping is recommended according to the medical record, so that accurate and reliable DRG grouping can be realized, and the device has strong explainability and adaptability in the clinical use process and can meet the actual application requirements.

[0126] Based on the above embodiment, the main diagnosis selection unit 520 is configured to:

[0127] determine a target diagnosis disease list based on the diagnosed disease and the diagnosis disease list;

[0128] generate an abstract based on the target diagnosis disease list and the medical record to obtain a diagnosis abstract corresponding to each diagnosis disease in the target diagnosis disease list;

[0129] select a main diagnosis based on each diagnosis disease and the corresponding diagnosis abstract to obtain a main diagnosis disease of the patient to be recommended.

[0130] Based on the above embodiment, the main diagnosis selection unit 520 is configured to:

[0131] determine diagnosis knowledge corresponding to each diagnosis disease, and perform vectorization processing on the diagnosis knowledge and the medical record to obtain a diagnosis knowledge vector and a medical record segment vector;

[0132] perform medical record retrieval based on the diagnosis knowledge vector and the medical record segment vector to obtain a medical record diagnosis segment corresponding to each diagnosis disease in the medical record;

[0133] generate an abstract based on the medical record diagnosis segment corresponding to each diagnosis disease to obtain a diagnosis abstract corresponding to each diagnosis disease in the target diagnosis disease list.

[0134] Based on the above embodiment, the diagnosis abstract includes a diagnosis disease abstract and a diagnosis treatment abstract.

[0135] The main diagnosis selection unit 520 is configured to:

[0136] Based on the diagnosis diseases, diagnosis disease abstracts corresponding to the diagnosis diseases, and treatment disease abstracts corresponding to the diagnosis diseases, diagnosis validity verification and treatment validity verification are performed to obtain verification results;

[0137] Based on the verification results, candidate diagnosis diseases are determined from the diagnosis diseases;

[0138] Based on the candidate diagnosis diseases and diagnosis abstracts corresponding to the candidate diagnosis diseases, main diagnosis selection is performed to obtain main diagnosis diseases of the patient to be recommended.

[0139] Based on the above embodiment, the verification results include diagnosis validity verification results and treatment validity verification results.

[0140] The main diagnosis selection unit 520 is configured to:

[0141] In a case where the diagnosis validity verification result and the treatment validity verification result corresponding to any diagnosis disease are both verification passed, the diagnosis disease is taken as a candidate diagnosis disease.

[0142] Based on the above embodiment, the device further includes a missed diagnosis recommendation unit configured to:

[0143] From the diagnosis diseases, diagnosis diseases with diagnosis validity verification results passed in the verification results are selected as diagnosis diseases to be selected;

[0144] Based on the diagnosis diseases to be selected and the diagnosis disease list, missed diagnosis recommendation is performed.

[0145] Based on the above embodiment, the enrollment recommendation unit 530 is configured to:

[0146] Based on the main diagnosis diseases, disease diagnosis related group (DRG) enrollment recommendation is performed to determine initial enrollment recommendation results of the patient to be recommended.

[0147] Based on the confirmed diagnosis disease and other diagnosis diseases in the diagnosis disease list, the initial enrollment recommendation results are updated and optimized to obtain DRG enrollment recommendation results of the patient to be recommended.

[0148] Figure 6 An example of an entity structure diagram of an electronic device is shown in FIG. 1. Figure 6As shown, the electronic device can include a processor 610, a communications interface 620, a memory 630, and a communications bus 640, wherein the processor 610, the communications interface 620, and the memory 630 complete mutual communication through the communications bus 640. The processor 610 can invoke a logical instruction in the memory 630 to execute the large model-based DRG group recommendation method, which includes determining a medical record of a patient to be recommended and a diagnosis disease list corresponding to the medical record; performing disease extraction on the medical record based on a medical model to obtain a confirmed diagnosis disease, and performing main diagnosis selection based on the confirmed diagnosis disease, the diagnosis disease list, and the medical record to obtain a main diagnosis disease of the patient to be recommended; and performing disease diagnosis related group (DRG) group recommendation based on the main diagnosis disease to obtain a DRG group recommendation result of the patient to be recommended.

[0149] In addition, the logical instruction in the memory 630 described above can be implemented in the form of a software function unit and sold or used as an independent product, and can be stored in a computer readable storage medium. Based on such understanding, the technical solutions of the present application essentially or the part that contributes to the prior art or part of the technical solutions can be embodied in the form of a software product. The computer software product is stored in a storage medium, and includes a plurality of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the method described in the various embodiments of the present application. The aforementioned storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a magnetic disk or an optical disk, and various media that can store program codes.

[0150] On the other hand, the present application also provides a computer program product, which includes a computer program stored on a non-transitory computer readable storage medium, and the computer program includes program instructions, when the program instructions are executed by a computer, the computer can execute the large model-based DRG group recommendation method provided by the above-mentioned methods, which includes: determining a medical record of a patient to be recommended and a diagnosis disease list corresponding to the medical record; performing disease extraction on the medical record based on a medical model to obtain a confirmed diagnosis disease, and performing main diagnosis selection based on the confirmed diagnosis disease, the diagnosis disease list, and the medical record to obtain a main diagnosis disease of the patient to be recommended; and performing disease diagnosis related group (DRG) group recommendation based on the main diagnosis disease to obtain a DRG group recommendation result of the patient to be recommended.

[0151] In yet another aspect, the present application also provides a non-transitory computer-readable storage medium having stored thereon a computer program, which, when executed by a processor, implements a DRG grouping recommendation method based on a large model, the method comprising: determining a medical record of a patient to be recommended, and a list of diagnosed diseases corresponding to the medical record; performing disease extraction on the medical record based on a medical model to obtain a confirmed disease, and performing primary diagnosis selection based on the confirmed disease, the list of diagnosed diseases and the medical record to obtain a primary diagnosis disease of the patient to be recommended; and performing DRG grouping recommendation based on the primary diagnosis disease to obtain a DRG grouping recommendation result of the patient to be recommended.

[0152] The device embodiments described above are merely illustrative, wherein the units described as separate components can or can not be physically separated, and the components displayed as units can or can not be physical units, i.e., they can be located in one place or distributed on multiple network units. Part or all of the modules can be selected to achieve the purpose of the present embodiment according to actual needs. Those skilled in the art can understand and implement without creative labor.

[0153] From the above description of the embodiments, those skilled in the art can clearly understand that the embodiments can be realized by means of software plus necessary universal hardware platforms, and of course, can also be realized by hardware. Based on such understanding, the above technical solutions, essentially or in other words, the part that contributes to the prior art, can be embodied in the form of a software product, which can be stored in a computer readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes a number of instructions to make a computer device (which can be a personal computer, a server, or a network device, etc.) execute the methods described in each embodiment or some parts of the embodiments.

[0154] Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present application, and not to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that: it can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement to some technical features; and these modifications or replacements do not make the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A large model-based DRG group recommendation method, characterized in that, The method comprises the following steps: determining the medical record of a patient to be recommended and a list of diagnosed diseases corresponding to the medical record; extracting diseases from the medical record based on a medical model to obtain confirmed diseases, and determining a target list of diagnosed diseases based on the confirmed diseases and the list of diagnosed diseases; generating summaries based on the target list of diagnosed diseases and the medical record to obtain a diagnosis summary corresponding to each diagnosed disease in the target list of diagnosed diseases; the diagnosis summary comprises a diagnosis disease summary and a diagnosis treatment summary; performing diagnosis validity verification and treatment validity verification based on each diagnosed disease and the diagnosis disease summary and the diagnosis treatment summary corresponding to each diagnosed disease to obtain verification results; the verification results comprise diagnosis validity verification results and treatment validity verification results; in the case that the diagnosis validity verification result and the treatment validity verification result corresponding to any diagnosed disease are both verified, the any diagnosed disease is taken as a candidate diagnosed disease; performing main diagnosis selection based on the candidate diagnosed disease and the diagnosis summary corresponding thereto to obtain a main diagnosed disease of the patient to be recommended; performing disease diagnosis-related group (DRG) group entry recommendation based on the main diagnosed disease to determine an initial group entry recommendation result of the patient to be recommended; updating and optimizing the initial group entry recommendation result based on the confirmed diseases and other diagnosed diseases in the list of diagnosed diseases to obtain a DRG group entry recommendation result of the patient to be recommended. 2.The large model-based DRG admission recommendation method according to claim 1, characterized in that, The method of generating summaries based on the target list of diagnosed diseases and the medical record to obtain a diagnosis summary corresponding to each diagnosed disease in the target list of diagnosed diseases comprises the following steps: determining diagnosis knowledge corresponding to each diagnosed disease, and performing vectorization processing on the diagnosis knowledge and the medical record to obtain diagnosis knowledge vectors and medical record segment vectors; performing medical record retrieval based on the diagnosis knowledge vectors and the medical record segment vectors to obtain medical record diagnosis segments corresponding to each diagnosed disease in the medical record; generating summaries based on the medical record diagnosis segments corresponding to each diagnosed disease to obtain diagnosis summaries corresponding to each diagnosed disease in the target list of diagnosed diseases. 3.The large model-based DRG admission recommendation method of claim 1, wherein, The method of performing diagnosis validity verification and treatment validity verification based on each diagnosed disease and the diagnosis disease summary and the diagnosis treatment summary corresponding to each diagnosed disease to obtain verification results further comprises the following steps: selecting, from the diagnosed diseases, a diagnosed disease whose diagnosis validity verification result in the verification results is verified as passed as a to-be-selected diagnosed disease; performing missed diagnosis recommendation based on the to-be-selected diagnosed disease and the list of diagnosed diseases. 4.A device for DRG group recommendation based on a large model, characterized in that, The method comprises the following steps: a medical record determining unit configured to determine a medical record of a patient to be recommended and a list of diagnosed diseases corresponding to the medical record; The main diagnosis selection unit is configured to: perform disease extraction on the medical record based on a medical model to obtain a confirmed disease; determine a target diagnosis disease list based on the confirmed disease and the diagnosis disease list; perform summary generation on the target diagnosis disease list and the medical record to obtain a diagnosis summary corresponding to each diagnosis disease in the target diagnosis disease list; the diagnosis summary includes a diagnosis disease summary and a diagnosis treatment summary; perform diagnosis validity verification and treatment validity verification based on the each diagnosis disease and the diagnosis disease summary and the diagnosis treatment summary corresponding to the each diagnosis disease to obtain a verification result; the verification result includes a diagnosis validity verification result and a treatment validity verification result; In a case where the diagnosis validity verification result and the treatment validity verification result corresponding to any diagnosis disease are both verified, the any diagnosis disease is taken as a candidate diagnosis disease; perform main diagnosis selection based on the candidate diagnosis disease and the diagnosis summary corresponding to the candidate diagnosis disease to obtain a main diagnosis disease of the patient to be recommended; The group recommendation unit is configured to perform disease diagnosis related group (DRG) group recommendation based on the main diagnosis disease to determine an initial group recommendation result of the patient to be recommended; Perform update optimization on the initial group recommendation result based on the confirmed disease and other diagnosis diseases in the diagnosis disease list to obtain a DRG group recommendation result of the patient to be recommended.

5. An electronic device comprising a memory, a processor, and a computer program stored on the memory and running on the processor, characterized in that, The computer program is executed by the processor to implement the DRG group recommendation method based on a large model according to any one of claims 1 to 3.

6. A non-transitory computer-readable storage medium having stored thereon a computer program, characterized in that, The computer program is executed by the processor to implement the DRG group recommendation method based on a large model according to any one of claims 1 to 3.

Images