Complaint text generation method, system and equipment based on large language model and medium

Abstract

The invention provides an appeal text generation method, system and device based on a large language model and a medium. The method comprises the steps of obtaining medical data needing to be processed currently; processing the medical data needing to be processed currently to obtain corresponding text description information as current medical record text information; generating a corresponding prompt word based on the current medical record text information and current reference text information corresponding to the current medical data needing to be processed; inputting the current medical record text information and the cue word into a large language model to obtain a corresponding analysis result as a current processing result; and verifying the current processing result to determine whether the current processing result is a usable result, and if the current processing result is determined to be a usable result, generating a corresponding appeal text and an intermediate record text by using the large language model. According to the invention, the generation efficiency and accuracy of the appeal text can be improved.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular to a method, system, device and medium for generating complaint text based on a large language model. Background Art

[0002] In some application scenarios, data analysis is required to determine whether the associated content meets the specified conditions. If so, a corresponding complaint document is generated. For example, in medical scenarios, complaint materials must be submitted to file a medical insurance appeal. Currently, manual analysis of data and the generation of corresponding text content are often performed, which is inefficient and inaccurate. Summary of the Invention

[0003] In view of the above technical problems, the technical solution adopted by the present invention is:

[0004] According to a first aspect of the present invention, a method for generating a complaint text based on a large language model is provided, the method comprising the following steps:

[0005] S100, obtaining medical data that currently needs to be processed, wherein the medical data includes case data and cost data associated with the case data.

[0006] S200: Process the medical data that currently needs to be processed to obtain corresponding text description information as current medical record text information.

[0007] S300, generating corresponding prompt words based on current medical record text information and current reference text information corresponding to the medical data currently to be processed; the current reference text information includes at least one case description information allowing for appeal.

[0008] S400: Input the current medical record text information and the prompt word into a large language model to obtain a corresponding analysis result as the current processing result.

[0009] S500, verifying the current processing result to determine whether the current processing result is a usable result. If it is determined that the current processing result is a usable result, using the large language model to generate a corresponding complaint text and an intermediate record text, wherein the intermediate record text includes process description information of the large language model obtaining the current processing result and description information representing that the similarity between the current medical record text information and at least one case description information in the current reference text information is greater than a set similarity threshold.

[0010] According to a second aspect of the present invention, a complaint text generation system based on a large language model is provided, the system comprising:

[0011] The data acquisition module is used to acquire medical data that currently needs to be processed, wherein the medical data includes case data and cost data associated with the case data.

[0012] The data processing module is used to process the medical data that currently needs to be processed to obtain corresponding text description information as the current medical record text information.

[0013] The prompt word generation module is used to generate corresponding prompt words based on the current medical record text information and the current reference text information corresponding to the medical data currently to be processed; the current reference text information includes at least one case description information that allows for appeal.

[0014] The analysis module is used to input the current medical record text information and the prompt word into the large language model to obtain a corresponding analysis result as the current processing result.

[0015] A verification module is configured to verify the current processing result to determine whether the current processing result is a usable result. If the current processing result is determined to be a usable result, the large language model is used to generate a corresponding complaint text and an intermediate record text. The intermediate record text includes a description of the process by which the large language model obtains the current processing result, and description information indicating that the similarity between the current medical record text information and at least one case description in the current reference text information is greater than a set similarity threshold. According to a third aspect of the present invention, an electronic device is provided, comprising a processor and a memory; the processor is configured to execute the steps of the method described in the first aspect of the present invention by calling a program or instruction stored in the memory.

[0016] According to a fourth aspect of the present invention, there is provided a computer-readable storage medium storing a program or instructions, wherein the program or instructions enable a computer to execute the steps of the method according to the first aspect of the present invention.

[0017] The present invention has at least the following beneficial effects:

[0018] The method for generating a complaint text based on a large language model provided by an embodiment of the present invention can realize the automatic generation of the declaration text due to the use of the large language model technology, and can improve the generation efficiency and accuracy of the complaint text.

[0019] It should be understood that the content described in this section is not intended to identify the key or important features of the embodiments of the present invention, nor is it intended to limit the scope of the present invention. Other features of the present invention will become readily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0021] Figure 1 A flowchart of a method for generating complaint text based on a large language model provided in an embodiment of the present invention;

[0022] Figure 2 This is a structural block diagram of the complaint text generation system based on a large language model provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0023] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only 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 those skilled in the art without making any creative efforts shall fall within the scope of protection of the present invention.

[0024] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art to which this invention pertains. The terms used herein in the specification of the present invention are for the purpose of describing specific embodiments only and are not intended to limit the present invention. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0025] It should be noted that some exemplary embodiments are described as processes or methods depicted as flowcharts. Although the flowcharts describe the steps as sequential processes, many of the steps can be performed in parallel, concurrently, or simultaneously. In addition, the order of the steps can be rearranged. A process can be terminated when its operation is completed, but can also have additional steps not included in the accompanying drawings. A process can correspond to a method, function, procedure, subroutine, subprogram, etc.

[0026] The embodiment of the present invention provides a method for generating complaint text based on a large language model, such as Figure 1 As shown, the method includes the following steps:

[0027] S100, obtaining medical data that currently needs to be processed, wherein the medical data includes case data and cost data associated with the case data.

[0028] In an embodiment of the present invention, relevant case data and associated cost data can be obtained from an authorized electronic medical record system through an API interface or database connection. Case data may include all recorded information during the patient's medical treatment process, such as diagnosis results, treatment plans, surgical records, etc.

[0029] In an embodiment of the present invention, the medical data that currently needs to be processed may be medical data that is determined by relevant departments to be non-compliant with preset regulations, for example, medical data that is determined by the Medical Insurance Bureau to be non-compliant with medical insurance rules.

[0030] S200: Process the medical data that currently needs to be processed to obtain corresponding text description information as current medical record text information.

[0031] In an embodiment of the present invention, case data and cost data may be processed separately to generate text descriptions of the case data and cost data, for example, what surgery was performed at a certain time, what medicine was prescribed at a certain time and how much money was charged, etc.

[0032] S300, generating corresponding prompt words based on current medical record text information and current reference text information corresponding to the medical data currently to be processed; the current reference text information includes at least one case description information allowing for appeal.

[0033] In a specific embodiment of the present invention, the reference text information may be the medical insurance rules, and the case description information may be the case conditions for applying for special exceptions. For example, the case description information may include: the number of hospitalizations exceeds 5 times (inclusive) the average hospitalization days of the disease group (DRG) / disease type (DIP) of the same level designated medical institutions in the first half of the year (each pooling area may appropriately reduce the multiple according to the specific circumstances); the number of days the intensive care unit bed is used exceeds 60% (inclusive) of the total number of days the hospitalization bed is used for the case.

[0034] In the embodiment of the present invention, the prompt word may be determined based on actual conditions.

[0035] In an exemplary embodiment, S300 may specifically include:

[0036] S301 , obtaining the similarity between the current medical record text information and each case description information in the current reference text information, and obtaining at least one similarity.

[0037] S302: If there is a similarity greater than a set similarity threshold among the at least one similarity, the similarity greater than the set similarity threshold is used as a candidate similarity.

[0038] In the embodiment of the present invention, the similarity threshold may be set as an empirical value.

[0039] Those skilled in the art know that if there is no similarity greater than a set similarity threshold among the at least one similarity, a corresponding prompt word can be generated directly based on the current medical record text information and the current reference text information corresponding to the medical data currently to be processed.

[0040] S303: Obtain the score corresponding to each candidate similarity, and obtain a weighted sum of each candidate similarity and the corresponding score to obtain at least one weighted sum.

[0041] In an embodiment of the present invention, the score corresponding to the candidate similarity can be determined based on the medical insurance reimbursement expense corresponding to the case description information corresponding to the candidate similarity, and can specifically be positively correlated with the medical insurance reimbursement expense, that is, the greater the medical insurance reimbursement expense, the greater the corresponding score.

[0042] In an embodiment of the present invention, the weight of similarity and the weight of score can be determined based on actual conditions. In one exemplary embodiment, the weight of similarity and the weight of score can be the same. In another exemplary embodiment, the weight of similarity can be less than the weight of score.

[0043] S304: Taking the case description information corresponding to the maximum weighted sum in the at least one weighted sum as information of interest, and generating prompt word prompt information based on the information of interest.

[0044] In the embodiment of the present invention, the prompt word is used to guide the reasoning of the large language model. The prompt word prompt information can be determined based on the actual situation, for example, it can be "Please prioritize matching information of interest".

[0045] S305, if an input instruction for generating a prompt word based on the prompt word prompt information is received, the prompt word is generated based on the current medical record text information, the current reference text information and the prompt word prompt information; if no input instruction for generating a prompt word based on the prompt word prompt information is received, the prompt word is generated based on the current medical record text information and the current reference text information.

[0046] In an embodiment of the present invention, if the prompt word is generated based on the current medical record text information, the current reference text information and the prompt word prompt information, it can save processing time and make the matching results more accurate compared to the method of generating the prompt word based only on the current medical record text information and the current reference text information.

[0047] In an exemplary embodiment of the present invention, an example of generating the prompt word based on the current medical record text information and the current reference text information may be as follows:

[0048]

[0049]

[0050] Example:

[0051] Assume the following rules:

[0052] Serial number 16: The same-day charging code contains exactly 003107010030000-310701003 and the number of charges on that day is greater than 1.

[0053] Rule 21: The same-day expense codes exactly match 003315010520000-331501052 and the expense codes exactly match 003302040010000-330204001.

[0054] Assume that the medical record contains the following:

[0055] Patient Yang XX, female, 27 years old, was admitted to the hospital on November 17, 2024, and discharged on November 23, 2024, with an actual hospital stay of 6 days. Discharged from the Department of Gynecology, Medical Insurance Settlement Level: Level 3, Diagnosis Codes and Names: N73.003: Acute Female Pelvic Inflammatory Disease, B37.301: Candidal Vaginitis, Z39.100: Lactation Supervision, A09.02.07.03: Pelvic Inflammatory Disease. On December 31, 2024, 003107010030000-310701003 was used twice. On December 30, 2024, 003315010520000-331501052 was used.

[0056] Output the results in Chinese:

[0057]

[0058]

[0059] S400: Input the current medical record text information and the prompt word into a large language model to obtain a corresponding analysis result as the current processing result.

[0060] In the embodiment of the present invention, the large language model may be an existing open source large language model.

[0061] S500: Verify the current processing result to determine whether the current processing result is a usable result. If it is determined that the current processing result is a usable result, use the large language model to generate a corresponding complaint text and intermediate record text.

[0062] In an embodiment of the present invention, the intermediate record text includes process description information of the large language model obtaining the current processing result and description information representing that the similarity between the current medical record text information and at least one case description information in the current reference text information is greater than a set similarity threshold.

[0063] Those skilled in the art will appreciate that a large language model can automatically generate intermediate record text based on the analysis results. The similarity can be semantic-based similarity, feature-based similarity, etc., preferably semantic similarity.

[0064] Furthermore, in an exemplary embodiment, in S500, the checking of the current processing result to determine whether the current processing result is a usable result specifically includes:

[0065] S501, if the current processing result is a result that represents that the similarity between the current medical record text information and at least one case description information in the current reference text information is greater than the set similarity threshold, it means that the expenses to be reimbursed for the medical data currently to be processed are in compliance with the expense reimbursement rules specified in the current reference text information, and execute S502.

[0066] S502 , determining a reference value of complaint fees corresponding to the medical data currently requiring processing based on the occurrence time of the medical data currently requiring processing and the number of occurrences of the fee-related object in the geographical area corresponding to the medical data currently requiring processing.

[0067] In an embodiment of the present invention, the geographical area may be an area divided according to an administrative unit. The expense-related object may be a disease requiring treatment recorded in the medical data. The reference value of the appeal expense may be a reimbursable medical insurance reimbursement expense.

[0068] It is known to those skilled in the art that each case description information corresponds to a reference value for complaint fees.

[0069] S503, compare the cost information in the current processing result with the complaint cost reference value corresponding to the medical data currently to be processed. If the comparison result indicates that the absolute value of the difference between the cost information in the current processing result and the complaint cost reference value is less than the set value, then determine that the current processing result is a usable result; otherwise, determine that the current processing result is not a usable result.

[0070] In an embodiment of the present invention, the set value may be determined based on actual conditions, for example, an empirical value. If the comparison result indicates that the absolute value of the difference between the cost information in the current processing result and the reference value of the complaint cost is less than the set value, it indicates that the cost reimbursement situation in the medical data currently being processed complies with the reimbursement situation in the geographical area and is relatively reasonable.

[0071] In an embodiment of the present invention, the appeal fee reference value is determined based on a preset function expression, and the preset function expression is determined based on the product of the first cost influencing factor and the first association weight and the product of the second cost influencing factor and the second association weight, wherein the first cost influencing factor is the time interval between the occurrence time of the medical data and the effective start time of the corresponding latest reference text information, and the second cost influencing factor is the number of occurrences of the cost-related object corresponding to the medical data in the geographical area corresponding to the medical data.

[0072] In one exemplary embodiment, the preset function expression can be expressed as: y = f(r1, r2), where y is the reference value of the complaint fee, r1 is the first fee influencing factor, r2 is the second fee influencing factor, and f() is the preset function expression. f() can be set based on actual needs. In one exemplary embodiment, f() can be a multi-linear function of r1 and r2, preferably a linear function of r1 and r2, that is, y = k1 × r1 + k2 × r2 + a. k1 is the first association weight, k2 is the second association weight, and a is a constant.

[0073] Furthermore, in an embodiment of the present invention, the first association weight and the second association weight in the preset function expression are obtained by the following steps:

[0074] S10, obtaining a historical medical data set within the geographical area corresponding to the medical data currently to be processed, wherein the historical medical data set includes m historical data, each historical medical data includes the corresponding complaint fee, the time interval between the occurrence time of the corresponding medical data and the effective start time of the corresponding latest reference text information, and the number of occurrences of the fee-related object corresponding to the medical data in the said geographical area; m>1.

[0075] In the embodiment of the present invention, the appeal fee refers to the medical insurance reimbursement fee. It is known to those skilled in the art that each case description information corresponds to one appeal fee.

[0076] S11, based on the historical medical data set and the preset function expression, obtain the associated weight combination set C = (C1, C2, ..., C i ,……,C m-1 ), C i is the i-th association weight combination in C, C i =(C ir1 , C ir2 ), C ir1 is the first association weight of the i-th item, C ir2 is the i-th second association weight, where i ranges from 1 to m-1.

[0077] In the embodiment of the present invention, each piece of historical data can be represented by a function expression, that is, an equation. In this way, a set of associated weight combinations can be obtained by solving m sets of equations.

[0078] S12, draw a square with a length of L1 along the x-axis and a length of L2 along the y-axis in the rectangular coordinate system; wherein L1=(C r1 max -C r1 min ), C r1 max is the maximum time interval association weight in C, C r1 min is the minimum time interval association weight in C, C r1 max =max((C ir1 ) i＝1……m-1 ), C r1 min =min((C ir1 ) i＝1……m-1 ), L2=(C r2 max -C r2 min ), C r2 max is the maximum occurrence association weight in C, C r2 min is the minimum occurrence association weight in C, C r2 max =max((C ir2 ) i＝1……m-1 ), C r2 min =min((C ir2 ) i＝1……m-1 ); the coordinates of the upper left corner of the square are (C r1 min , C r2 min ).

[0079] In the embodiment of the present invention, the scale unit of the x-axis in the rectangular coordinate system may be days, and the scale unit of the y-axis may be times.

[0080] S13, obtain min((C jr1 -C (j-1)r1 )) j＝2……m-1 As the length of the cell in the x-axis direction and get min((C jr2 -C (j-1)r2 )) j＝2……m-1 As the length of the cell in the y-axis direction; the value of j ranges from 2 to m-1.

[0081] S14, using the cells to divide the square into n cells.

[0082] In the embodiment of the present invention, since L1 may not be min((C jr1 -C (j-1)r1 )) j＝2……m-1 Integer multiples of L2 may not be min((C jr2 -C (j-1)r2 )) j＝2……m-1 The total area of n units can be greater than or equal to the area of the square.

[0083] S15, for the rth cell among n cells, traverse C, if C ir1 Belongs to the rth cell, C ir1 Add to the counting set corresponding to the r-th cell. The initial value of the counting set corresponding to the r-th cell is empty. The value of r ranges from 1 to n, and the initial value is 1.

[0084] S16, taking the cell corresponding to the counting set with the most association weight combinations among the n counting sets as the target cell.

[0085] In the embodiment of the present invention, if the number of association weights in a certain counting set is the largest, it means that more association weights are located in the cells corresponding to the counting set, which can be used as representative association weights.

[0086] S17: Obtain the first association weight and the festival association weight in the preset function expression based on the target cell.

[0087] Furthermore, in an exemplary embodiment, the first association weight is the x-coordinate value of the center coordinate of the target cell, and the second association weight is the y-coordinate value of the center coordinate of the target cell.

[0088] Furthermore, in another exemplary embodiment, the first association weight is the x-coordinate value of the lower right corner coordinate of the target cell, and the second association weight is the y-coordinate value of the lower right corner coordinate of the target cell.

[0089] Furthermore, in another embodiment of the present invention, in S500, the checking of the current processing result to determine whether the current processing result is a usable result specifically includes:

[0090] S510 , if the current processing result indicates that the similarity between the current medical record text information and at least one case description information in the current reference text information is greater than a set similarity threshold, execute S511 .

[0091] S511, determining a reference value of a complaint fee corresponding to the medical data currently requiring processing based on the occurrence time of the medical data currently requiring processing and the number of occurrences of the fee-related object in the geographical area corresponding to the medical data currently requiring processing;

[0092] S512, compare the cost information in the current processing result with the complaint cost reference value corresponding to the medical data currently to be processed. If the comparison result indicates that the absolute value of the difference between the cost information in the current processing result and the complaint cost reference value is less than the set value, then determine that the current processing result is a usable result, otherwise, execute S513.

[0093] S513, determine whether the effective start time corresponding to the current reference text information is earlier than the effective end time of the previous reference text information. If the effective start time corresponding to the current reference text information is earlier than the effective end time of the previous reference text information, and the score of the analysis result corresponding to the current reference text information is greater than the score of the analysis result corresponding to the previous reference text information, determine that the current processing result is a usable result.

[0094] In an embodiment of the present invention, the score of the analysis result is determined based on the medical insurance reimbursement expenses corresponding to the corresponding case description information, and may be positively correlated with the medical insurance reimbursement expenses, that is, the greater the medical insurance reimbursement expenses, the greater the corresponding score.

[0095] Compared with the previous embodiment, this embodiment further determines whether the current processing result is a usable result based on the effective start time corresponding to the current reference text information and the effective end time of the previous reference text information when the comparison result indicates that the absolute value of the difference between the fee information in the current processing result and the complaint fee reference value is greater than or equal to the set value, thereby making the obtained processing result more accurate and improving the accuracy of complaint text generation.

[0096] Furthermore, in the embodiment of the present invention, if the effective start time corresponding to the current reference text information is later than the effective end time of the previous reference text information, it indicates that the used reference text information is accurate, and the current processing result is determined to be not a usable result.

[0097] Furthermore, in an embodiment of the present invention, if the effective start time corresponding to the current reference text information is earlier than the effective end time of the previous reference text information, and the score of the analysis result corresponding to the current reference text information is greater than the score of the analysis result corresponding to the previous reference text information, determining that the current processing result is a usable result may specifically include:

[0098] If the effective start time corresponding to the current reference text information is earlier than the effective end time of the previous reference text information, instructing the large language model to perform analysis based on the current medical record text information and the previous reference text information to obtain a corresponding current processing result as a reference result;

[0099] If the score of the analysis result corresponding to the current reference text information is greater than the score of the analysis result corresponding to the previous reference text information, it is determined that the current processing result is a usable result.

[0100] Furthermore, S513 further includes the following steps:

[0101] If the score corresponding to the current processing result is less than the score corresponding to the reference result, the current processing result is determined to be a usable result, and the large language model is instructed to generate the complaint text and intermediate record text based on the reference result.

[0102] In another embodiment of the present invention, for target medical data that is determined by relevant departments to meet preset regulations, the following operations may be performed:

[0103] Determine whether the effective start time corresponding to the reference text information used by the target medical data is earlier than the effective end time of the previous reference text information. If the effective start time corresponding to the reference text information used is earlier than the effective end time of the previous reference text information, and the score of the analysis result corresponding to the reference text information used is less than the score of the analysis result corresponding to the previous reference text information, then generate a corresponding complaint text based on the analysis result corresponding to the previous reference text information.

[0104] Those skilled in the art know that if the effective start time corresponding to the reference text information used is earlier than the effective end time of the previous reference text information, and the score of the analysis result corresponding to the reference text information used is greater than the score of the analysis result corresponding to the previous reference text information, no processing will be performed.

[0105] In the embodiment of the present invention, since the above operations are performed on the target medical data that is determined by the relevant departments to comply with the preset regulations, the processing of the medical data can be made more accurate.

[0106] Based on the same inventive concept, the embodiment of the present invention provides a complaint text generation system based on a large language model, such as Figure 2 As shown, the system may include:

[0107] The data acquisition module 1 is used to acquire medical data that currently needs to be processed, wherein the medical data includes case data and cost data associated with the case data.

[0108] The data processing module 2 is used to process the medical data that currently needs to be processed to obtain corresponding text description information as the current medical record text information.

[0109] The prompt word generation module 3 is used to generate corresponding prompt words based on the current medical record text information and the current reference text information corresponding to the medical data currently to be processed; the current reference text information includes at least one case description information that allows for appeal.

[0110] The analysis module 4 is used to input the current medical record text information and the prompt word into the large language model to obtain the corresponding analysis result as the current processing result.

[0111] Verification module 5 is used to verify the current processing result to determine whether the current processing result is a usable result. If the current processing result is determined to be a usable result, the corresponding complaint text and intermediate record text are generated using the large language model. The intermediate record text includes process description information of the large language model obtaining the current processing result and description information representing that the similarity between the current medical record text information and at least one case description information in the current reference text information is greater than a set similarity threshold.

[0112] The system can be used to perform Figure 1 Therefore, for the functions that can be realized by each functional module of the system, please refer to Figure 1 The description of the illustrated embodiment is omitted for brevity.

[0113] An embodiment of the present invention also provides an electronic device, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are configured to execute the method described in the embodiment of the present invention.

[0114] An embodiment of the present invention further provides a computer-readable storage medium storing computer-executable instructions, wherein the computer instructions are used to execute the method described in the embodiment of the present invention.

[0115] It should be understood that the various forms of the processes shown above can be used to reorder, add, or delete steps. For example, the steps described in the present invention can be performed in parallel, sequentially, or in a different order, as long as the desired results of the technical solutions disclosed in the present invention can be achieved. This is not limited herein.

[0116] The above specific embodiments do not limit the scope of protection of the present invention. Those skilled in the art will appreciate that various modifications, combinations, sub-combinations, and substitutions may be made based on design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention are intended to be included within the scope of protection of the present invention.

Claims

1. A method for generating complaint text based on a large language model, characterized in that: The method comprises the following steps: S100, obtaining medical data that currently needs to be processed, wherein the medical data includes case data and cost data associated with the case data; S200, processing the medical data currently to be processed to obtain corresponding text description information as current medical record text information; S300, generating corresponding prompt words based on current medical record text information and current reference text information corresponding to the medical data currently to be processed; the current reference text information includes at least one case description information allowing for appeal; S400, inputting the current medical record text information and the prompt word into a large language model to obtain corresponding analysis results as current processing results; S500, verifying the current processing result to determine whether the current processing result is a usable result. If it is determined that the current processing result is a usable result, using the large language model to generate a corresponding complaint text and an intermediate record text, wherein the intermediate record text includes process description information of the large language model obtaining the current processing result and description information representing that the similarity between the current medical record text information and at least one case description information in the current reference text information is greater than a set similarity threshold.

2. The method according to claim 1, characterized in that In S500, the checking of the current processing result to determine whether the current processing result is a usable result specifically includes: S501, if the current processing result is a result indicating that the similarity between the current medical record text information and at least one case description information in the current reference text information is greater than a set similarity threshold, executing S502; S502, determining a reference value of a complaint fee corresponding to the medical data currently requiring processing based on the occurrence time of the medical data currently requiring processing and the number of occurrences of the fee-related object in the geographical area corresponding to the medical data currently requiring processing; S503, compare the cost information in the current processing result with the complaint cost reference value corresponding to the medical data currently to be processed. If the comparison result indicates that the absolute value of the difference between the cost information in the current processing result and the complaint cost reference value is less than the set value, then determine that the current processing result is a usable result; otherwise, determine that the current processing result is not a usable result.

3. The method according to claim 1, characterized in that In S500, the checking of the current processing result to determine whether the current processing result is a usable result specifically includes: S510, if the current processing result is a result indicating that the similarity between the current medical record text information and at least one case description information in the current reference text information is greater than a set similarity threshold, executing S511; S511, determining a reference value of a complaint fee corresponding to the medical data currently requiring processing based on the occurrence time of the medical data currently requiring processing and the number of occurrences of the fee-related object in the geographical area corresponding to the medical data currently requiring processing; S512, comparing the cost information in the current processing result with the appeal cost reference value corresponding to the medical data currently requiring processing; if the comparison result indicates that the absolute value of the difference between the cost information in the current processing result and the appeal cost reference value is less than a set value, determining that the current processing result is an acceptable result; otherwise, executing S513; S513, determine whether the effective start time corresponding to the current reference text information is earlier than the effective end time of the previous reference text information. If the effective start time corresponding to the current reference text information is earlier than the effective end time of the previous reference text information, and the score of the analysis result corresponding to the current reference text information is greater than the score of the analysis result corresponding to the previous reference text information, determine that the current processing result is a usable result.

4. The method according to claim 2 or 3, characterized in that The appeal fee reference value is determined based on a preset function expression, which is determined based on the product of a first fee influencing factor and a first association weight, and the product of a second fee influencing factor and a second association weight, wherein the first fee influencing factor is the time interval between the occurrence time of the medical data and the effective start time of the corresponding latest reference text information, and the second fee influencing factor is the number of occurrences of the fee-associated object corresponding to the medical data in the geographical area corresponding to the medical data; The first association weight and the second association weight in the preset function expression are obtained by the following steps: S10, obtaining a historical medical data set within a geographical area corresponding to the medical data currently to be processed, wherein the historical medical data set includes m historical data, each historical medical data set including a corresponding complaint fee, a time interval between a time when the corresponding medical data occurs and a time when the corresponding latest reference text information takes effect, and a number of occurrences of a fee-related object corresponding to the medical data in the geographical area; m>1; S11, based on the historical medical data set and the preset function expression, obtain the associated weight combination set C = (C1, C2, ..., C i ,……,C m-1 ), C i is the i-th association weight combination in C, C i =(C ir1 , C ir2 ), C ir1 is the first association weight of the i-th item, C ir2 is the second association weight of the i-th order, where i ranges from 1 to m-1; S12, draw a square with a length of L1 along the x-axis and a length of L2 along the y-axis in the rectangular coordinate system; wherein L1=(C r1 max -C r1 min ), C r1 max is the maximum time interval association weight in C, C r1 min is the minimum time interval association weight in C, C r1 max =max((C ir1 ) i＝1……m-1 ), C r1 min =min((C ir1 ) i＝1……m-1 ), L2=(C r2 max -C r2 min ), C r2 max is the maximum occurrence association weight in C, C r2 min is the minimum occurrence association weight in C, C r2 max =max((C ir2 ) i＝1……m-1 ), C r2 min =min((C ir2 ) i＝1……m-1 ); the coordinates of the upper left corner of the square are (C r1 min , C r2 min ); S13, obtain min((C jr1 -C (j-1)r1 )) j＝2……m-1 As the length of the cell in the x-axis direction and get min((C jr2 -C (j-1)r2 )) j＝2……m-1 As the length of the cell in the y-axis direction; j ranges from 2 to m-1; S14, using the cells to divide the square into n cells; S15, for the rth cell among n cells, traverse C, if C ir1 Belongs to the rth cell, C ir1 Add to the counting set corresponding to the r-th cell. The initial value of the counting set corresponding to the r-th cell is empty. The value of r ranges from 1 to n, and the initial value is 1. S16, taking the cell corresponding to the counting set with the largest combination of associated weights among the n counting sets as the target cell; S17: Obtain the first association weight and the festival association weight in the preset function expression based on the target cell.

5. The method according to claim 4, characterized in that The first association weight is the x-coordinate value of the center coordinate of the target cell, and the second association weight is the y-coordinate value of the center coordinate of the target cell.

6. The method according to claim 4, characterized in that The first association weight is the x-coordinate value of the lower right corner coordinate of the target cell, and the second association weight is the y-coordinate value of the lower right corner coordinate of the target cell.

7. The method according to claim 1, characterized in that S300 specifically includes: S301, obtaining the similarity between the current medical record text information and each case description information in the current reference text information, and obtaining at least one similarity; S302: If there is a similarity greater than a set similarity threshold among the at least one similarity, the similarity greater than the set similarity threshold is used as a candidate similarity; S303, obtaining a score corresponding to each candidate similarity, and obtaining a weighted sum of each candidate similarity and the corresponding score to obtain at least one weighted sum; S304, taking the case description information corresponding to the maximum weighted sum in the at least one weighted sum as information of interest, and generating prompt word prompt information based on the information of interest; S305, if an input instruction for generating a prompt word based on the prompt word prompt information is received, the prompt word is generated based on the current medical record text information, the current reference text information and the prompt word prompt information; if no input instruction for generating a prompt word based on the prompt word prompt information is received, the prompt word is generated based on the current medical record text information and the current reference text information.

8. A complaint text generation system based on a large language model, characterized by: The system comprises: A data acquisition module is used to acquire medical data that currently needs to be processed, wherein the medical data includes case data and cost data associated with the case data; The data processing module is used to process the medical data that needs to be processed to obtain the corresponding text description information as the current medical record text information; A prompt word generation module, configured to generate corresponding prompt words based on current medical record text information and current reference text information corresponding to the medical data currently to be processed; the current reference text information includes at least one case description information allowing for appeal; An analysis module, configured to input the current medical record text information and the prompt word into a large language model to obtain a corresponding analysis result as a current processing result; A verification module is used to verify the current processing result to determine whether the current processing result is a usable result. If the current processing result is determined to be a usable result, the corresponding complaint text and intermediate record text are generated using the large language model. The intermediate record text includes process description information of the large language model obtaining the current processing result and description information representing that the similarity between the current medical record text information and at least one case description information in the current reference text information is greater than a set similarity threshold.

9. An electronic device, characterized in that: including processor and memory; The processor is configured to execute the steps of the method according to any one of claims 1 to 7 by calling the program or instructions stored in the memory.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium is used to store a program or instruction, and the program or instruction enables a computer to execute the steps of the method according to any one of claims 1 to 7.

Images