Method and system for automatically generating modification type knowledge base based on text report analysis

Abstract

The invention provides a method for automatically generating a modification type knowledge base based on text report analysis. The method comprises the following steps: acquiring modification type data before and after a preset number of image reports which are manually labeled are modified; defining a database formed by modification type data before and after the preset number of image reports are modified as a modification type basic library; obtaining a single disease instance library; each single disease instance library comprises cue words and answers; and automatically generating a modification type knowledge base according to a preset rule based on the cue word of the large language model, the modification type basic library and the single disease instance library. The invention further discloses a system for automatically generating the modification type knowledge base based on text report analysis. According to the method, a standardized knowledge base can be continuously integrated and expanded for standardizing writing and teaching of reports, and intelligent teaching and performance evaluation can be designed based on report content extraction and type modification.

Description

Technical Field

[0001] The present invention relates to the field of medical information, and more specifically, to a method and system for automatically generating a modification type knowledge base based on text report analysis. Background Art

[0002] In the process of writing imaging reports, the same report will be initially written by junior doctors and reviewed and revised by senior doctors. In order to evaluate the quality of report writing by junior doctors and to fill in the gaps in the knowledge of junior doctors in a targeted manner, it is necessary to first classify the modifications of diagnostic impressions in medical imaging reports. In the existing technology, traditional NLP can recognize many words and grammars in imaging diagnostic reports, and can even perform entity alignment, but there is no clear definition method for the relationship between the recognized words and the report modification type. Even if anatomical and pathological features can be extracted, it cannot be used to define the nature and type of review modifications based on the changes before and after the review of these issues. Summary of the invention

[0003] In view of this, the main purpose of the present invention is to provide a method and system for automatically generating a modification type knowledge base based on text report analysis, which can solve the problem in the prior art that there is no clear correspondence between vocabulary and report modification types.

[0004] To achieve the above object, the technical solution of the present invention is achieved as follows:

[0005] On the one hand, the present invention provides a method for automatically generating a modification type knowledge base based on text report analysis, including: obtaining modification type data of a preset number of manually annotated image reports before and after modification; defining a database formed by the modification type data of the preset number of image reports before and after modification as a modification type basic library; obtaining a single disease example central library; each of the single disease example central libraries includes a prompt word and an answer; based on the prompt word of a large language model, the modification type basic library, and the single disease example central library, automatically generating a modification type knowledge base according to preset rules.

[0006] Preferably, the preset rules include single rule judgment classification data and multiple rule judgment classification data.

[0007] Preferably, the single rule judgment classification data includes: consistency: the content of the imaging report is consistent before and after modification; diagnostic logic error: the importance ranking of disease types is incorrect, only the order of sentences is modified, and other contents are not modified; text grammar and format errors; inconsistent diagnostic confidence: the text of possibility appears before and after modification, or the expression of possibility is inconsistent; inconsistent language habits: the writing habits before and after modification are inconsistent, and synonyms are used; degree judgment error: the description of the disease degree before and after modification is inconsistent; the disease degree includes: mild, slight, slightly, severe, serious, and significant; clinical communication problems: clinical questions are not answered, normal before modification, and normal or disease is denied after modification; clinical communication problems: prompts for clinical work-contrast: only contrast and comparison appear before or after modification; clinical communication problems: prompts for clinical work-suggestions: only suggestions appear before or after modification.

[0008] Preferably, the multi-rule judgment classification data includes: judgment criteria and priority confirmation, wherein the judgment criteria include: missed diagnosis, misdiagnosis, positioning error, unrecognized features, and feature recognition errors; priority confirmation includes the following steps: confirmation of positioning error and missed diagnosis or misdiagnosis: first determine the type of disease before and after modification, and then determine the anatomical location. If the disease type is the same and the anatomical location is close, the modification type is positioning error; if the disease type is the same and the anatomical location is far, the modification type is missed diagnosis or misdiagnosis; positioning error and unrecognized features or feature recognition errors: first determine whether there are image features before and after modification. If there are no image features and only the position is inconsistent, the modification type is positioning error; if there are image features and the image features are inconsistent, the modification type is unrecognized features or feature recognition errors; unrecognized features or feature recognition errors and missed diagnosis or misdiagnosis: confirm whether the inconsistent content before and after modification is the disease type or the image feature. If the disease type is inconsistent, the modification type is missed diagnosis or misdiagnosis; if the image features are inconsistent, the modification type is unrecognized features or feature recognition errors.

[0009] Preferably, the method also includes: automatically generating teaching materials, the specific steps are as follows: obtaining all pre-modification report contents and post-modification report contents of the same doctor within a preset time period; based on the modification type knowledge base, automatically determining the modification type of each report, judging the complexity, completeness and clarity of the report content according to the modification type, and outputting analysis results: comparing the analysis results with the doctor's learning requirements, and automatically generating teaching materials for the doctor.

[0010] Preferably, after obtaining all the pre-modification report contents and post-modification report contents of the same doctor within a preset time period, the method also includes: obtaining the number of control clicks, the distance the control mouse moves, the number of times the text edit box is filled in, the number of characters filled in the text edit box, the number of manual edits of the imaging diagnosis, the time taken to write the report, and the time taken to review the report; calculating the above data to generate the analysis results; comparing the analysis results with the performance requirements, and automatically outputting the performance results of the doctor within the preset time period.

[0011] On the other hand, the present invention also provides a system for automatically generating a modification type knowledge base based on text report analysis, the system comprising: a first acquisition module, a second acquisition module and a modification type knowledge base module, wherein the first acquisition module is connected to the modification type knowledge base module, and is used to obtain the modification type data of a preset number of manually annotated image reports before and after modification; the database formed by the modification type data before and after modification of the preset number of image reports is defined as a modification type basic library; the second acquisition module is connected to the modification type knowledge base module, and is used to obtain a single disease example library; each of the single disease example libraries contains a prompt word and an answer; the modification type knowledge base module is respectively connected to the first acquisition module and the second acquisition module, and is used to automatically generate a modification type knowledge base according to preset rules based on the prompt word of a large language model, the modification type basic library, and the single disease example library.

[0012] Preferably, the preset rules include single-rule judgment classification data and multi-rule judgment classification data, wherein the single-rule judgment classification data includes: consistency; diagnostic logic error; text grammar and format error; inconsistent diagnostic confidence; inconsistent language habits; degree judgment error; clinical communication problem: clinical question not answered, normal before modification, denied normal or denied disease after modification; clinical communication problem: prompts for clinical work-comparison: only comparison and comparison appear before or after modification; clinical communication problem: prompts for clinical work-suggestions: only suggestions appear before or after modification; the multi-rule judgment classification data includes: judgment criteria and priority confirmation: judgment criteria include: missed diagnosis, misdiagnosis, positioning error, feature not recognized, feature recognition error; priority confirmation steps are as follows: Confirmation of positioning error and missed diagnosis or misdiagnosis: first determine the type of disease before and after modification, and then determine the anatomical location. If the disease type is the same and the anatomical location is close, the modification type is positioning error; if the disease type is the same and the anatomical location is far, the modification type is missed diagnosis or misdiagnosis; Positioning error and unrecognized features or incorrect feature recognition: first determine whether there are imaging features before and after modification. If there are no imaging features and only the position is inconsistent, the modification type is positioning error; if there are imaging features and the imaging features are inconsistent, the modification type is unrecognized features or incorrect feature recognition; Unrecognized features or incorrect feature recognition and missed diagnosis or misdiagnosis: confirm whether the inconsistent content before and after modification is the disease type or the imaging feature. If the disease type is inconsistent, the modification type is missed diagnosis or misdiagnosis; if the imaging features are inconsistent, the modification type is unrecognized features or incorrect feature recognition.

[0013] Preferably, the system also includes: an analysis module and a teaching material generation module, wherein the analysis module is connected to the modification type knowledge base module, and is used to analyze all pre-modification report contents and post-modification report contents of the same doctor within a preset time period, and automatically determine the modification type of each report based on the modification type knowledge base, judge the complexity, completeness and clarity of the report content according to the modification type, and output the analysis results; the teaching material generation module is connected to the analysis module, and is used to compare the analysis results with the doctor's learning requirements and automatically generate the doctor's teaching materials.

[0014] Technical effects of the present invention:

[0015] The method of the present invention defines a basic knowledge base of modification types and a basic knowledge base of single disease examples. Based on the basic knowledge base of modification types and the basic knowledge base of single disease examples, a large language model is trained to obtain batches of image feature data, and a modification type knowledge base is automatically generated. The standardized knowledge base can be continuously integrated and expanded for standardized report writing and teaching, and intelligent teaching and performance evaluation can be designed based on report content extraction and modification type. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] The drawings described herein are used to provide a further understanding of the present invention and constitute a part of this application. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings:

[0017] Figure 1 A flow chart of a method for automatically generating a modification type knowledge base based on text report analysis according to Embodiment 1 of the present invention is shown;

[0018] Figure 2 A schematic diagram of the system structure for automatically generating a modification type knowledge base based on text report analysis according to Embodiment 2 of the present invention is shown;

[0019] Figure 3 A schematic diagram of the system structure for automatically generating a modification type knowledge base based on text report analysis according to Embodiment 3 of the present invention is shown. DETAILED DESCRIPTION

[0020] The present invention will be described in detail below with reference to the accompanying drawings and in combination with embodiments.

[0021] Embodiment 1

[0022] Figure 1 FIG. 4 shows a flow chart of a method for automatically generating a modification type knowledge base based on text report analysis according to Embodiment 1 of the present invention; Figure 1 As shown, the method comprises the following steps:

[0023] Step S101, obtaining the modification type data of the preset number of image reports before and after modification that are manually annotated; defining a database formed by the modification type data of the preset number of image reports before and after modification as a modification type basic database;

[0024] Among them, the manually annotated modification type data of the imaging report before and after modification is a set of label libraries. Based on the text content of the comparison report modification, the modification type can be judged to establish a general knowledge base and a disease-specific personalized knowledge base. Each knowledge base entry is a combination of a set of label definitions and label property definitions. These labels are manually extracted and standardized.

[0025] Step S102, obtaining a single disease example library; each of the single disease example libraries includes a prompt word and an answer;

[0026] Step S103, based on the prompt words of the large language model, the modification type basic library, and the single disease example library, a modification type knowledge base is automatically generated according to preset rules;

[0027] The preset rules include single rule judgment classification data and multiple rule judgment classification data.

[0028] The single rule for determining the classification data includes:

[0029] Consistent: The contents of the imaging report are consistent before and after modification;

[0030] Diagnostic logic errors: the importance of disease types is not ranked correctly, only the order of sentences is changed, and other contents are not changed;

[0031] The text contains grammatical and formatting errors;

[0032] Inconsistent diagnostic confidence: The text of the possibility before and after the modification, or the expression of the possibility is inconsistent;

[0033] Inconsistent language habits: The writing habits before and after the revision are inconsistent, and synonyms are used;

[0034] Error in judging the severity: The description of the severity of the disease before and after the revision is inconsistent; the severity of the disease includes: mild, slight, slightly, severe, serious, and significant;

[0035] Clinical communication issues: clinical questions were not answered, normal before modification, and denied normality or disease after modification;

[0036] Clinical communication issues: Tips for clinical work - Contrast: Contrast and comparison only appear before or after modification;

[0037] Clinical communication issues: Tips and suggestions for clinical work: Suggestions only appear before or after modification.

[0038] The multi-rule judgment classification data includes: judgment criteria and priority confirmation, wherein:

[0039] The judgment criteria include: missed diagnosis, misdiagnosis, incorrect positioning, unrecognized features, and incorrect feature recognition;

[0040] The priority is confirmed as follows:

[0041] Confirmation of positioning error and missed or misdiagnosis: first determine the type of disease before and after the modification, then determine the anatomical location. If the disease type is the same and the anatomical location is close, the modification type is positioning error; if the disease type is the same and the anatomical location is far, the modification type is missed or misdiagnosis;

[0042] Positioning error and feature not recognized or feature recognition error: First determine whether there are image features before and after the modification. If there are no image features and only the position is inconsistent, the modification type is positioning error; if there are image features and the image features are inconsistent, the modification type is feature not recognized or feature recognition error;

[0043] Unrecognized features or incorrectly identified features vs. missed or misdiagnosed features: confirm whether the inconsistent content before and after the modification is the disease type or the imaging feature. If the disease type is inconsistent, the modification type is missed or misdiagnosed; if the imaging features are inconsistent, the modification type is unrecognized features or incorrectly identified features.

[0044] Among them, the method also includes: automatically generating teaching materials, the specific steps are as follows: obtaining all pre-modification report contents and post-modification report contents of the same doctor within a preset time period; based on the modification type knowledge base, automatically determining the modification type of each report, judging the complexity, completeness and clarity of the report content according to the modification type, and outputting analysis results: comparing the analysis results with the doctor's learning requirements, and automatically generating teaching materials for the doctor.

[0045] Among them, after obtaining all the report contents before and after modification of the same doctor within a preset time period, the method also includes: obtaining the number of control clicks, the distance the control mouse moves, the number of times the text edit box is filled in, the number of characters filled in the text edit box, the number of manual edits of the imaging diagnosis, the time taken to write the report, and the time taken to review the report; calculating the above data to generate the analysis results; comparing the analysis results with the performance requirements, and automatically outputting the performance results of the doctor within the preset time period.

[0046] The method further includes: using a large language model to extract various information from medical books and generate a medical knowledge base, and the extracted content includes:

[0047] Disease type: The standard name of a disease type in medical books;

[0048] Disease description: Glossary of disease types in medical books;

[0049] Report template: Generate report template and image feature vocabulary for the current disease type based on the current disease type and historical report text analysis;

[0050] In the summary of the report writing process, the words that need to be searched in the knowledge base are selected, and the relevant content is viewed in the knowledge base module, which provides convenience for the reporting doctor.

[0051] The embodiments of the present invention define a basic knowledge base of modification types and a basic knowledge base of single disease examples. Based on the basic knowledge base of modification types and the basic knowledge base of single disease examples, a large language model is trained to obtain batches of image feature data, and a modification type knowledge base is automatically generated. The standardized knowledge base can be continuously integrated and expanded for standardizing report writing and teaching, and intelligent teaching and performance evaluation can be designed based on report content extraction and modification type.

[0052] Embodiment 2

[0053] Figure 2 FIG. 4 shows a schematic diagram of a system structure for automatically generating a modification type knowledge base based on text report analysis according to Embodiment 2 of the present invention; Figure 2 As shown, the system includes: a first acquisition module 10, a second acquisition module 20 and a modification type knowledge base module 30, wherein:

[0054] The first acquisition module 10 is connected to the modification type knowledge base module 30, and is used to obtain the modification type data of the preset number of image reports before and after the modification; define the database formed by the modification type data of the preset number of image reports before and after the modification as the modification type basic library;

[0055] Among them, the manually annotated modification type data of the imaging report before and after modification is a set of label libraries. Based on the text content of the comparison report modification, the modification type can be judged to establish a general knowledge base and a disease-specific personalized knowledge base. Each knowledge base entry is a combination of a set of label definitions and label property definitions. These labels are manually extracted and standardized.

[0056] The second acquisition module 20 is connected to the modification type knowledge base module 30, and is used to obtain a library of single disease examples; each of the single disease example libraries includes a prompt word and an answer;

[0057] The modification type knowledge base module 30 is connected to the first acquisition module 10 and the second acquisition module 20 respectively, and is used to automatically generate a modification type knowledge base according to preset rules based on the prompt words of the large language model, the modification type basic library, and the single disease example central library.

[0058] The preset rules include single rule judgment classification data and multiple rule judgment classification data.

[0059] The single rule for determining the classification data includes:

[0060] Consistent: The contents of the imaging report are consistent before and after modification;

[0061] Diagnostic logic errors: the importance of disease types is not ranked correctly, only the order of sentences is changed, and other contents are not changed;

[0062] The text contains grammatical and formatting errors;

[0063] Inconsistent diagnostic confidence: The text of the possibility before and after the modification, or the expression of the possibility is inconsistent;

[0064] Inconsistent language habits: The writing habits before and after the revision are inconsistent, and synonyms are used;

[0065] Error in judging the severity: The description of the severity of the disease before and after the revision is inconsistent; the severity of the disease includes: mild, slight, slightly, severe, serious, and significant;

[0066] Clinical communication issues: clinical questions were not answered, normal before modification, and denied normality or disease after modification;

[0067] Clinical communication issues: Tips for clinical work - Contrast: Contrast and comparison only appear before or after modification;

[0068] Clinical communication issues: Tips and suggestions for clinical work: Suggestions only appear before or after modification.

[0069] The multi-rule judgment classification data includes: judgment criteria and priority confirmation, wherein:

[0070] The judgment criteria include: missed diagnosis, misdiagnosis, incorrect positioning, unrecognized features, and incorrect feature recognition;

[0071] The priority is confirmed as follows:

[0072] Confirmation of positioning error and missed or misdiagnosis: first determine the type of disease before and after the modification, then determine the anatomical location. If the disease type is the same and the anatomical location is close, the modification type is positioning error; if the disease type is the same and the anatomical location is far, the modification type is missed or misdiagnosis;

[0073] Positioning error and feature not recognized or feature recognition error: First determine whether there are image features before and after the modification. If there are no image features and only the position is inconsistent, the modification type is positioning error; if there are image features and the image features are inconsistent, the modification type is feature not recognized or feature recognition error;

[0074] Unrecognized features or incorrectly identified features vs. missed or misdiagnosed features: confirm whether the inconsistent content before and after the modification is the disease type or the imaging feature. If the disease type is inconsistent, the modification type is missed or misdiagnosed; if the imaging features are inconsistent, the modification type is unrecognized features or incorrectly identified features.

[0075] Embodiment 3

[0076] Figure 3 FIG. 4 shows a schematic diagram of a system structure for automatically generating a modification type knowledge base based on text report analysis according to Embodiment 3 of the present invention; Figure 3 As shown, the system also includes: an analysis module 40 and a teaching material generation module 50, wherein:

[0077] The analysis module 40 is connected to the modification type knowledge base module 30, and is used to analyze all pre-modification and post-modification report contents of the same doctor within a preset time period, automatically determine the modification type of each report based on the modification type knowledge base, judge the complexity, completeness and clarity of the report content according to the modification type, and output the analysis result;

[0078] The teaching material generating module 50 is connected to the analyzing module 40 and is used to compare the analysis result with the doctor's learning requirements and automatically generate the teaching material for the doctor.

[0079] Among them, after obtaining all the report contents before and after modification of the same doctor within a preset time period, the analysis module 40 is also used to: obtain the number of control clicks, the distance the control mouse moves, the number of times the text edit box is filled in, the number of characters filled in the text edit box, the number of manual edits of the imaging diagnosis, the time taken to write the report, and the time taken to review the report; calculate the above data to generate the analysis results; compare the analysis results with the performance requirements, and automatically output the performance results of the doctor within the preset time period.

[0080] The system also includes: using a large language model to extract various information from medical books and generate a medical knowledge base, the extracted content includes:

[0081] Disease type: The standard name of a disease type in medical books;

[0082] Disease description: Glossary of disease types in medical books;

[0083] Report template: Generate report template and image feature vocabulary for the current disease type based on the current disease type and historical report text analysis;

[0084] In the summary of the report writing process, the words that need to be searched in the knowledge base are selected, and the relevant content is viewed in the knowledge base module, which provides convenience for the reporting doctor.

[0085] The embodiments of the present invention define a basic knowledge base of modification types and a basic knowledge base of single disease examples. Based on the basic knowledge base of modification types and the basic knowledge base of single disease examples, a large language model is trained to obtain batches of image feature data, and a modification type knowledge base is automatically generated. The standardized knowledge base can be continuously integrated and expanded for standardizing report writing and teaching, and intelligent teaching and performance evaluation can be designed based on report content extraction and modification type.

[0086] From the above description, it can be seen that the above-mentioned embodiments of the present invention achieve the following technical effects: the embodiments of the present invention define a basic knowledge base of modification types and a basic knowledge base of single disease examples, and based on the basic knowledge base of modification types and the basic knowledge base of single disease examples, train a large language model to obtain batches of image feature data, and automatically generate a modification type knowledge base. The standardized knowledge base can be continuously integrated and expanded for standardized report writing and teaching, and can be used to design intelligent teaching and performance evaluation based on report content extraction and modification type.

[0087] Obviously, those skilled in the art should understand that the above modules or steps of the present invention can be implemented by a general computing device, they can be concentrated on a single computing device, or distributed on a network composed of multiple computing devices, and optionally, they can be implemented by a program code executable by a computing device, so that they can be stored in a storage device and executed by the computing device, or they can be made into individual integrated circuit modules, or multiple modules or steps therein can be made into a single integrated circuit module for implementation. Thus, the present invention is not limited to any specific combination of hardware and software.

[0088] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for automatically generating a modification type knowledge base based on text report analysis, characterized in that: include: Obtain modification type data of a preset number of manually annotated image reports before and after modification; A database formed by the modification type data of the preset number of image reports before and after modification is defined as a modification type basic database; Obtain a single disease example database; each of the single disease example databases includes a prompt word and an answer; Based on the prompt words of the large language model, the modification type basic library, and the single disease example central library, a modification type knowledge base is automatically generated according to preset rules.

2. The method for automatically generating a modification type knowledge base based on text report analysis according to claim 1, characterized in that: The preset rules include single rule judgment classification data and multiple rule judgment classification data.

3. The method for automatically generating a modification type knowledge base based on text report analysis according to claim 2, characterized in that: The single rule judgment classification data includes: Consistent: The contents of the imaging report are consistent before and after modification; Diagnostic logic errors: the importance of disease types is not ranked correctly, only the order of sentences is changed, and other contents are not changed; The text contains grammatical and formatting errors; Inconsistent diagnostic confidence: The text of the possibility before and after the modification, or the expression of the possibility is inconsistent; Inconsistent language habits: The writing habits before and after the revision are inconsistent, and synonyms are used; Error in judging the severity: The description of the severity of the disease before and after the revision is inconsistent; the severity of the disease includes: mild, slight, slightly, severe, serious, and significant; Clinical communication issues: clinical questions were not answered, normal before modification, and denied normality or disease after modification; Clinical communication issues: Tips for clinical work - Contrast: Contrast and comparison only appear before or after modification; Clinical communication issues: Tips and suggestions for clinical work: Suggestions only appear before or after modification.

4. The method for automatically generating a modification type knowledge base based on text report analysis according to claim 2, characterized in that: The multi-rule judgment classification data includes: judgment criteria and priority confirmation, wherein: The judgment criteria include: missed diagnosis, misdiagnosis, incorrect positioning, unrecognized features, and incorrect feature recognition; The priority is confirmed as follows: Confirmation of positioning error and missed or misdiagnosis: first determine the type of disease before and after the modification, then determine the anatomical location. If the disease type is the same and the anatomical location is close, the modification type is positioning error; if the disease type is the same and the anatomical location is far, the modification type is missed or misdiagnosis; Positioning error and feature not recognized or feature recognition error: First determine whether there are image features before and after the modification. If there are no image features and only the position is inconsistent, the modification type is positioning error; if there are image features and the image features are inconsistent, the modification type is feature not recognized or feature recognition error; Unrecognized features or incorrectly identified features vs. missed or misdiagnosed features: confirm whether the inconsistent content before and after the modification is the disease type or the imaging feature. If the disease type is inconsistent, the modification type is missed or misdiagnosed; if the imaging features are inconsistent, the modification type is unrecognized features or incorrectly identified features.

5. The method for automatically generating a modification type knowledge base based on text report analysis according to claim 1, characterized in that: The method also includes automatically generating teaching materials, and the specific steps are as follows: Obtain all pre-modification and post-modification report contents of the same doctor within a preset time period; Based on the modification type knowledge base, the modification type of each report is automatically determined, the complexity, completeness and clarity of the report content are judged according to the modification type, and the analysis results are output: The analysis results are compared with the doctor's learning requirements, and teaching materials for the doctor are automatically generated.

6. The method for automatically generating a modification type knowledge base based on text report analysis according to claim 5, characterized in that: After obtaining all pre-modification report contents and post-modification report contents of the same doctor within a preset time period, the method further includes: Obtain the number of control clicks, the distance the control mouse moves, the number of text edit box entries, the number of characters filled in the text edit box, the number of manual edits of the imaging diagnosis, the time taken to write the report, and the time taken to review the report; Calculate the above data to generate the analysis result; The analysis results are compared with the performance requirements, and the performance results of the doctor within a preset time period are automatically output.

7. A system for automatically generating a modification type knowledge base based on text report analysis, characterized in that: The system includes: a first acquisition module, a second acquisition module and a modification type knowledge base module, wherein: The first acquisition module is connected to the modification type knowledge base module, and is used to obtain the modification type data of the preset number of image reports before and after the modification; define the database formed by the modification type data of the preset number of image reports before and after the modification as the modification type basic library; The second acquisition module is connected to the modification type knowledge base module, and is used to obtain a library of single disease examples; each of the single disease example libraries includes a prompt word and an answer; The modification type knowledge base module is connected to the first acquisition module and the second acquisition module respectively, and is used to automatically generate a modification type knowledge base according to preset rules based on the prompt words of the large language model, the modification type basic library, and the single disease example central library.

8. The system for automatically generating a modification type knowledge base based on text report analysis according to claim 7, characterized in that: The preset rules include single rule judgment classification data and multi-rule judgment classification data, wherein: The single rule judgment classification data includes: consistency; diagnostic logic errors; text grammar and format errors; inconsistent diagnostic confidence; inconsistent language habits; degree judgment errors; clinical communication problems: clinical questions were not answered, normal before modification, and normal or disease was denied after modification; clinical communication problems: prompts for clinical work-contrast: only contrast and comparison appear before or after modification; clinical communication problems: prompts for clinical work-suggestions: only suggestions appear before or after modification; The multi-rule judgment classification data includes: judgment criteria and priority confirmation: The judgment criteria include: missed diagnosis, misdiagnosis, incorrect positioning, unrecognized features, and incorrect feature recognition; The priority is confirmed as follows: Confirmation of positioning error and missed or misdiagnosis: first determine the type of disease before and after the modification, then determine the anatomical location. If the disease type is the same and the anatomical location is close, the modification type is positioning error; if the disease type is the same and the anatomical location is far, the modification type is missed or misdiagnosis; Positioning error and feature not recognized or feature recognition error: First determine whether there are image features before and after the modification. If there are no image features and only the position is inconsistent, the modification type is positioning error; if there are image features and the image features are inconsistent, the modification type is feature not recognized or feature recognition error; Unrecognized features or incorrectly identified features vs. missed or misdiagnosed features: confirm whether the inconsistent content before and after the modification is the disease type or the imaging feature. If the disease type is inconsistent, the modification type is missed or misdiagnosed; if the imaging features are inconsistent, the modification type is unrecognized features or incorrectly identified features.

9. The system for automatically generating a modification type knowledge base based on text report analysis according to claim 7, characterized in that: The system also includes: an analysis module and a teaching material generation module, wherein: The analysis module is connected to the modification type knowledge base module, and is used to analyze all pre-modification and post-modification report contents of the same doctor within a preset time period, automatically determine the modification type of each report based on the modification type knowledge base, judge the complexity, completeness and clarity of the report content according to the modification type, and output the analysis results; The teaching material generating module is connected to the analyzing module and is used for comparing the analysis result with the learning requirements of the doctor and automatically generating the teaching material for the doctor.

Images