An image diagnosis distribution method and system based on a large language model

Abstract

The application discloses an image diagnosis distribution method and system based on a large language model, relates to the fields of medical informatization and artificial intelligence application, and comprises the following steps: setting corresponding prompt words according to the treatment data of a patient, wherein the treatment data of the patient at least comprises an image examination item name, an application form and an electronic medical record; identifying and matching the image examination item of the patient with preset various special disease group classifications based on the prompt words through a large language model to obtain a target special disease group; and distributing the image examination report writing task of the patient to the shared perspective of the target special disease group; and under the premise of setting the special disease group in the imaging department, analyzing the content of the application form and the electronic medical record by using the LLM technology, determining the classification of the special disease group of the patient in advance, and greatly improving the diagnosis quality and efficiency of the department.

Description

Technical Field

[0001] This invention relates to the field of medical informatics and artificial intelligence applications, and more specifically, to an image diagnosis allocation method and system based on a large language model. Background Technology

[0002] Currently, the allocation of diagnostic reports in the radiology department is proactively based on the type and number of examinations, or by placing patients in a public pool for doctors to freely compete for. The goal of this approach is to ensure a basic balance between the workload and difficulty of diagnostic doctors, and to avoid favoritism.

[0003] However, diagnosing physicians have professional biases. They are efficient and make fewer mistakes when diagnosing cases within their own area of ​​expertise. For cases outside their area of ​​expertise, although they need to continuously learn to avoid having any significant weaknesses, their efficiency is reduced due to the different levels of knowledge in different specialties. In fact, the quality of their diagnosis is likely to be lower than that of physicians with strong expertise in specific diseases within the same department. If diagnostic tasks are allocated and managed solely based on the type and quantity of examinations, the match between the diagnostic type and the physician's professional bias is only around 30%.

[0004] Therefore, this application is hereby submitted. Summary of the Invention

[0005] The purpose of this invention is to provide an image diagnosis allocation method and system based on a large language model. Under the premise of setting up disease groups in the radiology department, the method uses LLM technology to analyze the contents of the application form and electronic medical record to determine the classification of the patient's disease group in advance, thereby significantly improving the diagnostic quality and efficiency of the department.

[0006] The above-mentioned technical objective of the present invention is achieved through the following technical solution: This application provides an image diagnosis assignment method based on a large language model, including the following specific steps: Based on the patient's medical records, set corresponding prompts. The patient's medical records should include at least the name of the imaging examination, the application form, and the electronic medical record. Based on prompt words, a large language model is used to identify and match the patient's imaging examination items with pre-defined disease groups to obtain the target disease group; The task of writing the patient's imaging examination report was assigned to the shared perspective of the target disease group.

[0007] Based on the above technical solution, the present invention can be further improved as follows.

[0008] Furthermore, based on the patient's medical records, corresponding prompts are set, including: When the patient's medical records contain the name of an imaging examination item, the system matches the name of the imaging examination item with a pre-set table corresponding to imaging examinations and disease group classifications. When a patient's medical record is a request form and the request form shows that a certain area needs to be examined, the text of the request form and the disease group type included in the area to be examined will be set as prompt words; when a patient's medical record is a request form and the request form shows that a certain area needs to be followed up, the core diagnostic conclusion of the patient's last imaging examination report and the disease group type included in the scanned area will be set as prompt words. When a patient's medical records are electronic medical records, set the entire text of the patient's first medical record and the disease group type included in the scanned area as prompt words.

[0009] Furthermore, the above method also includes: Monitor the image examination report writing tasks of patients in the shared perspective of each disease group. When the number of image examination report writing tasks of patients exceeds the preset value, the image examination report writing tasks of new patients will be allocated to the public resource pool.

[0010] Furthermore, the aforementioned target disease groups were obtained through the following methods: ; In the formula, To obtain the target disease group, For the patient feature vector, the first dimension, Indicates the disease control group The eigenvector of the first dimension, For similarity function, Indicates the weight.

[0011] Furthermore, the above methods also include: When an error occurs in the identification and matching of the target disease group, keywords are retrieved again through a large language model and / or specified by the assigned attending physician until the correct target disease group is obtained.

[0012] Secondly, this application provides an image diagnostic allocation system based on a large language model, applied to any of the image diagnostic allocation methods based on a large language model in the first aspect, comprising: The feature extraction module is used to set corresponding prompts based on the patient's medical records, which include at least the names of imaging examination items, application forms, and electronic medical records. The disease group matching module is used to identify and match the patient's imaging examination items with the pre-defined disease group categories based on prompt words and a large language model to obtain the target disease group; The doctor assignment module is used to assign the task of writing the patient's imaging examination report to the shared perspective of the target disease group.

[0013] Furthermore, based on the patient's medical records, corresponding prompts are set, including: When the patient's medical records contain the name of an imaging examination item, the system matches the name of the imaging examination item with a pre-set table corresponding to imaging examinations and disease group classifications. When a patient's medical record is a request form and the request form shows that a certain area needs to be examined, the text of the request form and the disease group type included in the area to be examined will be set as prompt words; when a patient's medical record is a request form and the request form shows that a certain area needs to be followed up, the core diagnostic conclusion of the patient's last imaging examination report and the disease group type included in the scanned area will be set as prompt words. When a patient's medical records are electronic medical records, set the entire text of the patient's first medical record and the disease group type included in the scanned area as prompt words.

[0014] Furthermore, the aforementioned system also includes: The monitoring module is used to monitor the image examination report writing tasks of patients in the shared perspective of each disease group. When the number of image examination report writing tasks of patients exceeds the preset value, the image examination report writing tasks of new patients will be allocated to the public resource pool.

[0015] Thirdly, this application provides an electronic device, including: at least one processor, at least one memory, and a data bus; In this system, the processor and memory communicate with each other via a data bus; the memory stores program instructions that can be executed by the processor, and the processor calls the program instructions to execute an image diagnosis allocation method based on a large language model, as described in any of the first aspects.

[0016] Fourthly, this application provides a non-transitory computer-readable storage medium that stores computer instructions that cause a computer to execute an image diagnostic allocation method based on a large language model, as described in any of the first aspects.

[0017] Compared with the prior art, the present invention has at least the following beneficial effects: In this application, a large language model (LLM) is used to optimize the medical imaging diagnostic process. By looking up examination items, analyzing application forms, electronic medical records, and other information, the professional field and complexity of the diagnosis are matched with the doctor's capabilities, so as to achieve precise allocation and dynamic adjustment of diagnostic tasks. This can effectively improve diagnostic efficiency and quality. Even without the use of other complex quality control methods, it can produce a significant improvement in diagnostic quality and has the potential to be promoted in medical institutions at all levels. Attached Figure Description

[0018] The accompanying drawings, which are included to provide a further understanding of embodiments of the invention and form part of this application, do not constitute a limitation thereof. In the drawings: Figure 1 This is a flowchart of the image diagnosis allocation method based on a large language model in an embodiment of the present invention; Figure 2 This is a schematic diagram illustrating the definition of a disease-specific group in an embodiment of the present invention; Figure 3 This is a schematic diagram illustrating the establishment of a disease-specific group in an embodiment of the present invention; Figure 4 This is a schematic diagram illustrating the configuration of LLM in an embodiment of the present invention; Figure 5 This is a schematic diagram illustrating the creation of new variables when analyzing application forms in an embodiment of the present invention; Figure 6 This is a schematic diagram illustrating the configuration for storing procedure extraction in an embodiment of the present invention; Figure 7 This is a schematic diagram illustrating the creation of new prompt words when analyzing an application form in an embodiment of the present invention; Figure 8 This is a schematic diagram of the format sent to the LLM model in an embodiment of the present invention; Figure 9 This is a schematic diagram of the LLM model returned in an embodiment of the present invention; Figure 10 This is a schematic diagram illustrating the addition of variables when combining electronic medical record analysis in an embodiment of the present invention; Figure 11 This is a schematic diagram illustrating the configuration for extracting the storage procedure of the first medical record in an embodiment of the present invention; Figure 12 This is a schematic diagram illustrating the creation of new prompt words when combining electronic medical record analysis in an embodiment of the present invention; Figure 13 This is a schematic diagram illustrating the configuration for triggering the execution of a stored procedure during patient registration in an embodiment of the present invention; Figure 14 This is a schematic diagram illustrating the configuration of the disease reporting perspective in an embodiment of the present invention; Figure 15 This is a schematic diagram illustrating the setup of a disease-specific group in an embodiment of the present invention; Figure 16 This is a schematic diagram illustrating the setup of a disease-specific group in an embodiment of the present invention. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0020] Therefore, the following detailed description of the embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention.

[0021] It should be noted that similar labels and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.

[0022] In the description of the embodiments of the present invention, "multiple" means at least two.

[0023] Example 1: To pre-determine patient disease group classifications and significantly improve the diagnostic quality and efficiency of departments, this example provides an image diagnosis allocation method based on a large-scale language model, such as... Figure 1 As shown, the specific steps include the following: S1. Set corresponding prompts based on the patient's medical records. The patient's medical records should include at least the name of the imaging examination, the application form, and the electronic medical record.

[0024] When the medical records are examination items, the classification of the disease group can be directly identified by recognizing the examination name. For these items, a corresponding table can be used directly, and the table can be consulted. For example, abdominal X-ray (KUB) and lumbar MRI can directly identify the urology disease group and the spinal joint disease group, respectively.

[0025] When the medical records are application forms, LLM analysis is used to determine the disease group. It is not possible to directly lock the disease group by the examination items. However, if the examination items include equipment type and site information, the range of disease groups that can be involved can be narrowed down. The text of the application form can be used as a prompt word, and the type of disease group included in that site can be used as a common prompt word. The prompt words are extracted by LLM and directly classified after matching.

[0026] If the extracted prompt is "stroke to be investigated", the specific disease areas can include: 1. Psychiatric imaging; 2. Brain tumors; 3. Cerebrovascular diseases; 4. Leukoencephalopathy; 5. Ophthalmological diseases; 6. Otological diseases; 7. Nasopharyngeal carcinoma and other head tumors; then the LLM large model matching result will be "cerebrovascular diseases".

[0027] When the application form contains "xx follow-up", the patient's previous report of the same type of imaging examination can be used as a reference, and LLM analysis and matching can be performed twice. The first time, the text of the previous report is sent to LLM to request it to extract the main core diagnostic conclusions. The second time, the core diagnostic conclusions and the disease group types that the examination item may involve are used as prompt words, and the disease group classification is given after processing by LLM.

[0028] When the medical records are electronic medical records, since electronic medical records include many chapters and the time of generation and the department / person submitting them are different, LLM can be used to identify and process the "initial medical record" and the set of names of the disease groups involved in this examination, thereby obtaining the disease groups involved, and can also provide the possible disease groups involved.

[0029] Specifically, if there is only one disease group in the LLM treatment results, then the patient can be classified according to that disease group; if there is more than one disease group in the LLM treatment results, or more than three disease groups in the LLM treatment results, then the patient will not be classified into a disease group, so as to avoid over-classification; in this case, manual intervention can be used to resolve the issue.

[0030] S2, based on prompt words, uses a large language model to identify and match the patient's imaging examination items with pre-defined disease groups to obtain the target disease group.

[0031] Examples of the pre-defined disease group classifications are as follows: Head: 1. Mental imaging; 2. Brain tumors; 3. Cerebrovascular diseases; 4. Leukoencephalopathy; 5. Ophthalmic diseases; 6. Otological diseases; 7. Nasopharyngeal carcinoma and other head tumors; Musculoskeletal: 8. Spinal joint diseases; 9. Bone tumors and tumor-like changes; 10. Sports medicine / trauma imaging; Cardiothoracic diseases: 11. Chest tumors; 12. Interstitial lung disease; 13. Congenital heart disease; 14. Cardiomyopathy; 15. Coronary heart disease; 16. Great vessel disease; 17. Breast disease; Abdomen: 18. Liver and gallbladder diseases; 19. Biliary tract diseases; 20. Pancreatic and spleen diseases; 21. Urogenital system diseases; 22. Gastrointestinal and small intestine diseases; 23. Rectal and colonic diseases; 24. Abdominal and retroperitoneal diseases.

[0032] Although the setup of disease-specific groups may vary slightly between different medical institutions’ radiology departments, a diagnostic physician may participate in 2-3 disease-specific groups at the same time.

[0033] Optionally, the aforementioned target disease group can be calculated using the following method: ; In the formula, To obtain the target disease group, For the patient feature vector, the first dimension, Indicates the disease control group The eigenvector of the first dimension, For similarity function, Indicates the weight.

[0034] S3, assign the task of writing the patient's imaging examination report to the shared perspective of the target disease group; then, based on the number of reports to be written by each doctor in the target disease group, assign the attending physician.

[0035] Optionally, the above methods also include: S4. When the identification and matching of the target disease group is incorrect, keywords are retrieved again through a large language model and / or specified by the assigned attending physician until the correct target disease group is obtained.

[0036] In this case, if a diagnosing physician discovers that a patient belongs to another disease group while writing a report for a non-disease group, they can continue writing the report; they can also designate the patient to another disease group and submit the report to that disease group; or they can re-analyze the report using LLM.

[0037] Specifically, if a disease-specific group has insufficient doctors or too many patients on a given day, an upper limit can be set by multiplying the number of doctors in the disease-specific group by the maximum number of reports each doctor in the disease-specific group can complete. New patients exceeding this limit will no longer be submitted to the disease-specific group's perspective but will instead be sent to a public resource pool that is not categorized by disease, allowing all doctors to freely access them. If there are too few patients in the disease-specific management perspective, this perspective can be merged into the public perspective visible to all doctors, placed at the top, and highlighted with a special color. These disease-specific patients are still not visible to doctors outside of the disease-specific group; they can only be seen in the public perspective without needing to click on the disease-specific group's perspective to find them.

[0038] The following examples further illustrate this solution: (1) Define disease groups, such as Figure 2 As shown, it can be defined in the following form: Head: 1. Mental imaging; 2. Brain tumors; 3. Cerebrovascular diseases; 4. Leukoencephalopathy; 5. Ophthalmic diseases; 6. Otological diseases; 7. Nasopharyngeal carcinoma and other head tumors; Musculoskeletal: 8. Spinal joint diseases; 9. Bone tumors and tumor-like changes; 10. Sports medicine / trauma imaging; Cardiothoracic diseases: 11. Chest tumors; 12. Interstitial lung disease; 13. Congenital heart disease; 14. Cardiomyopathy; 15. Coronary heart disease; 16. Great vessel disease; 17. Breast disease; Abdomen: 18. Liver and gallbladder diseases; 19. Biliary tract diseases; 20. Pancreatic and spleen diseases; 21. Urogenital system diseases; 22. Gastrointestinal and small intestine diseases; 23. Rectal and colonic diseases; 24. Abdominal and retroperitoneal diseases.

[0039] (2) When the medical records are a table of examination items, this can be achieved by directly specifying the disease group, such as... Figure 3 As shown, examination items such as abdominal X-ray (KUB) and lumbar spine MRI can be directly specified for a specific disease group.

[0040] (3) When processing application forms and electronic medical records using the LLM model, first configure the LLM call interface address, see [link to relevant documentation]. Figure 4 The LLM model is a relatively mature large language model, and its introduction can be achieved through conventional techniques; when combined with the application form to determine the disease group, such as Figure 5 As shown, create new variables: last similar imaging examination, examination items, clinical diagnosis, examination purpose, etc.; the last similar imaging examination (variable) is retrieved through a stored procedure, which can be accessed through methods such as... Figure 6 The configuration shown is implemented.

[0041] A. Create new suggestion words: such as Figure 7 As shown, the application form analyzes the disease group. Based on the input information, the LLM analyzes and determines which disease group the examination belongs to. The determination rules are as follows: 1. Disease Group IDs and Name Definitions: 1. Mental Imaging; 2. Brain Tumors; 3. Cerebrovascular Diseases; 4. Leukoencephalopathy; 5. Ophthalmological Diseases; 6. Otological Diseases; 7. Nasopharyngeal Carcinoma and other head tumors; 8. Spinal and Joint Diseases; 9. Bone Tumors and Tumor-like Degeneration; 10. Sports Medicine / Trauma Imaging; 11. Thoracic Tumors; 12. Interstitial Lung Disease; 13. Congenital Heart Disease; 14. Cardiomyopathy; 15. Coronary Artery Disease; 16. Great Vessel Diseases; 17. Breast Diseases; 18. Hepatobiliary Diseases; 19. Biliary Tract Diseases; 20. Pancreatic and Spleen Diseases; 21. Genitourinary System Diseases; 22. Gastrointestinal and Small Intestinal Diseases; 23. Rectal and Colonic Diseases; 24. Abdominal and Retroperitoneal Diseases.

[0042] 2. If the application form contains "xx pending investigation", a specific disease group will be assigned based on the information from this imaging examination.

[0043] 3. If the application form includes "xx follow-up", extract the main core diagnostic conclusions based on the previous similar imaging examination report and combine them with the examination items to give a disease group.

[0044] 4. Output Format: A single line of JSON format can be used, for example: {"Disease Group ID":"2","Disease Group Name":"Brain Tumor"}; Disease Group ID: Output the disease group ID defined in the output definition; if no such group exists, output "None". Disease Group Name: Output the disease group name defined in the output definition; if no such group exists, output "None".

[0045] 5. Do not output information, explanations, or descriptions that are irrelevant to the task.

[0046] B. User suggestion word configuration: The following information is analyzed: Previous similar imaging examination: <<Previous similar imaging examination>>; Information on this imaging examination: Examination items: <<Examination items>>; Clinical diagnosis: <<Clinical diagnosis>>; Purpose of examination: <<Purpose of examination>>.

[0047] After completion, it can be done as follows Figure 8 The format shown is sent to the LLM model for analysis. The analysis results returned by the LLM model are as follows: Figure 9 As shown.

[0048] Disease group determination using LLM in conjunction with electronic medical records: 1. Add variables: initial medical record, such as... Figure 10 As shown, the initial medical record (variable) can be retrieved through a stored procedure, as follows: Figure 11 The configuration shown is implemented.

[0049] 2. New prompt: Electronic Medical Record Analysis Specialty Group, see [link / reference] Figure 12 .

[0050] 3. System prompt word configuration: Based on the input information, analyze and determine which disease group the examination belongs to, and the determination rules are as follows: Disease group IDs and names defined as follows: 1. Mental imaging; 2. Brain tumors; 3. Cerebrovascular diseases; 4. Leukoencephalopathy; 5. Ophthalmological diseases; 6. Otological diseases; 7. Nasopharyngeal carcinoma and other cranial tumors; 8. Spinal and joint diseases; 9. Bone tumors and tumor-like changes; 10. Sports medicine / trauma imaging; 11. Thoracic tumors; 12. Interstitial lung disease; 13. Congenital heart disease; 14. Cardiomyopathy; 15. Coronary heart disease; 16. Great vessel disease; 17. Breast diseases; 18. Hepatobiliary diseases; 19. Biliary tract diseases; 20. Pancreatic and splenic diseases; 21. Genitourinary system diseases; 22. Gastrointestinal and small intestinal diseases; 23. Rectal and colorectal diseases; 24. Abdominal and retroperitoneal diseases.

[0051] 4. If more than one disease group is configured, the patient will not be classified into a disease group, and the output will be "None".

[0052] 5. Output Format: Use a single line of JSON format, for example: {"Disease Group ID":"2","Disease Group Name":"Brain Tumor"}. Disease Group ID: Output the disease group ID defined in the output definition; if no such group exists, output "None". Disease Group Name: Output the disease group name defined in the output definition; if no such group exists, output "None".

[0053] 6. Do not output information, explanations, or descriptions that are irrelevant to the task.

[0054] Specifically, in practical applications, the InterRis trigger table T_I_STATUS_SQL can be configured to trigger the execution of the stored procedure P_LLM_SET_SD to add an LLM task during patient registration, such as... Figure 13 As shown; further, from the perspective of configuring disease reports in practical applications: unwritten, unreviewed, and reviewed, the SQL conditions include the disease group ID as a limiting condition. See [link / reference]. Figure 14 Configure disease-specific perspective permissions for each user and user group. A diagnosing physician can participate in 2-3 disease-specific groups simultaneously. In case of incorrect grouping of disease-specific groups, such as... Figure 15 As shown, in the work list, select the patient, click "Set Disease Group" in OEM, and modify or remove the patient's disease group; see also Figure 16 Handling situations where there are too many or too few patients in a specific disease group. Too many patients in a specific disease group: When a patient's examination is completed, the InterRIS stored procedure P_LLM_SET_SD_EX is triggered. If the number of examinations completed in the patient's specific disease group exceeds the maximum number of reports a doctor can complete, the patient's F_SD_ID is set to 0. Too few patients in a specific disease group: Configure a public view, sorted in descending order by F_SD_ID. Set the list status so that records with F_SD_ID > 0 are displayed in blue. Set the SQL query conditions for this view for users; when users query this view, these conditions must be included.

[0055] Specifically, by leveraging the general text analysis capabilities of LLM, the content of application forms and related electronic medical records is analyzed to determine the area of ​​expertise and complexity of the diagnosis, and based on this determination, it is decided what level of physician should be assigned to complete the diagnostic task. 1. Predefined tables.

[0056] 1.1 The scenario predefined table is shown in Table 1: Table 1 1 Chest CT Health screening / physical examination Age <50 years old The risk is extremely low, and there is no need to waste resources; The system utilizes AI to perform image analysis; doctors then write preliminary reports; and the system can self-review without requiring doctor review. 2 Chest CT Preoperative examination Age <50 years; Previous chest CT report: Main lesion was ground-glass nodules Young, low risk of relocation; First, use AI to perform image analysis; then, assign a junior physician to write a preliminary report; finally, assign a non-specialist physician to review the report. 3 Chest CT Postoperative follow-up Surgical pathology results: Small cell carcinoma The possibility of relocation is high, and it is relatively covert; Arrange for junior doctors specializing in a particular disease to write preliminary reports; arrange for specialist doctors to review the reports; 1.11. Purpose of application: To use LLM to analyze the application form to see if it matches the feature.

[0057] 1.12. Patient Medical History: Analyze the patient's medical history using LLM within the electronic medical record to see if these features match simultaneously. Since electronic medical records may be divided into multiple sub-modules, if so, use the noun before the colon to select these modules. For example, for "surgical pathology results," select the pathology sub-module to extract information.

[0058] 1.13. Process Method Description: Define a specific process method based on the specific application purpose and patient medical history conditions; mainly including three aspects: whether to arrange the use of a specific AI module; who to assign to write the preliminary report; who to assign to complete the review report; the RIS system RBA (process robot) uses LLM to analyze the text to determine the direction of the process.

[0059] 1.2. Predefined personnel competency table, example shown in Table 2: Table 2 1 Zhang San 1234 Reviewing Doctor Brain tumors, cerebrovascular diseases 2 Li Si 5678 Preliminary report by the doctor Coronary heart disease, large vessel disease 1.3 List of definitions for specific diseases Head: 1. Mental imaging; 2. Brain tumors; 3. Cerebrovascular diseases; 4. Leukoencephalopathy; 5. Ophthalmic diseases; 6. Otological diseases; 7. Nasopharyngeal carcinoma and other cranial tumors; Musculoskeletal: 8. Spinal joint diseases; 9. Bone tumors and tumor-like changes; 10. Sports medicine / trauma imaging; Cardiothoracic diseases: 11. Chest tumors; 12. Interstitial lung disease; 13. Congenital heart disease; 14. Cardiomyopathy; 15. Coronary heart disease; 16. Great vessel disease; 17. Breast disease; Abdomen: 18. Liver and gallbladder diseases; 19. Biliary tract diseases; 20. Pancreatic and spleen diseases; 21. Urogenital system diseases; 22. Gastrointestinal and small intestine diseases; 23. Rectal and colonic diseases; 24. Abdominal and retroperitoneal diseases.

[0060] 2. Diagnostic task allocation 2.1 Preliminary Report Specialty Perspective vs. Review Report Specialty Perspective: If a patient is assigned to a specific disease perspective, the initial report will be presented in the specific disease perspective, and the review report will be presented in the specific disease perspective. These two perspectives are accessible to the respective doctors who will be reporting the initial report and reviewing the review report for patients who meet the specific disease criteria.

[0061] 2.2 Personal disease-specific learning perspective and personal disease-specific competence perspective: A preliminary reporting physician can participate in the learning of multiple disease specialties; a reviewing physician can also have multiple professional strengths. For each of them, the collection of all the perspectives they participate in forms their respective "personal disease-specific learning perspective" (for the preliminary reporting physician) and "personal disease-specific competence perspective" (for the reviewing reporting physician).

[0062] It's conceivable that everyone's combination of specialized perspectives will likely be different, but it will certainly be a professional type that they want to learn or are good at.

[0063] As mentioned above, matching the professional field and complexity of diagnosis with the doctor's ability can effectively improve diagnostic efficiency and quality; even without using other complex quality control methods, it can produce a significant improvement in diagnostic quality and has the potential to be promoted in medical institutions at all levels.

[0064] Example 2: This application provides an image diagnostic allocation system based on a large language model, applied to the image diagnostic allocation method based on a large language model in Example 1, including: The feature extraction module is used to set corresponding prompts based on the patient's medical records, which include at least the names of imaging examination items, application forms, and electronic medical records. The disease group matching module is used to identify and match the patient's imaging examination items with the pre-defined disease group categories based on prompt words and a large language model to obtain the target disease group; The doctor assignment module is used to assign the task of writing the patient's imaging examination report to the shared perspective of the target disease group.

[0065] Furthermore, the aforementioned system also includes: The reassignment module is used to retrieve keywords again through a large language model and / or specify them through the assigned attending physician when the identification and matching of the target disease group is incorrect, until the correct target disease group is obtained.

[0066] Example 3: This application provides an electronic device, including: at least one processor, at least one memory, and a data bus; The processor and memory communicate with each other via a data bus; the memory stores program instructions that can be executed by the processor, and the processor calls the program instructions to execute an image diagnosis allocation method based on a large language model, as described in Embodiment 1.

[0067] Example 4: This application provides a non-transitory computer-readable storage medium that stores computer instructions, which cause the computer to execute an image diagnosis allocation method based on a large language model according to Example 1.

[0068] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0069] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0070] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0071] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0072] Those skilled in the art will understand that all or part of the steps in the above facts and methods can be implemented by a program instructing related hardware. The program or the program described therein can be stored in a computer-readable storage medium. When the program is executed, it includes the following steps: at this time, the corresponding method steps are introduced. The storage medium can be ROM / RAM, magnetic disk, optical disk, etc.

[0073] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. An image diagnostic allocation method based on a large-scale language model, characterized in that, The specific steps include the following: Based on the patient's medical records, set corresponding prompts, wherein the patient's medical records include at least the names of imaging examination items, application forms, and electronic medical records; Based on the prompt words, the patient's imaging examination items are identified and matched with the pre-defined disease groups using a large language model to obtain the target disease group; The task of writing the patient's imaging examination report was assigned to the shared perspective of the target disease group.

2. The image diagnosis allocation method based on a large language model according to claim 1, characterized in that, The method involves setting corresponding prompts based on the patient's medical records, including: When the patient's medical records contain the name of an imaging examination item, the system matches the name of the imaging examination item with a pre-set table corresponding to imaging examinations and disease group classifications. When a patient's medical record is an application form and the application form shows that a certain area needs to be examined, the text of the application form and the disease group type included in the area to be examined are set as prompt words; when the patient's medical record is an application form and the application form shows that a certain area needs to be followed up, the core diagnostic conclusion of the patient's last imaging examination report and the disease group type included in the scanned area are set as prompt words. When a patient's medical records are electronic medical records, set the entire text of the patient's first medical record and the disease group type included in the scanned area as prompt words.

3. The image diagnostic allocation method based on a large language model according to claim 1, characterized in that, The method also includes: Monitor the image examination report writing tasks of patients in the shared perspective of each disease group. When the number of image examination report writing tasks of patients exceeds the preset value, the image examination report writing tasks of new patients will be allocated to the public resource pool.

4. The image diagnostic allocation method based on a large language model according to claim 1, characterized in that, The method further includes: When an error occurs in the identification and matching of the target disease group, keywords are retrieved again through a large language model and / or specified by the assigned attending physician until the correct target disease group is obtained.

5. The image diagnostic allocation method based on a large language model according to claim 1, characterized in that, The target disease group was obtained through the following methods: ； In the formula, To obtain the target disease group, For the patient feature vector, the first dimension, Indicates the disease control group The eigenvector of the first dimension, For similarity function, Indicates the weight.

6. An image diagnosis and allocation system based on a large-scale language model, characterized in that, include: The feature extraction module is used to set corresponding prompts based on the patient's medical records, wherein the patient's medical records include at least the names of imaging examination items, application forms, and electronic medical records; The disease group matching module is used to identify and match the patient's imaging examination items with the preset disease group classifications based on the prompt words and through a large language model to obtain the target disease group. The doctor assignment module is used to assign the task of writing the patient's imaging examination report to the shared perspective of the target disease group.

7. The image diagnostic allocation system based on a large language model according to claim 5, characterized in that, The method involves setting corresponding prompts based on the patient's medical records, including: When the patient's medical records contain the name of an imaging examination item, the system matches the name of the imaging examination item with a pre-set table corresponding to imaging examinations and disease group classifications. When a patient's medical record is an application form and the application form shows that a certain area needs to be examined, the text of the application form and the disease group type included in the area to be examined are set as prompt words; when the patient's medical record is an application form and the application form shows that a certain area needs to be followed up, the core diagnostic conclusion of the patient's last imaging examination report and the disease group type included in the scanned area are set as prompt words. When a patient's medical records are electronic medical records, set the entire text of the patient's first medical record and the disease group type included in the scanned area as prompt words.

8. The image diagnostic allocation system based on a large language model according to claim 5, characterized in that, The system also includes: The monitoring module is used to monitor the image examination report writing tasks of patients in the shared perspective of each disease group. When the number of image examination report writing tasks of patients exceeds the preset value, the image examination report writing tasks of new patients will be allocated to the public resource pool.

9. An electronic device, characterized in that, include: At least one processor, at least one memory, and a data bus; The processor and the memory communicate with each other via the data bus. The memory stores program instructions that can be executed by the processor, which invokes the program instructions to execute an image diagnostic allocation method based on a large language model as described in any one of claims 1-5.

10. A non-transitory computer-readable storage medium, characterized in that, The non-transitory computer-readable storage medium stores computer instructions that cause the computer to execute the image diagnostic allocation method based on a large language model according to any one of claims 1-5.

Images