An ai electronic medical record pre-filling method and system based on diagnosis dialogue

Abstract

The application discloses an AI electronic medical record pre-filling method and system based on diagnosis dialogue, relates to the technical field of artificial intelligence, and comprises the following steps: acquiring multi-source information of a current patient, wherein the multi-source information comprises but is not limited to original doctor-patient diagnosis dialogue text, a patient payment receipt, a historical medical record, an examination result and a doctor's medical record draft; a scene quadrant corresponding to each disease is pre-configured, and a historical benchmark medical record is set; the multi-source information of the current patient is matched with the historical benchmark medical record through a large model, and the scene quadrant to which the current scene belongs is determined in real time; and a differentiated filling strategy of an AI electronic medical record is executed according to different scene quadrants, so that the problem that the prior art cannot realize differentiated completion of AI electronic medical record pre-filling based on diagnosis dialogue is solved.

Description

Technical Field

[0001] This invention relates to the field of artificial intelligence technology, and in particular to an AI-based electronic medical record pre-filling method and system based on consultation dialogue. Background Technology

[0002] Current AI-assisted electronic medical record pre-filling technologies mostly use a single, fixed template for automatic record filling, without adapting to the specific differences in real clinical consultation scenarios. Clinical consultation scenarios can be divided into four different categories based on complexity and frequency of visits: high-frequency simple consultations, low-frequency simple consultations, high-frequency complex consultations, and low-frequency complex consultations. The requirements for record writing, information completeness, and focus of filling all differ significantly. Current generic pre-filling solutions cannot differentiate between these diverse consultation scenarios, employing a uniform filling logic throughout. This easily leads to problems such as redundant and complex records for simple consultations and missing key information in complex consultations, resulting in poor relevance of the pre-filled content. Furthermore, existing technologies lack a benchmark medical record comparison and recognition mechanism, failing to adapt to the current consultation situation based on historical high-quality medical records. This significantly increases the workload for doctors to manually delete and supplement medical records later, reducing the efficiency of clinical diagnosis and treatment and medical record writing.

[0003] Therefore, there is an urgent need for a method that can achieve AI-based pre-filling of electronic medical records based on the differentiation of consultation dialogue. Summary of the Invention

[0004] In view of this, the present invention proposes an AI-based electronic medical record pre-filling method and system based on consultation dialogue, which can achieve data security isolation and privacy protection, and ensure the security and compliant use of data throughout the entire process.

[0005] To achieve the above objectives, the present invention provides the following technical solution: An AI-based electronic medical record pre-filling method based on online consultation includes: Obtain multi-source information of the current patient, including original doctor-patient consultation dialogue text, patient payment slip, historical medical records, examination results, and doctor's medical record drafts; Pre-configure the scenario quadrants corresponding to each disease and set historical benchmark medical records; By matching the multi-source information of the current patient with the historical benchmark medical records through a large model, the scene quadrant to which the current scene belongs can be determined in real time. Implement differentiated filling strategies for AI electronic medical records based on different scenario quadrants.

[0006] Based on the above technical solution, the present invention can be further improved as follows: Optionally, the scene quadrant includes simple high-frequency scene, simple low-frequency scene, complex high-frequency scene, and complex low-frequency scene.

[0007] Optionally, the differentiated filling strategy for AI electronic medical records according to different scenario quadrants includes: When the current scenario belongs to the simple and high-frequency scenario quadrant, the corresponding dedicated medical record template, term template and default value configuration are invoked; The original doctor-patient consultation dialogue text was copied and filled in. Logical consistency verification of the populated content is performed based on a knowledge graph. Randomly sample and review AI electronic medical records that have completed logical consistency verification.

[0008] Optionally, the differentiated filling strategy for AI electronic medical records according to different scenario quadrants further includes: When the current scenario belongs to the complex and high-frequency scenario quadrant, a customized medical record template marked with auto-fill, core verification, and default content attributes is invoked. The AI ​​electronic medical record framework is filled with auto-fill fields and default content fields, and the core verification fields are highlighted for doctors to confirm.

[0009] Optionally, the differentiated filling strategy for AI electronic medical records according to different scenario quadrants further includes: When the current scenario belongs to the scenario quadrant of simple low-frequency scenario, a general medical record template is used for filling, terminology normalization is performed, and low confidence fields are highlighted.

[0010] Optionally, the differentiated filling strategy for AI electronic medical records according to different scenario quadrants further includes: When the current scene belongs to the complex low-frequency scene quadrant, a complete AI electronic medical record is not generated, but only auxiliary materials are generated. The auxiliary materials include dialogue summaries, medical history summaries, examination results, and payment item lists.

[0011] Optionally, after implementing the differentiated filling strategy steps for AI electronic medical records according to different scenario quadrants, the process may also include: The AI ​​electronic medical record is subjected to quality inspection to obtain the quality inspection data of the AI ​​electronic medical record; The scene quadrant is adaptively upgraded or downgraded based on the quality inspection data.

[0012] An AI-powered electronic medical record pre-filling system based on online consultation includes: The information acquisition module is used to acquire multi-source information of the current patient, including the original doctor-patient consultation dialogue text, patient payment slip, historical medical records, examination results and doctor's medical record drafts. The preset configuration module is used to pre-configure the scenario quadrants corresponding to each disease and set historical benchmark medical records; The scene determination module is used to match the current patient's multi-source information with the historical benchmark medical records through a large model, and determine the scene quadrant to which the current scene belongs in real time. The differentiated entry module is used to execute differentiated entry strategies for AI electronic medical records according to different scenario quadrants.

[0013] An electronic device includes a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the processor executes the computer program to implement the steps of the method described herein.

[0014] A non-transitory computer-readable storage medium having a computer program stored thereon, the computer program implementing the steps of the method when executed by a processor.

[0015] The present invention has the following advantages: This invention presents an AI-based electronic medical record pre-filling method based on consultation dialogue. It integrates various information sources, including consultation dialogues and examination reports, and combines pre-defined disease-specific quadrants with historical benchmark medical records. Leveraging a large-scale model, it accurately identifies the current consultation scenario. Differentiated medical record filling strategies are adopted for different scenarios, adapting to various clinical consultation needs, including simple, complex, high-frequency, and low-frequency cases. This effectively improves the accuracy and relevance of electronic medical record pre-filling, reduces manual modifications by medical staff, lowers the workload of medical record writing, improves the efficiency of medical record compilation, and ensures standardized and consistent medical record content, better meeting the requirements of actual clinical use. Attached Figure Description

[0016] For illustrative and not limiting purposes, the present invention will now be described in conjunction with embodiments and accompanying drawings, wherein: Figure 1 This is a flowchart illustrating the AI-based electronic medical record pre-filling method based on consultation dialogue in an embodiment of the present invention. Figure 2 This is a schematic diagram of the main components of the AI ​​electronic medical record pre-filling system based on consultation dialogue in an embodiment of the present invention; Figure 3 This is a schematic diagram of the physical structure of the electronic device provided by the present invention. Detailed Implementation

[0017] To enable those skilled in the art to better understand the present invention, 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 a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0018] It should be noted that the terms "first," "second," etc., in the specification and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be used interchangeably where appropriate for the embodiments of the invention described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0019] It should be noted that, where there is no conflict, the embodiments and features of the present invention can be combined with each other. The embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0020] Figure 1 This is a flowchart illustrating the AI-based electronic medical record pre-filling method based on consultation dialogue in an embodiment of the present invention, as shown below. Figure 1 As shown, the AI ​​electronic medical record pre-filling method based on consultation dialogue provided in this embodiment of the invention includes the following steps S101 to S104.

[0021] S101, Obtain multi-source information for the current patient.

[0022] Multi-source information includes original doctor-patient consultation dialogue texts, patient payment slips (including examination items, drug lists, treatment items, inpatient items, amounts, etc.), historical medical records (including those from this department, other departments, other hospitals, etc.), examination results, and doctor's draft medical records.

[0023] S102, pre-configure the scene quadrants corresponding to each disease and set historical benchmark medical records.

[0024] The scenario quadrants include simple high-frequency scenarios, simple low-frequency scenarios, complex high-frequency scenarios, and complex low-frequency scenarios.

[0025] S103 uses a large model to match the current patient's multi-source information with historical benchmark medical records, and determines the scene quadrant to which the current scene belongs in real time.

[0026] S104, implement differentiated filling strategies for AI electronic medical records according to different scenario quadrants.

[0027] When the current scenario belongs to the simple and high-frequency scenario quadrant, the corresponding dedicated medical record template, term template and default value configuration are invoked; The term template supports configuring standardized terms according to scenarios; The original doctor-patient consultation dialogue text was copied and filled in. Logical consistency verification of the populated content is performed based on a knowledge graph. The knowledge graph includes a standard medical knowledge graph and a customizable medical knowledge graph. The standard medical knowledge graph covers standard terminology systems such as ICD, while the customizable medical knowledge graph supports the configuration of synonyms and common expressions at the hospital / department level.

[0028] Randomly sample and review AI electronic medical records that have completed logical consistency verification.

[0029] When the current scenario belongs to the complex and high-frequency scenario quadrant, a customized medical record template marked with auto-fill, core verification, and default content attributes is invoked. The AI ​​electronic medical record framework is filled with auto-fill fields and default content fields, and the core verification fields are highlighted for doctors to confirm.

[0030] Customized medical record templates: Supports configuring field structure and field attributes by disease type; When the current scenario belongs to the scenario quadrant of simple low-frequency scenario, a general medical record template is used for filling, terminology normalization is performed, and low confidence fields are highlighted.

[0031] When the current scene belongs to the complex low-frequency scene quadrant, a complete AI electronic medical record is not generated, but only auxiliary materials are generated. The auxiliary materials include dialogue summaries (categorized by timeline / topic), a summary of past medical history (extracted from historical medical records), examination results, and a list of payment items.

[0032] The following dual-track mapping mechanism is adopted: Automated selection of medical record templates; Fill template (original information); The large model automatically fills the template based on the template and the original input information; The filling process is divided into selecting a term template and filling in atomic information.

[0033] The content to be filled in is: the original (dialogue) text; Candidate knowledge retrieval: Based on the original dialogue text, candidate terms are retrieved from both the customized knowledge base and the standard knowledge base; Large model normalization: The large model combines the original text and two candidate knowledge lists to normalize colloquial expressions into knowledge base corpus, and provides confidence scores and reasons for normalization; Confidence-based splitting: High confidence mapping: Directly fill in the medical record; Low confidence mapping: Enter the doctor confirmation queue and highlight it; Priority rule: Customizable knowledge graph > Standard knowledge graph > Large model freely generated (ensuring that the hospital's localized configuration is not covered by the general model).

[0034] The AI ​​electronic medical record is subjected to quality inspection to obtain the quality inspection data of the AI ​​electronic medical record; The scene quadrant is adaptively upgraded or downgraded based on the quality inspection data.

[0035] Upgrade Trigger: When the number of cases for a certain disease accumulates to a threshold in the low-frequency quadrant, and the doctor's quality inspection pass rate is higher than the set value, the system prompts the administrator to configure a dedicated template to upgrade to the high-frequency quadrant. Degradation Trigger: When the sampling error rate in a high-frequency scenario exceeds the threshold, the system will automatically downgrade and issue an alarm.

[0036] One embodiment is as follows: Taking a real-world dental clinic scenario as an example, this paper explains the complete system operation process under the four scenario quadrants of this invention one by one, which is consistent with the actual clinical treatment process and intuitively reflects the differentiated medical record filling logic.

[0037] 1. Simple, high-frequency scenario: Tooth filling (follow-up appointment) Patient input information: Consultation dialogue: Doctor: Which tooth is bothering you? Patient: The filling on my lower left molar fell out, and it's a bit sore when I eat.

[0038] Doctor: How long has it been? Patient: It's been two days, and I haven't experienced any spontaneous pain.

[0039] Doctor: Okay, let me take a look. (After examination) Tooth #36 has secondary caries and needs to be prepared and filled again.

[0040] (Remaining dialogue information)... Payment slip: Composite resin filling ×1, tooth preparation ×1.

[0041] Historical medical records: Record of filling treatment for tooth 36 6 months ago.

[0042] Appointment Information: Patient Wang, female, 28 years old, came for a follow-up visit due to "the filling in her lower left posterior tooth falling out 2 days ago".

[0043] Quadrant determination: The large model matches and compares the patient's multi-source input information with historical benchmark medical records of "follow-up visits for dental caries filling". If the matching degree meets the standard, the current consultation scenario is determined to be a simple and high-frequency scenario.

[0044] System processing flow: The first step is to access the medical record template for follow-up appointments for dental fillings and the accompanying dental terminology database.

[0045] The second step is to directly copy and fill the core medical record fields based on the original consultation dialogue text: the chief complaint is directly copied as "left lower posterior tooth filling fell off 2 days ago"; the present illness history is directly extracted from the dialogue content and intelligently spliced ​​as "left lower posterior tooth filling fell off 2 days ago, soreness when eating, no spontaneous pain"; the oral examination content is directly copied as "secondary caries in tooth 36".

[0046] The third step is to perform medical terminology normalization: the spoken description "the lower left molar" is entered into the dental tooth position customization knowledge base to generate candidate terms [36th tooth, mandibular first molar]. The large model combines the treatment context of composite resin filling in the payment slip and finally normalizes it into the standard medical term "36th tooth" with a confidence level of 0.97; the spoken symptom "acidity" is normalized into the standard oral term "pain from hot or cold stimuli" with a confidence level of 0.91.

[0047] The fourth step is to automatically fill in the system default values: fill the vital signs field with the system default value "No abnormalities found"; and directly use the patient's historical medical record "None" for the allergy history field.

[0048] The fifth step involves conducting a logical consistency check based on the medical knowledge graph: the system verifies that the diagnosis result "secondary caries" matches the treatment item "composite resin filling", the medical record treatment logic is correct, and the check passes.

[0049] System outputs medical records: Chief complaint: Filling in the lower left posterior tooth fell out 2 days ago; Present illness: The patient underwent filling treatment for tooth 36 at our hospital 6 months ago. The filling fell out 2 days ago. The patient experiences pain when eating, but no spontaneous pain or night pain.

[0050] Examination: Tooth 36 has secondary caries, filling has fallen out, sensitive to probing, negative percussion, and no redness or swelling of the gums.

[0051] Diagnosis: Secondary caries in tooth 36; Treatment: Tooth preparation for tooth 36, and filling with composite resin; Doctor's action: This medical record is automatically entered into the system's random sampling pool. Since it was not sampled this time, the medical record can be archived directly without the doctor needing to perform any manual editing.

[0052] 2. Complex and frequent scenarios: Chronic periodontitis (initial visit) Appointment Information: Patient Li, male, 52 years old, made his first visit to the Department of Stomatology due to "recurrent bleeding gums and halitosis for half a year".

[0053] Quadrant determination: The initial diagnosis of chronic periodontitis requires the collection of multiple tooth positions and multiple periodontal specialist examination indicators, making the diagnosis and treatment logic complex; at the same time, this disease is a common disease in dental clinics, with a high frequency of visits. The large model comprehensively determines that this scenario is a complex and high-frequency scenario.

[0054] Administrator preset template field attributes: |Field|Attribute|Description| |---|---|---| |Chief complaint|auto|Extracted from dialogue| |Present medical history|auto|Extracted from dialogue| |Periodontal probing depth (PD)|key|Required for doctors, clinical measurement| |Loss of Attachment (AL)|key|Required for physicians, clinical measurement| |Looseness|key|Required for Doctor| |Diagnosis|key|Required for Doctor| |Oral Hygiene Guidance|default|Default value: "Oral hygiene education has been provided, including guidance on proper brushing techniques and flossing."| |Treatment Plan|auto|Retrieved from Conversation + Payment Slip| System processing flow: The system automatically fills in the content of all auto attribute fields, fills in the preset default content of the default attribute field, and highlights all key attribute fields, waiting for the doctor to supplement and verify.

[0055] Doctor's operation: Doctors only need to fill in the on-site clinical measurement data such as periodontal probing depth, attachment loss, and tooth mobility in the highlighted core field area, and at the same time supplement the diagnosis results of mild, moderate and severe periodontitis. The other non-highlighted fields do not need to be modified or edited.

[0056] 3. Simple, low-frequency scenario: Minor oral mucosal ulcers (initial visit) Appointment Information: Patient Zhang, female, 35 years old, came to see a doctor because of "painful swelling in the oral cavity for 3 days".

[0057] Quadrant assessment: The patient's symptoms were simple and the condition was straightforward, but the frequency of visits to the oral ulcer specialist clinic was much lower than that for dental pulp and periodontal diseases. There was no customized medical record template. Based on the comprehensive assessment of the large model, this scenario was determined to be a simple and low-frequency scenario.

[0058] System processing flow: The first step is to use a standard electronic medical record template for dentistry to build a standard medical record text framework.

[0059] The second step involves normalizing medical terminology: the patient's verbal description of "a blister in my mouth" is entered into a standard medical knowledge base to generate candidate terms [oral ulcer, oral mucosal ulcer, herpetic stomatitis]; the large model, combined with the patient's 3-day history of pain without fever, finally normalizes it into the standard term "oral mucosal ulcer" with a confidence level of 0.82, and includes it in the manual confirmation queue; the specific location of the ulcer was not clearly stated in the consultation dialogue, and the confidence level of this field was only 0.41. The system highlights this field to remind the doctor to supplement and improve it.

[0060] The third step is to automatically generate a draft medical record, with all low-confidence identification fields highlighted in yellow.

[0061] Doctor's procedure: The doctor performs a quality check on the entire medical record, focusing on checking the highlighted suspicious fields, supplementing and improving the specific location of the ulcer as "left buccal mucosa", and after confirming that the medical terminology automatically normalized by the system is correct, completes the medical record signing confirmation.

[0062] 4. Complex low-frequency scenarios: Infections in the oral and maxillofacial spaces Appointment Information: Patient Zhao, male, 45 years old, presented to the emergency department with "swelling on the right side of the face, limited mouth opening, fever of 38.5℃, and a course of illness of 3 days". The patient has a history of diabetes.

[0063] Quadrant assessment: Maxillofacial space infection requires differentiation of multiple space infection types and the development of a treatment plan based on the patient's history of chronic diseases, making the condition complex; at the same time, this disease belongs to the oral emergency, with an extremely low incidence rate in daily visits. The large model comprehensively judges this scenario as a complex low-frequency scenario.

[0064] System Processing Flow: Considering the lack of a fixed diagnostic and treatment paradigm and the scarcity of clinical samples in this type of scenario, the system does not directly generate complete electronic medical records. Instead, it intelligently integrates information from multiple sources and outputs structured diagnostic and treatment auxiliary materials, as detailed below: [Dialogue Summary] Main symptoms: Swelling on the right side of the face for 3 days, with progressive worsening of symptoms; Accompanying symptoms: Limited mouth opening, body temperature of 38.5℃, and pain when swallowing; Suspected cause: The patient reported a history of right posterior tooth pain one week ago; General health: History of diabetes, regularly taking metformin to lower blood sugar.

[0065] [Past Medical History Summary (Retrieve Historical Medical Records)] In March 2024, the patient was diagnosed with chronic periapical periodontitis in tooth 47, and the root canal treatment was not completed; in 2022, the patient was diagnosed with type 2 diabetes in the Department of Internal Medicine and has been managing blood sugar with medication for a long time.

[0066] [Examination and Payment List] Blood routine examination and panoramic oral radiograph examination to be completed; Payment items: incision and drainage procedure ×1, panoramic oral radiograph examination ×1.

[0067] Doctor's Operation: Based on all the auxiliary materials output by the system and combined with the results of on-site clinical examination, the doctor can independently complete the writing of a complete medical record, including core clinical content such as the determination of the range of infection interval, the surgical plan for incision and drainage, and the specific management of blood glucose in the perioperative period.

[0068] 5. Specific examples of scene quadrant adaptive upgrading and downgrading (1) Example of scene upgrade: Oral mucosal minor ulcers were initially classified as a simple, low-frequency scenario. During clinical operation at a top-tier tertiary dental hospital, a total of 38 cases of this disease were treated, with an overall doctor's record quality inspection pass rate of 97.4%, exceeding the system's preset upgrade threshold (patient volume ≥ 30 cases, quality inspection pass rate ≥ 95%). The system automatically pushed a notification to the administrator: "Oral mucosal minor ulcers have met the scenario upgrade conditions; it is recommended to configure a dedicated record template for this disease." After the administrator approved the upgrade, a dedicated template was configured for this disease, and oral mucosal minor ulcers were officially upgraded from a simple, low-frequency scenario to a simple, high-frequency scenario.

[0069] (2) Example of scenario degradation: A dental hospital initially categorized pericoronitis of wisdom teeth as a simple, high-frequency scenario. After two months of continuous system operation, random sampling of medical records showed that the automatic entry error rate for this condition reached 12%, exceeding the system's preset error rate threshold of 8%. The system automatically downgraded the corresponding consultation scenario for this condition from a simple, high-frequency scenario to a simple, low-frequency scenario, and simultaneously issued an operational alert to the administrator, prompting the administrator to review the medical record template parameters and benchmark medical record configuration rules for this condition.

[0070] Figure 2 This is a schematic diagram illustrating the main components of the AI-based electronic medical record pre-filling system based on consultation dialogue in an embodiment of the present invention. Figure 2 As shown, the AI ​​electronic medical record pre-filling system 1 based on consultation dialogue provided in this embodiment of the invention includes an information acquisition module 10, a preset configuration module 20, a scene determination module 30, and a differentiated filling module 40.

[0071] Information acquisition module 10 is used to acquire multi-source information of the current patient, including original doctor-patient consultation dialogue text, patient payment slip, historical medical records, examination results and doctor's medical record draft; The preset configuration module 20 is used to pre-configure the scene quadrants corresponding to each disease and set historical benchmark medical records; The scene determination module 30 is used to match the multi-source information of the current patient with the historical benchmark medical records through a large model, and determine the scene quadrant to which the current scene belongs in real time. The differentiated filling module 40 is used to execute differentiated filling strategies for AI electronic medical records according to different scenario quadrants.

[0072] Figure 3This is a schematic diagram of the physical structure of an electronic device provided in an embodiment of the present invention, such as... Figure 3 As shown, the electronic device 50 includes: a processor 501, a memory 502, and a bus 503; The processor 501 and the memory 502 communicate with each other via the bus 503. The processor 501 is used to call program instructions in the memory 502 to execute the methods provided in the above-described method embodiments, and to execute the methods provided in the embodiments of the present invention.

[0073] This embodiment provides a non-transitory computer-readable storage medium that stores computer instructions, which cause a computer to execute the method provided in this embodiment of the invention.

[0074] Those skilled in the art will understand that all or part of the steps of the above method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When the program is executed, it performs the steps of the above method embodiments. The aforementioned storage medium includes various storage media capable of storing program code, such as ROM, RAM, magnetic disk, or optical disk.

[0075] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can occur depending on design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A method for pre-filling AI-based electronic medical records based on consultation dialogue, characterized in that, include: Obtain multi-source information of the current patient, including original doctor-patient consultation dialogue text, patient payment slip, historical medical records, examination results, and doctor's medical record drafts; Pre-configure the scenario quadrants corresponding to each disease and set historical benchmark medical records; By matching the multi-source information of the current patient with the historical benchmark medical records through a large model, the scene quadrant to which the current scene belongs can be determined in real time. Implement differentiated filling strategies for AI electronic medical records based on different scenario quadrants.

2. The AI-based electronic medical record pre-filling method based on consultation dialogue as described in claim 1, characterized in that, The scenario quadrants include simple high-frequency scenarios, simple low-frequency scenarios, complex high-frequency scenarios, and complex low-frequency scenarios.

3. The AI-based electronic medical record pre-filling method based on consultation dialogue as described in claim 1, characterized in that, The differentiated filling strategy for AI electronic medical records according to different scenario quadrants includes: When the current scenario belongs to the simple and high-frequency scenario quadrant, the corresponding dedicated medical record template, term template and default value configuration are invoked; The original doctor-patient consultation dialogue text was copied and filled in. Logical consistency verification of the populated content is performed based on a knowledge graph. Randomly sample and review AI electronic medical records that have completed logical consistency verification.

4. The AI-based electronic medical record pre-filling method based on consultation dialogue as described in claim 1, characterized in that, The differentiated filling strategy for AI electronic medical records based on different scenario quadrants also includes: When the current scenario belongs to the complex and high-frequency scenario quadrant, a customized medical record template marked with auto-fill, core verification, and default content attributes is invoked. The AI ​​electronic medical record framework is filled with auto-fill fields and default content fields, and the core verification fields are highlighted for doctors to confirm.

5. The AI-based electronic medical record pre-filling method based on consultation dialogue as described in claim 1, characterized in that, The differentiated filling strategy for AI electronic medical records based on different scenario quadrants also includes: When the current scenario belongs to the scenario quadrant of simple low-frequency scenario, a general medical record template is used for filling, terminology normalization is performed, and low confidence fields are highlighted.

6. The AI-based electronic medical record pre-filling method based on consultation dialogue as described in claim 1, characterized in that, The differentiated filling strategy for AI electronic medical records based on different scenario quadrants also includes: When the current scene belongs to the complex low-frequency scene quadrant, a complete AI electronic medical record is not generated, but only auxiliary materials are generated. The auxiliary materials include dialogue summaries, medical history summaries, examination results, and payment item lists.

7. The AI-based electronic medical record pre-filling method based on consultation dialogue as described in claim 1, characterized in that, After implementing differentiated filling strategies for AI electronic medical records according to different scenario quadrants, the process also includes: The AI ​​electronic medical record is subjected to quality inspection to obtain the quality inspection data of the AI ​​electronic medical record; The scene quadrant is adaptively upgraded or downgraded based on the quality inspection data.

8. An AI-based electronic medical record pre-filling system based on online consultation, characterized in that, include: The information acquisition module is used to acquire multi-source information of the current patient, including the original doctor-patient consultation dialogue text, patient payment slip, historical medical records, examination results and doctor's medical record drafts. The preset configuration module is used to pre-configure the scenario quadrants corresponding to each disease and set historical benchmark medical records; The scene determination module is used to match the current patient's multi-source information with the historical benchmark medical records through a large model, and determine the scene quadrant to which the current scene belongs in real time. The differentiated entry module is used to execute differentiated entry strategies for AI electronic medical records according to different scenario quadrants.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the method as described in any one of claims 1 to 7.

10. A non-transitory computer-readable medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method as described in any one of claims 1 to 7.

Images