Internet hospital multi-modal chief complaint conflict grading processing and doctor collaboration method, system, electronic device and storage medium

Abstract

The application discloses an internet hospital multi-modal chief complaint conflict grading processing and doctor cooperation method, system, electronic equipment and storage medium. The method comprises the following steps: receiving multi-modal chief complaint data; independently analyzing the multi-modal chief complaint data to generate structured chief complaints; performing field-level comparison on the structured chief complaints to check whether there is a conflict description; calculating a conflict severity score according to the field-level comparison result; performing three-level shunt processing according to the conflict severity score and two preset grading thresholds; and generating a doctor takeover list when the doctor cooperation evaluation processing and / or forced locking of the multi-modal chief complaint data is triggered. The application can improve the efficiency of processing chief complaint conflicts.

Description

Technical Field

[0001] This application belongs to the technical field of medical data processing, specifically relating to a method, system, electronic device, and storage medium for multimodal complaint conflict classification and doctor collaboration in internet hospitals. Background Technology

[0002] With the rapid development of internet hospitals and online consultation services, multimodal chief complaints (the main symptoms or signs described by the patient during their visit and their duration, which doctors usually use to make a primary diagnosis) collection technology has become an industry standard. Patients submit multi-source medical information, including text descriptions, voice, and images, through internet hospital terminals. After systematic recognition and analysis, this information is structured into chief complaint content for use by artificial intelligence systems or doctors for diagnostic reference. However, due to individual patient perception and knowledge differences, the chief complaint content often contains descriptions that conflict with medical expertise.

[0003] In existing technologies, multimodal complaint processing solutions mainly fall into three categories. The first is a single-modal processing solution, which only supports text or voice input and cannot cross-validate cross-modal medical information. The second is a rule-based multimodal fusion solution, which simply fuses the parsing results from different modalities according to set rules, but lacks a quantitative evaluation mechanism for medical conflicts in the fusion results. The third is a general multimodal large language model solution, which utilizes the multimodal understanding capabilities of a large-scale pre-trained model for complaint parsing, but lacks a dedicated mechanism for handling medical conflicts and often suffers from hallucinations in the generated content. For example, Chinese patent CN114386740A discloses a doctor-patient risk assessment method based on multimodal data, which focuses on conflict warning at the doctor-patient communication level and does not address conflict handling at the medical information level. Another example is the visual hallucination detection method based on evidence conflict or the multimodal large language model optimization method for adversarial modal conflict, which mainly addresses the model robustness problem, i.e., how to reduce hallucination outputs under multimodal input, rather than the safety decision-making and handling problems after medical conflicts occur.

[0004] Therefore, it is evident that existing technologies still have several technical shortcomings when dealing with multimodal complaint conflict handling scenarios in internet hospitals. Firstly, there is a lack of quantitative assessment of medical conflicts. Existing technologies primarily rely on qualitative judgments to determine multimodal information conflicts, failing to provide quantifiable severity scores at the field level based on medical ontology databases. This makes it difficult for the system to differentiate its handling based on the actual medical risk of the conflict. Secondly, the conditions for triggering manual intervention are vague. After detecting a conflict, existing technologies either silently ignore diagnostic suggestions directly output by AI or directly trigger an alarm requiring doctor intervention, without establishing a tiered triage mechanism based on conflict severity. This results in low-risk conflicts consuming excessive doctor resources, while high-risk conflicts fail to receive timely and mandatory handling. Thirdly, the information for doctor intervention is incomplete. In conflict scenarios requiring doctor intervention, existing technologies only present the original conflict data to the doctor, who must then reorganize and verify the items themselves, leading to low intervention efficiency and the potential for overlooking crucial information. Fourthly, there is a lack of auditable conflict handling logs. The conflict handling process in existing technologies does not generate structured log records, failing to meet the compliance requirements of internet hospital medical supervision for traceable and auditable high-risk AI decisions. Summary of the Invention

[0005] To address one or more of the aforementioned problems, this application provides a method for hierarchical processing and doctor collaboration in multimodal chief complaint conflicts in internet hospitals, a system for hierarchical processing and doctor collaboration in multimodal chief complaint conflicts in internet hospitals, an electronic device, and a storage medium to improve the efficiency of processing chief complaint conflicts.

[0006] The technical solution of this application is as follows.

[0007] The first aspect is a method for multimodal complaint conflict classification and doctor collaboration in internet hospitals, including: Receive multimodal chief complaint data from the patient, which includes at least text, voice, and images; Each modality in the multimodal chief complaint data is independently analyzed by calling the corresponding intelligent model, and the analysis results are merged to generate a structured chief complaint that includes multiple medical fields. Each medical field includes at least the field value, confidence level, and modality source. Perform field-level comparison checks on the structured complaint to identify any conflicting descriptions; Calculate the conflict severity score based on the results of the field-level comparison; Based on the conflict severity score and two preset grading thresholds, a three-level triage process is implemented. The three-level triage process includes: when the conflict severity score is less than the first grading threshold, automatic repair of the multimodal complaint data is triggered; when the conflict severity score is greater than or equal to the first grading threshold and less than the second grading threshold, physician collaborative assessment and processing of the multimodal complaint data is triggered; when the conflict severity score is greater than or equal to the second grading threshold, forced locking of the multimodal complaint data and an alarm are triggered. When triggering collaborative assessment and processing of multimodal complaint data by doctors, and / or triggering forced locking of multimodal complaint data and alarm, information on the corresponding conflict events, patient context, medical ontology, and evidence chain is extracted, and then a structured list of doctors to take over is generated.

[0008] Preferably, field-level comparison involves comparing the semantic consistency of field values ​​for each medical field under different modal sources, and identifying conflicting descriptions when semantics are inconsistent.

[0009] Preferably, the calculation of conflict severity score includes: For conflicting medical fields, determine the corresponding basic field weights, field importance correction factors, and conflict flags; Calculate the conflict severity score based on the field's base weight, field importance correction factor, and conflict flag.

[0010] Preferably, the generated physician takeover list includes: Extract conflict fields, modality sources, and confidence levels from conflict events; Retrieve basic patient information and medical history from the patient's context; Search for the corresponding conceptual relationships in the medical ontology database; Tracing back multimodal chief complaint data from the chain of evidence; Based on the rule template, check items and suggested actions are generated and sorted in descending order of conflict severity score.

[0011] Preferably, the process includes the following steps before the three-stage diversion process: The conflict level is determined based on field whitelist rules and patient stratification rules; The field whitelist rule is used to force an upgrade of the three-level triage process when a preset high-risk medical field is involved. The patient stratification rule is used to force an upgrade of the three-tier triage process when the patient is in an acute or acute exacerbation phase of a chronic disease.

[0012] Preferred automatic repair methods include: For medical fields with conflicting descriptions, a high-confidence fusion algorithm is used to automatically resolve the conflicts. The high-confidence fusion algorithm selects the corresponding optimal field value based on the confidence level of each modality.

[0013] Preferred options also include: Write the three-level traffic splitting process into a structured log; The structured log includes timestamps, patient identifiers, conflict events, conflict severity scores, automatic repair results, and physician takeover records.

[0014] The second aspect is an internet hospital multimodal complaint conflict classification and doctor collaboration system, including: The multimodal acquisition module is used to receive multimodal complaint data from the patient, which includes at least text, voice, and images. The multimodal parsing and fusion module is used to independently parse each modality in the multimodal complaint data by calling the corresponding intelligent model, and then fuse the parsing results to generate a structured complaint that includes multiple medical fields. Each medical field includes at least the field value, confidence level, and modality source. The conflict comparison module is used to perform field-level comparison checks on structured complaints to check for conflicting descriptions. The severity scoring module is used to calculate the conflict severity score based on the results of field-level comparisons. The triage module is used to perform three-level triage based on the conflict severity score and two preset grading thresholds. The three-level triage process includes: when the conflict severity score is less than the first grading threshold, triggering automatic repair of the multimodal complaint data; when the conflict severity score is greater than or equal to the first grading threshold and less than the second grading threshold, triggering collaborative assessment and processing of the multimodal complaint data by doctors; when the conflict severity score is greater than or equal to the second grading threshold, triggering forced locking of the multimodal complaint data and triggering an alarm. The list generation module is used to extract relevant conflict events, patient context, medical ontology, and evidence chain information when triggering physician collaborative assessment and processing of multimodal chief complaint data, and / or triggering forced locking of multimodal chief complaint data and alarm, and then generate a structured physician takeover list.

[0015] Thirdly, an electronic device includes a memory, a processor, and a computer program stored on the memory and running on the processor, wherein the processor, when executing the computer program, implements the first aspect's method for multimodal complaint conflict classification and doctor collaboration in internet hospitals.

[0016] Fourthly, a storage medium, which is a computer-readable storage medium, stores computer instructions that enable the computer to implement the first aspect of the Internet hospital's multimodal chief complaint conflict classification and doctor collaboration method.

[0017] Compared with the prior art, the advantages of this application are as follows: To improve the safety of medical diagnosis, enable quantitative assessment and differentiated handling of conflict severity, improve physician takeover efficiency and information integrity, enhance the traceability and compliance of the handling process, improve the handling efficiency of low-severity conflicts, and achieve refined adjustment of conflict levels. Attached Figure Description

[0018] Figure 1 This is a flowchart illustrating one of the methods for handling multimodal complaints and tiered processing of conflicts and coordinating physicians in internet hospitals, as described in this application.

[0019] Figure 2 This is a schematic diagram of the structural framework of one of the internet hospital multimodal complaint conflict classification and doctor collaboration systems in this application.

[0020] Figure 3 This is a schematic diagram of the frame of one of the electronic devices in this application. Detailed Implementation

[0021] Referring to the illustrations, the principles of this application are illustrated by way of example implementation in a suitable operating environment. The following description is based on the illustrative specific embodiments of this application and should not be construed as limiting other specific embodiments not detailed herein, such as corresponding adjustments to the order of steps of the method of this application based on the principles of the technical concept.

[0022] This application declares that it only performs tag generation processing on standardized medical data records that have been anonymized or de-identified in the business systems of medical institutions, and does not directly process sensitive data such as patients' personal identity information or privacy content in free text of medical records; the large language model involved is deployed on local computing nodes or medical private networks, and the medical-related knowledge graph and database are also located in a controlled network environment, which complies with the relevant provisions of the "Personal Information Protection Law of the People's Republic of China" and the "Data Security Law of the People's Republic of China" regarding de-identification, minimum necessary processing and data security.

[0023] like Figure 1 As shown in the figure, this application discloses a method for hierarchical processing of multimodal chief complaint conflicts and doctor collaboration in Internet hospitals, including the following steps.

[0024] S1. Receive multimodal complaint data from the patient. The multimodal complaint data includes at least text, voice, and images.

[0025] In this embodiment, patients submit multimodal complaint data through an internet hospital terminal (mobile APP, webpage, or mini-program), which includes at least three modalities: text description, voice expression, and images. Among them, images may include test reports, photos of the affected area, imaging examination reports, etc. After receiving the raw data of each modality, a unified timestamp is added to each piece of raw data to achieve temporal alignment of the multimodal complaint data, providing a temporal benchmark for subsequent cross-modal comparisons.

[0026] S2. For each modality in the multimodal complaint data, call the corresponding intelligent model to perform independent analysis, and merge the analysis results to generate a structured complaint that includes multiple medical fields. Each medical field includes at least the field value, confidence level, and modality source.

[0027] In this embodiment, different medical intelligent models are invoked for independent analysis of different modalities. Specifically, for the text modality, a large language model finely tuned for the clinical medical field is invoked to extract standardized symptom elements from the free text provided by the patient. For the speech modality, a dialect-robust speech recognition model is first invoked to transcribe the speech into text, and then a medical named entity recognition model is used to extract symptom elements. For the image modality, an optical character recognition model is invoked to extract text and indicator information from the test report, while a medical image classification model is invoked to identify abnormal regions and signs in the image.

[0028] After completing the independent parsing of each modality, all parsing results are semantically aligned and fused according to a unified medical field system to generate a structured chief complaint. In this embodiment, it is preferred to include 12 core medical fields, namely: chief complaint location, symptom type, severity, allergy history, medication history, family history, abnormal examination, vital signs, imaging findings, laboratory indicators, previous diagnosis, and current request.

[0029] S3. Perform a field-level comparison check on the structured complaint to identify any conflicting descriptions.

[0030] In this embodiment, field-level comparison is performed by comparing the semantic consistency of the field values ​​of each medical field under different modal sources. When the semantics are inconsistent, it indicates that there is a conflict description.

[0031] Optionally in this embodiment, a medical conflict matrix is ​​constructed to store the medical meaning of conflict descriptions of medical fields at corresponding severity levels. The medical conflict matrix is ​​a two-dimensional structure of 12 core medical fields and 5 severity levels. The basis for constructing the medical conflict matrix includes the conceptual relationships of the SNOMED-CT clinical medical ontology library, the ICD-10 disease classification standard, and the UpTo Date clinical evidence-based medicine knowledge base. It has been verified and corrected by clinical experts from tertiary hospitals through multiple rounds of Delphi method.

[0032] In this embodiment, optionally, during field-level comparison, all 12 medical fields of the structured chief complaint are traversed, and the semantic consistency of the field values ​​of each field under different source modalities is compared. When there is semantic inconsistency between the field values ​​of the same field across modalities, it is determined that there is a conflict description, and the severity and medical meaning corresponding to the conflict are matched from the medical conflict matrix. The severity level of field conflict is determined according to the clinical relevance. For example, the contradiction of the chief complaint site can be matched between levels that are negligible to life-threatening based on the clinical relevance of the diseased organ.

[0033] S4. Calculate the conflict severity score based on the results of the field-level comparison.

[0034] In this embodiment, calculating the conflict severity score includes: For conflicting medical fields, determine the corresponding basic field weights, field importance correction factors, and conflict flags; Calculate the conflict severity score based on the field's base weight, field importance correction factor, and conflict flag.

[0035] Optionally in this embodiment, for all conflicting medical fields, the conflict severity score is calculated using the following formula. : ; in, The field's basic weights range from 0 to 1. They are obtained through pre-training on the SNOMED-CT medical ontology library. The weight values ​​are determined by the analytic hierarchy process (AHP). The weights are scored independently by multiple physicians with associate chief physician titles or above from tertiary hospitals. After the consistency test shows that the consistency ratio (CR) is less than 0.1, the geometric mean is taken as the final weight. This represents the importance adjustment factor for the field, which is assigned a value based on the clinical risk level of the field. Typical assignments include, for example, chief complaint site = 1.0, symptom type = 0.9, severity = 1.2, and allergy history = 1.5. This indicates the conflict flag; it takes a value of 1 when a conflict exists and a value of 0 when there is no conflict. f This indicates which medical field is in conflict. F The total number of conflicting medical fields; the final conflict severity score. The value ranges from 0 to 10, with higher scores indicating higher medical safety risks associated with the conflict.

[0036] S5. Based on the conflict severity score and two preset grading thresholds, a three-level triage process is performed. The three-level triage process includes: when the conflict severity score is less than the first grading threshold, automatic repair of the multimodal complaint data is triggered; when the conflict severity score is greater than or equal to the first grading threshold and less than the second grading threshold, physician collaborative assessment and processing of the multimodal complaint data is triggered; when the conflict severity score is greater than or equal to the second grading threshold, forced locking of the multimodal complaint data and an alarm are triggered.

[0037] In this embodiment, before the three-level triage process, the following steps are included: determining the conflict level based on field whitelist rules and patient stratification rules; field whitelist rules are used to forcibly upgrade the three-level triage process when a preset high-risk medical field is involved; patient stratification rules are used to forcibly upgrade the three-level triage process when the patient is in an acute or acute exacerbation phase of a chronic disease.

[0038] In this embodiment, automatic repair includes: for medical fields with conflicting descriptions, a high-confidence fusion algorithm is used to automatically resolve the conflicts, and the high-confidence fusion algorithm selects the corresponding optimal field value according to the confidence of each modality.

[0039] S6. When triggering collaborative evaluation and processing of multimodal complaint data by doctors, and / or triggering forced locking of multimodal complaint data and alarm, extract the relevant conflict events, patient context, medical ontology, and evidence chain information, and then generate a structured doctor takeover list.

[0040] In this embodiment, generating the doctor takeover list includes: extracting conflict fields, modality sources, and confidence levels from conflict events; querying patient basic information and past medical history from the patient context; searching for corresponding conceptual relationships from the medical ontology; tracing back multimodal chief complaint data from the evidence chain; and generating verification items and suggested actions based on rule templates, sorted in descending order of conflict severity scores.

[0041] S7. Write the three-level triage process into a structured log; the structured log includes timestamps, patient identifiers, conflict events, conflict severity scores, automatic repair results, and physician takeover records.

[0042] In this embodiment, the structured log meets the regulatory requirements of the "Internet Diagnosis and Treatment Management Measures" and the "Guidelines for Ethical Review of Medical Artificial Intelligence Algorithms" regarding the traceability and auditability of high-risk artificial intelligence decisions, and supports post-event compliance audits and medical quality reviews.

[0043] Compared with the prior art, the advantages of this application are as follows: To improve the safety of medical diagnosis, this application performs field-level comparison checks on the structured complaints generated by independently parsing and fusing the various modalities in the multimodal complaint data to check for conflicts, thus achieving cross-validation of cross-modal medical information. By calculating the conflict severity score and performing three-level triage, conflict descriptions in the complaint content can be identified in a timely manner and trigger mandatory locking and alarms, avoiding the silent output of diagnostic suggestions while ignoring high-risk conflicts, thereby reducing medical safety risks. This application achieves quantitative assessment and differentiated handling of conflict severity by calculating conflict severity scores for conflicting medical fields. Through a three-level triage process, differentiated handling strategies are triggered for conflicts of different severity levels to avoid low-risk conflicts from excessively consuming physician resources, while ensuring that high-risk conflicts are prioritized for handling. This application improves the efficiency and completeness of physician takeovers by creating a structured physician takeover checklist. This eliminates the need for physicians to reorganize multimodal data themselves, allowing them to directly check each item on the checklist, thus improving takeover efficiency and reducing the risk of omissions. Enhance the traceability and compliance of the treatment process; this application writes the three-level triage process into a structured log to form a complete and traceable chain of evidence, meeting the auditable requirements of Internet hospital diagnosis and treatment supervision for high-risk artificial intelligence decision-making; Improve the efficiency of handling low-severity conflicts; reduce unnecessary intervention by doctors through a three-tiered triage system, thereby improving the overall efficiency of online consultations at internet hospitals; This application enables refined adjustment of conflict levels; it uses a whitelist to forcibly escalate high-risk conflicts, achieving refined conflict adjustment based on medical risks and individual patient characteristics, thereby further ensuring medical safety.

[0044] like Figure 2 As shown in the figure, this application discloses a multimodal chief complaint conflict classification and doctor collaboration system for internet hospitals, including: The multimodal acquisition module is used to receive multimodal complaint data from the patient, which includes at least text, voice, and images. The multimodal parsing and fusion module is used to independently parse each modality in the multimodal complaint data by calling the corresponding intelligent model, and then fuse the parsing results to generate a structured complaint that includes multiple medical fields. Each medical field includes at least the field value, confidence level, and modality source. The conflict comparison module is used to perform field-level comparison checks on structured complaints to check for conflicting descriptions. The severity scoring module is used to calculate the conflict severity score based on the results of field-level comparisons. The triage module is used to perform three-level triage based on the conflict severity score and two preset grading thresholds. The three-level triage process includes: when the conflict severity score is less than the first grading threshold, triggering automatic repair of the multimodal complaint data; when the conflict severity score is greater than or equal to the first grading threshold and less than the second grading threshold, triggering collaborative assessment and processing of the multimodal complaint data by doctors; when the conflict severity score is greater than or equal to the second grading threshold, triggering forced locking of the multimodal complaint data and triggering an alarm. The list generation module is used to extract relevant conflict events, patient context, medical ontology, and evidence chain information when triggering physician collaborative assessment and processing of multimodal chief complaint data, and / or triggering forced locking of multimodal chief complaint data and alarm, and then generate a structured physician takeover list.

[0045] The log module is used to write the three-level triage process into a structured log; the structured log includes timestamps, patient identifiers, conflict events, conflict severity scores, automatic repair results, and doctor takeover records.

[0046] The system described in the above embodiments is used to execute the corresponding Internet hospital multimodal chief complaint conflict classification and doctor collaboration methods in the foregoing embodiments, and has the beneficial effects of the corresponding method embodiments, which will not be repeated here.

[0047] like Figure 3 As shown, based on the same application concept, and corresponding to the methods of any of the above embodiments, this application also discloses an electronic device, including a memory, a processor, and a computer program stored on the memory and running on the processor. When the processor executes the computer program, it implements the above-mentioned method for multimodal complaint conflict classification processing and doctor collaboration in Internet hospitals.

[0048] Specifically, the device includes: a processor 1010, a memory 1020, an input / output interface 1030, a communication interface 1040, and a bus 1050. The processor 1010, memory 1020, input / output interface 1030, and communication interface 1040 are interconnected within the device via the bus 1050.

[0049] The processor 1010 can be implemented using a general-purpose CPU (Central Processing Unit), microprocessor, application-specific integrated circuit (ASIC), GPU (Graphics Processing Unit), or one or more integrated circuits, to implement relevant programs and achieve the technical solutions provided in the embodiments of this specification.

[0050] The memory 1020 can be implemented in the form of ROM (Read Only Memory), RAM (Random Access Memory), static storage device, dynamic storage device, etc. The memory 1020 can store the operating system and other application programs. When the technical solutions provided in the embodiments of this specification are implemented through software or firmware, the relevant program code is stored in the memory 1020 and called by the processor 1010. The input / output interface 1030 is used to connect input / output modules to realize information input and output. Input / output modules can be configured as components in the device (not shown in the figure) or externally connected to the device to provide corresponding functions. Input devices can include keyboards, mice, touch screens, microphones, various sensors, etc., and output devices can include displays, projectors, speakers, vibrators, indicator lights, etc.

[0051] The communication interface 1040 is used to connect a communication module (not shown in the figure) to enable communication between this device and other devices. The communication module can communicate via wired means (such as USB (Universal Serial Bus), network cable, etc.) or wireless means (such as mobile network, WIFI (Wireless Fidelity), Bluetooth, etc.).

[0052] Bus 1050 includes a pathway for transmitting information between various components of the device, such as processor 1010, memory 1020, input / output interface 1030, and communication interface 1040.

[0053] It should be noted that although the above-described device only shows the processor 1010, memory 1020, input / output interface 1030, communication interface 1040, and bus 1050, in specific implementations, the device may also include other components necessary for normal operation. Furthermore, those skilled in the art will understand that the above-described device may include only the components necessary for implementing the embodiments of this specification, and not necessarily all the components shown in the figures.

[0054] The electronic devices described in the above embodiments are used to implement the corresponding Internet hospital multimodal chief complaint conflict classification and doctor collaboration methods in any of the foregoing embodiments, and have the beneficial effects of the corresponding method embodiments, which will not be repeated here.

[0055] Based on the same application concept, and corresponding to any of the above embodiments, this application also discloses a computer-readable storage medium that stores computer instructions for enabling a computer to implement the above-described method for multimodal complaint conflict classification and doctor collaboration in internet hospitals.

[0056] The computer-readable storage medium of this embodiment includes permanent and non-permanent, removable and non-removable media, and information storage can be implemented by any method or technology. Information can be computer-readable instructions, data structures, program modules, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic magnetic disk storage or other magnetic storage devices, or any other non-transfer medium, which can be used to store information accessible by a computing device. The computer instructions stored in the storage medium of the above embodiments are used to enable the computer to implement the multimodal complaint conflict classification and doctor collaboration method for Internet hospitals as described in any of the above embodiments, and have the beneficial effects of the corresponding method embodiments, which will not be repeated here.

[0057] The above description is merely a preferred embodiment and the technical principles employed in this application. This application is not limited to the specific embodiments or combinations thereof, and various obvious changes, readjustments, and substitutions that can be made by those skilled in the art will not depart from the scope of protection of this application. Therefore, although this application has been described in detail through the above embodiments, this application is not limited to the above embodiments, and may include more other equivalent embodiments without departing from the concept of this application, the scope of which is determined by the scope of the claims.

Claims

1. A method for hierarchical handling of multimodal chief complaint conflicts and doctor collaboration in internet hospitals, characterized in that, include: Receive multimodal complaint data from the patient, wherein the multimodal complaint data includes at least text, voice, and images; Each modality in the multimodal complaint data is independently analyzed by calling the corresponding intelligent model, and the analysis results are merged to generate a structured complaint including multiple medical fields. Each medical field includes at least a field value, confidence level, and modality source. Perform a field-level comparison check on the structured complaint to identify any conflicting descriptions. Calculate the conflict severity score based on the results of the field-level comparison; Based on the conflict severity score and two preset grading thresholds, a three-level triage process is performed. The three-level triage process includes: when the conflict severity score is less than the first grading threshold, automatic repair of the multimodal complaint data is triggered; when the conflict severity score is greater than or equal to the first grading threshold and less than the second grading threshold, physician collaborative assessment and processing of the multimodal complaint data is triggered; when the conflict severity score is greater than or equal to the second grading threshold, forced locking of the multimodal complaint data and an alarm are triggered. When the physician collaborative assessment of multimodal complaint data is triggered, and / or when the forced locking of multimodal complaint data and alarm are triggered, the relevant conflict events, patient context, medical ontology, and evidence chain information are extracted, and then a structured physician takeover list is generated.

2. The method for multimodal chief complaint conflict classification and doctor collaboration in internet hospitals according to claim 1, characterized in that, Field-level comparison involves comparing the semantic consistency of the field values ​​for each medical field under different modal sources, and identifying conflicting descriptions when semantics are inconsistent.

3. The method for multimodal chief complaint conflict classification and doctor collaboration in internet hospitals according to claim 1, characterized in that, The calculation of conflict severity score includes: For conflicting medical fields, determine the corresponding basic field weights, field importance correction factors, and conflict flags; The conflict severity score is calculated based on the field's base weight, the field's importance correction factor, and the conflict flag.

4. The method for multimodal chief complaint conflict classification and doctor collaboration in internet hospitals according to claim 1, characterized in that, Generating the doctor takeover list includes: Extract the conflict field, modality source, and confidence level from the conflict event; Retrieve basic patient information and medical history from the patient's context; Search for the corresponding conceptual relationships in the medical ontology database; Tracing back the multimodal chief complaint data from the aforementioned chain of evidence; Based on the rule template, check items and suggested actions are generated and sorted in descending order of conflict severity score.

5. The method for multimodal chief complaint conflict classification and doctor collaboration in internet hospitals according to claim 1, characterized in that, Prior to the three-level traffic splitting process, the following is also included: The conflict level is determined based on field whitelist rules and patient stratification rules; The field whitelist rule is used to force an upgrade of the three-level triage process when a preset high-risk medical field is involved. The patient stratification rule is used to forcibly upgrade the three-tier triage process when the patient is in an acute or acute exacerbation phase of a chronic disease.

6. The method for multimodal chief complaint conflict classification and doctor collaboration in internet hospitals according to claim 1, characterized in that, The automatic repair includes: For medical fields with conflicting descriptions, a high-confidence fusion algorithm is used to automatically resolve the conflicts. The high-confidence fusion algorithm selects the corresponding optimal field value based on the confidence level of each modality.

7. The method for multimodal chief complaint conflict classification and doctor collaboration in internet hospitals according to claim 1, characterized in that, Also includes: Write the three-level traffic splitting process into a structured log; The structured log includes timestamps, patient identifiers, conflict events, conflict severity scores, automatic repair results, and doctor takeover records.

8. A multimodal complaint conflict classification and doctor collaboration system for internet hospitals, characterized in that, include: A multimodal acquisition module is used to receive multimodal complaint data from the patient, wherein the multimodal complaint data includes at least text, voice, and images; The multimodal parsing and fusion module is used to independently parse each modality in the multimodal complaint data by calling the corresponding intelligent model, and then fuse the parsing results to generate a structured complaint including multiple medical fields. Each medical field includes at least a field value, confidence level, and modality source. The conflict comparison module is used to perform field-level comparison checks on the structured complaint to check for conflicting descriptions. The severity scoring module is used to calculate the conflict severity score based on the results of the field-level comparison. The triage module is used to perform three-level triage processing based on the conflict severity score and two preset grading thresholds. The three-level triage processing includes: when the conflict severity score is less than the first grading threshold, triggering automatic repair of the multimodal complaint data; when the conflict severity score is greater than or equal to the first grading threshold and less than the second grading threshold, triggering collaborative assessment and processing of the multimodal complaint data by doctors; when the conflict severity score is greater than or equal to the second grading threshold, triggering forced locking of the multimodal complaint data and triggering an alarm. The list generation module is used to extract the corresponding conflict events, patient context, medical ontology, and evidence chain information when triggering the doctor collaborative assessment and processing of multimodal chief complaint data, and / or triggering the forced locking of multimodal chief complaint data and alarm, and then generate a structured doctor takeover list.

9. An electronic device, characterized in that, It includes a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, it implements the Internet hospital multimodal chief complaint conflict classification processing and doctor collaboration method according to any one of claims 1-7.

10. A storage medium, which is a computer-readable storage medium, characterized in that, The computer contains computer instructions that enable the computer to implement the method for multimodal complaint conflict classification and doctor collaboration in Internet hospitals as described in any one of claims 1-7.

Images