Remote consultation method and system based on artificial intelligence hospital guidance

Abstract

The invention discloses a remote consultation method and system based on artificial intelligence hospital guidance, and belongs to the technical field of medical informatization. The method comprises the following steps: acquiring multi-modal identity verification data of a patient, completing identity recognition, and associating with an electronic health record; patient symptom description is collected, semantic understanding and reasoning are carried out based on the dynamically updated multi-source knowledge base and the large language model, and a department recommendation result is generated; a remote doctor terminal is matched according to the recommendation result, and a multi-mode consultation channel is established; and after the consultation is completed, generating prescription data, triggering a medical insurance payment interface to complete expense settlement, and sending the prescription data to a medicine distribution system to realize medicine distribution. The system comprises an identity recognition module, an intelligent hospital guide module, a remote consultation module, a payment and distribution module and a data interface module. According to the method, the process integration of hospital guide and remote consultation is realized, the hospital guide accuracy and the doctor seeing efficiency are improved, and the method is suitable for basic medical and remote medical scenes.

Description

Technical Field

[0001] This invention relates to the field of medical information technology, and more specifically, to a remote consultation method and system based on artificial intelligence-guided diagnosis. Background Technology

[0002] Existing medical triage and telemedicine systems generally employ a combination of "rule tables + manual triage" or "static knowledge base + video communication." The triage side relies on symptom keyword-department lookup tables or simple decision trees, while the telemedicine side primarily uses independently deployed video conferencing systems. Neither system forms a closed loop in terms of data, processes, or business operations. Rule-based triage systems periodically import disease-symptom-department mapping relationships in batches from paper guides or localized databases, with update cycles often quarterly or even annually. This fails to promptly incorporate rapidly evolving medical evidence, public health emergency warnings, or information on new drugs / technologies, leading to significant discrepancies between triage results for rare diseases, cross-system symptoms, and emerging infectious diseases and clinical reality. Clinical statistics show that the error rate of such systems is generally between 15% and 30%, requiring patients to re-register or transfer to different departments multiple times, increasing the workload of specialists.

[0003] In terms of process architecture, traditional solutions decouple registration, triage, consultation, payment, prescription, and medication dispensing into heterogeneous subsystems. Patients must repeatedly enter basic information, queue again, or switch terminals each time they enter the next step, increasing the average consultation time by more than 40% compared to an integrated process. For elderly people with lower digital literacy, frequent interface switching can easily cause process interruptions and reduce service accessibility.

[0004] In terms of personalization, existing systems mostly provide one-time recommendations based on a single symptom input, lacking dynamic integration of multi-dimensional data such as the patient's medical history, long-term medication, allergy information, age, and complications. This results in weak generalization of triage results and a high likelihood of patients being referred to the wrong department. Technically, keyword matching or single-layer decision trees cannot handle mixed text, voice, and image inputs. They also lack sufficient depth in identifying latent symptoms (such as whether chest tightness radiates to the left shoulder) or overlapping symptoms (such as chest pain involving cardiology, respiratory, and gastroenterology departments), requiring secondary manual questioning to complete triage.

[0005] While remote consultation systems have been deployed in many medical institutions, they generally focus on video communication, file transfer, and electronic whiteboard functions, physically isolated from the triage process. Patients must select a department before initiating a remote consultation; if the initial selection is incorrect, they must return to the queue, resulting in redundant processes. Because the triage results lack a standardized data structure, doctors on the remote consultation side cannot directly reuse the symptom semantic information from the triage process, requiring them to inquire about the patient's medical history again, extending online time and consuming doctor resources. In summary, traditional solutions have significant shortcomings in terms of knowledge timeliness, process coherence, personalized reasoning, and multimodal interaction, making it difficult to meet the actual needs of primary healthcare institutions for accurate triage, remote collaboration, and integrated end-to-end services. Summary of the Invention

[0006] The present invention aims to solve at least one of the aforementioned technical problems existing in the prior art.

[0007] Therefore, the first aspect of the present invention provides a remote consultation method based on artificial intelligence-guided diagnosis.

[0008] A second aspect of the present invention provides a remote consultation system based on artificial intelligence-guided diagnosis.

[0009] This invention provides a remote consultation method based on artificial intelligence-guided diagnosis, comprising: The system acquires patient identity verification data, which is any one or more of multimodal identity information; identifies the patient's identity based on the identity verification data; and associates the multimodal identity information with the patient's electronic health record. Acquire patient symptom description data, perform semantic understanding and reasoning on patient symptom descriptions based on multi-source knowledge base and large language model, and generate department recommendation results; Based on the departmental recommendations, match and connect the corresponding remote doctor terminals to establish a multimodal remote consultation channel; After the consultation is completed, prescription data is generated, and the medical insurance payment interface is triggered to settle the expenses; The prescription data is sent to the drug delivery system, and the delivery status is tracked.

[0010] The remote consultation method based on artificial intelligence-guided diagnosis according to the above-described technical solution of the present invention may further have the following additional technical features: In the above technical solution, the types of multimodal identity information include ID cards, electronic health cards, and facial recognition.

[0011] In the above technical solution, the multi-source knowledge base includes medical literature, clinical guidelines, historical cases and reflection data on erroneous cases, and is stored and updated through vectorized encoding.

[0012] In the above technical solution, the large language model clarifies symptom details through multi-turn dialogue and outputs department recommendation results in conjunction with a retrieval enhancement generation mechanism.

[0013] In the above technical solution, the combined retrieval enhancement generation mechanism includes: using a retrieval enhancement generation framework to retrieve similar cases from the multi-source knowledge base to assist large model reasoning.

[0014] This invention provides a remote consultation system based on artificial intelligence-guided diagnosis, applicable to the remote consultation method based on artificial intelligence-guided diagnosis as described in any of the above technical solutions. The system includes: The identity verification module is used to acquire and verify the patient's identity verification data and link it to the electronic health record; The intelligent triage module is used to perform semantic parsing of symptom descriptions and recommend departments based on a multi-source knowledge base and a large language model; The remote consultation module is used to match doctor terminals based on departmental recommendations and establish a multimodal remote consultation channel. The payment and delivery module is used to receive prescription data, call the medical insurance payment interface, and trigger the drug delivery process; The data interface module is used to enable data interaction with hospital information systems, medical insurance platforms, and drug delivery systems.

[0015] In the above technical solution, the intelligent triage module includes: The dynamic knowledge base update unit is used to periodically ingest external medical data and store it in a multi-source knowledge base through BGE model vectorization. A multimodal dialogue engine is used to parse text and voice input, analyze patient symptom descriptions, combine historical medical records to generate personalized queries, and output accurate department recommendations. The decision support unit is used to obtain similar cases, historical experience cases, and error reflection knowledge from multi-source knowledge bases to enhance the reasoning context and assist reasoning with the search results.

[0016] In the above technical solution, the remote consultation module supports video and text communication methods, and integrates an electronic medical record writing tool for recording consultation content.

[0017] In the above technical solution, the payment and delivery module includes: The medical insurance interface call unit is used to calculate the reimbursement ratio; The drug delivery interface unit is used to generate delivery orders and provide feedback on logistics status.

[0018] In the above technical solution, the system architecture includes: The user terminal layer provides the access interface for patients and doctors. The data processing layer is used to realize identity recognition, intelligent triage, remote consultation, payment processing and delivery control; The data storage layer is used to store electronic health records, medical knowledge bases, prescriptions, and consultation records. The support service layer is used to provide interface services for hospital information systems, medical insurance platforms, payment gateways, and logistics systems.

[0019] In summary, due to the adoption of the above-mentioned technical features, the beneficial effects of the present invention are: This invention achieves dynamic synchronization between triage knowledge and clinical practice, avoiding misdiagnosis due to outdated knowledge. It seamlessly integrates triage, consultation, prescription, payment, and delivery processes, effectively reducing patient steps. Based on multi-source data (such as historical medical records, real-time dialogue, and medical knowledge bases), it enhances the personalization and accuracy of triage, and simplifies the department recommendation process for complex symptoms through artificial intelligence technology.

[0020] Specifically, this invention utilizes a dynamic knowledge update mechanism to write medical literature, clinical guidelines, and institutional medical records into a vector knowledge base at set intervals. This eliminates the reliance on quarterly or annual batch imports for triage rules, thereby removing the knowledge lag phenomenon mentioned in the background technology. The system sequentially triggers five stages—identity recognition, intelligent triage, remote consultation, medical insurance settlement, and drug delivery—through the same data interface. Patients no longer need to repeatedly enter basic information between the registration, triage, and payment subsystems, solving the problems of multiple system switching and repeated queuing in existing technologies. By inputting the current symptoms collected during the conversation along with the associated electronic health records into a large language model, and constraining the reasoning process by past medical history, allergy records, and age-related complications, the output department recommendations no longer rely solely on single keywords, reducing the probability of incorrect department registration. The dialogue engine simultaneously receives text and voice input and clarifies latent symptoms through multiple rounds of questioning, enabling initial screening of overlapping symptoms such as chest pain and dizziness at the triage stage, replacing the traditional keyword matching method that cannot handle complex descriptions. The triage results are transmitted to the doctor's terminal in structured fields. The doctor can directly view the semantics of symptoms and the basis for initial screening in the video or text consultation interface without having to collect medical history again. This improves the redundant process in the existing technology where triage and remote consultation are physically isolated and require returning to the department for reselection.

[0021] Additional aspects and advantages of the invention will become apparent in the following description or may be learned by practice of the invention. Attached Figure Description

[0022] The above and / or additional aspects and advantages of the present invention will become apparent and readily understood from the description of the embodiments taken in conjunction with the following drawings, in which: Figure 1 This is a flowchart of a remote consultation method based on artificial intelligence-guided diagnosis according to an embodiment of the present invention; Figure 2 This is an architecture diagram of a remote consultation system based on artificial intelligence-guided diagnosis, according to an embodiment of the present invention. Detailed Implementation

[0023] To better understand the above-mentioned objectives, features, and advantages of the present invention, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be noted that, unless otherwise specified, the embodiments and features described in these embodiments can be combined with each other.

[0024] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and therefore the scope of protection of the invention is not limited to the specific embodiments disclosed below.

[0025] The following reference Figures 1 to 2 This describes a remote consultation method based on artificial intelligence-guided diagnosis, provided according to some embodiments of the present invention.

[0026] Some embodiments of this application provide a remote consultation method based on artificial intelligence-guided diagnosis.

[0027] like Figure 1 As shown, the first embodiment of the present invention proposes a remote consultation method based on artificial intelligence-guided diagnosis, including the following steps S1 to S5.

[0028] S1. Obtain the patient's identity verification data, which is any one or more of multimodal identity information. Identify the patient's identity based on the identity verification data and associate the multimodal identity information with the patient's electronic health record.

[0029] In some embodiments, the types of multimodal identity information include ID cards, electronic health cards, and facial recognition. Patients can verify their identity using any of these methods, and the verification is automatically linked to the patient's electronic health record (EHR). This step effectively reduces the need for duplicate information entry.

[0030] S2. Obtain patient symptom description data, and based on a multi-source knowledge base and a large language model, perform semantic understanding and reasoning on the patient's symptom description to generate departmental recommendation results.

[0031] In some embodiments, the multi-source knowledge base includes medical literature, clinical guidelines, historical case data, and reflections on erroneous cases, and is stored and updated using vectorized encoding.

[0032] Specifically, the multi-source knowledge base is a dynamic knowledge base that integrates knowledge from multiple sources, including electronic health records, medical literature, and real-time clinical data. It is constructed using vectorized encoding (such as the BGE model) to create an updatable knowledge base. Through a dynamic knowledge management mechanism, it regularly updates external medical knowledge (such as guidelines and literature), experiential knowledge (successful referral cases), and reflective knowledge (reviews of erroneous cases), thereby improving the model's adaptability.

[0033] In some embodiments, the large language model clarifies symptom details through multi-turn dialogue and outputs departmental recommendations in conjunction with a retrieval-enhanced generation mechanism. LLM is employed for symptom semantic understanding and reasoning, supporting multi-turn dialogue to clarify latent symptoms, such as differentiating esophageal diseases from heart diseases by asking "Is the pain related to swallowing?"

[0034] Specifically, the combined retrieval enhancement generation mechanism includes: using a retrieval enhancement generation framework to retrieve similar cases from the multi-source knowledge base to assist large model reasoning.

[0035] Based on Large Language Model (LLM) and enhanced retrieval technology, the system parses patient symptom descriptions (text / voice) and combines them with historical medical records obtained from electronic health records to generate personalized queries and output accurate department recommendations.

[0036] S3. Based on the departmental recommendations, match and connect the corresponding remote doctor terminals to establish a multimodal remote consultation channel.

[0037] Among these features, the system automatically recommends online doctors from the Hospital Information System (HIS) based on the triage results, supporting multimodal consultations including video and text-based consultations. Doctors can integrate electronic medical record writing tools to record consultation content in real time and generate suggestions or prescriptions.

[0038] S4. After the consultation is completed, prescription data is generated and the medical insurance payment interface is triggered to settle the expenses. When settling expenses, the patient's co-payment ratio can be automatically calculated and online payment is supported.

[0039] S5. Send the prescription data to the drug delivery system and track the delivery status.

[0040] Specifically, after the prescription is approved, patients can choose to have the medication delivered by courier, or they can go directly to the pharmacy to purchase the medication with the prescription. When choosing to have the medication delivered by courier, the system tracks the delivery status and pushes the information to the patient for delivery status inquiry.

[0041] Other embodiments of the present invention provide a remote consultation system based on artificial intelligence-guided diagnosis, which is applied to the remote consultation method based on artificial intelligence-guided diagnosis as described in any of the above embodiments. The system includes: an identity recognition module, an intelligent diagnosis-guided diagnosis module, a remote consultation module, a payment and delivery module, and a data interface module.

[0042] The identity verification module is used to acquire and verify the patient's identity verification data and associate the patient's identity verification data with the corresponding electronic health record.

[0043] Specifically, the identity recognition module can recognize ID card information using OCR technology or verify identity by calling a facial recognition API. Alternatively, the identity recognition module can be configured to support fingerprint or iris authentication to enhance security. The electronic health card interface connects to the regional healthcare platform to obtain patients' historical medical records.

[0044] The intelligent triage module is used to perform semantic parsing of symptom descriptions and recommend departments based on a multi-source knowledge base and a large language model.

[0045] In some embodiments, the intelligent triage module includes: a dynamic knowledge base update unit, a multimodal dialogue engine, and a decision support unit.

[0046] The dynamic knowledge base update unit is used to periodically ingest external medical data and store it in a multi-source knowledge base through vectorization using the BGE model. The external medical data refers to the updated data of the aforementioned medical literature, clinical guidelines, historical cases, and error case reflection data. It can be updated weekly or monthly according to actual needs, or it can be flexibly configured according to other specified cycles.

[0047] A multimodal dialogue engine is used to parse text and voice input, analyze patient symptom descriptions, combine historical medical records to generate personalized queries, and output accurate department recommendations.

[0048] Specifically, the multimodal dialogue engine uses the LangChain framework to integrate LLM (such as a dedicated medical model), and the initialization prompt includes a task description and a list of candidate departments.

[0049] The decision support unit is used to obtain similar cases, historical experience cases, and error reflection knowledge from multi-source knowledge bases to enhance the reasoning context and assist reasoning with the search results.

[0050] In practice, a three-pronged approach is used: retrieving similar cases, historical experience cases, and error reflection knowledge from the knowledge base to enhance the reasoning context.

[0051] Example: A patient inputs "stomach pain after meals," and the model supplements the information with relevant knowledge about "Helicobacter pylori infection" and recommends a gastroenterology department.

[0052] The remote consultation module is used to match doctor terminals based on departmental recommendations and establish a multimodal remote consultation channel.

[0053] In some embodiments, the remote consultation module supports video and text communication methods and integrates an electronic medical record writing tool for recording consultation content.

[0054] In one specific embodiment, the remote consultation module integrates WebRTC technology to enable video consultations. An electronic medical record template is embedded on the doctor's end, supporting voice-to-text recording. After a doctor issues a prescription, an automatic review process is triggered, and compliant prescriptions are encrypted and transmitted to the pharmacy.

[0055] The payment and delivery module receives prescription data, calls the medical insurance payment interface, and triggers the drug delivery process. The medical insurance payment interface connects to medical insurance payment services, calculates the reimbursement rate, and generates an order. Third-party delivery services are called via a logistics API, allowing patients to track the delivery progress in real time.

[0056] In one specific embodiment, the payment and delivery module includes: The medical insurance interface call unit is used to calculate the reimbursement ratio and generate payment orders; The drug delivery interface unit is used to generate delivery orders and provide feedback on logistics status.

[0057] The data interface module is used to enable data interaction with hospital information systems, medical insurance platforms, and drug delivery systems. Specifically, based on the data interface module, each module connects with the hospital's HIS, payment system, and logistics platform through API interfaces to ensure seamless data flow.

[0058] In one specific embodiment, the operation process of each module of the disclosed system is illustrated using a patient with "dizziness accompanied by tinnitus" as an example: (1) After identity verification, the system retrieves the patient's medical history (such as previous history of hypertension).

[0059] (2) The intelligent triage module confirms the details of symptoms (such as the duration of tinnitus and the frequency of dizziness) through multiple rounds of dialogue, and recommends "otolaryngology" or "neurology" in combination with the knowledge base.

[0060] (3) The remote consultation module connects to the corresponding department doctors to provide remote diagnosis and prescribe examinations or prescriptions for patients.

[0061] (4) The payment and delivery module automatically calculates the medical insurance reimbursement ratio and pushes the generated order to the patient for payment. The patient selects medication delivery to complete the closed loop.

[0062] In another specific embodiment, the operation process of the disclosed system is illustrated using Mr. Wang (55 years old, with a history of hypertension), a community resident, as an example: (1) Identity recognition: Mr. Wang authenticates by scanning his face, and the system synchronizes his health record.

[0063] (2) Intelligent triage: Mr. Wang described "recent chest tightness and dizziness". The model confirmed the correlation between the symptoms and exercise through dialogue. The search of historical medical records revealed a history of hypertension, and the cardiology department was recommended as the priority department based on the knowledge base.

[0064] (3) Remote consultation: The system recommends Dr. Li from the Department of Cardiology. After the video consultation, the initial diagnosis requires an electrocardiogram examination, and an examination order and antihypertensive drugs are prescribed.

[0065] (4) Payment and delivery: Mr. Wang chose medical insurance payment, the examination form was electronically sent to the cooperating hospital, and the medicine was delivered to his home.

[0066] In some embodiments, such as Figure 2 As shown, the system architecture includes: The user terminal layer provides access interfaces for patients and doctors; specifically, it provides access points for patient-side apps / terminals and doctor-side workstations.

[0067] The data processing layer is used to realize identity recognition, intelligent triage, remote consultation, payment processing and delivery control; it can also be called the core module layer, which is used to process the five key links of identity recognition, intelligent triage (including LLM, multi-source knowledge base, dynamic enhancement and other modules), remote consultation, automatic deduction and drug delivery in sequence according to the business process.

[0068] The data storage layer is used to store electronic health records, medical knowledge bases, prescriptions and consultation records; specifically, the data storage layer stores business data including patient electronic medical records (EHRs), dynamic medical knowledge (disease databases, drug databases, etc.), prescription data, triage and consultation data, and payment and delivery data.

[0069] The support service layer is used to provide interface services for hospital information systems, medical insurance platforms, payment gateways, and logistics systems.

[0070] In this specification, the illustrative expressions of the terms used do not necessarily refer to the same embodiments or examples. Moreover, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0071] Any modifications, equivalent substitutions, or improvements made within the spirit and principles of this invention shall be included within the scope of protection of this invention.

Claims

1. A remote consultation method based on artificial intelligence-guided diagnosis, characterized in that, include: The system acquires patient identity verification data, which is any one or more of multimodal identity information; identifies the patient's identity based on the identity verification data; and associates the multimodal identity information with the patient's electronic health record. Acquire patient symptom description data, perform semantic understanding and reasoning on patient symptom descriptions based on multi-source knowledge base and large language model, and generate department recommendation results; Based on the departmental recommendations, match and connect the corresponding remote doctor terminals to establish a multimodal remote consultation channel; After the consultation is completed, prescription data is generated, and the medical insurance payment interface is triggered to settle the expenses; The prescription data is sent to the drug delivery system, and the delivery status is tracked.

2. The remote consultation method based on artificial intelligence-guided diagnosis according to claim 1, characterized in that, The types of multimodal identity information include ID cards, electronic health cards, and facial recognition.

3. The remote consultation method based on artificial intelligence-guided diagnosis according to claim 1, characterized in that, The multi-source knowledge base includes medical literature, clinical guidelines, historical cases, and reflections on erroneous cases, and is stored and updated using vectorized encoding.

4. The remote consultation method based on artificial intelligence-guided diagnosis according to claim 1, characterized in that, The large language model clarifies symptom details through multi-turn dialogue and outputs departmental recommendation results in conjunction with a retrieval enhancement generation mechanism.

5. The remote consultation method based on artificial intelligence-guided diagnosis according to claim 4, characterized in that, The combined retrieval-enhanced generation mechanism includes: employing a retrieval-enhanced generation framework to retrieve similar cases from the multi-source knowledge base to assist in large-scale model reasoning.

6. A remote consultation system based on artificial intelligence-guided diagnosis, characterized in that, The system, applied to the remote consultation method based on artificial intelligence triage as described in any one of claims 1 to 5, comprises: The identity verification module is used to acquire and verify the patient's identity verification data and link it to the electronic health record; The intelligent triage module is used to perform semantic parsing of symptom descriptions and recommend departments based on a multi-source knowledge base and a large language model; The remote consultation module is used to match doctor terminals based on departmental recommendations and establish a multimodal remote consultation channel. The payment and delivery module is used to receive prescription data, call the medical insurance payment interface, and trigger the drug delivery process; The data interface module is used to enable data interaction with hospital information systems, medical insurance platforms, and drug delivery systems.

7. The remote consultation system based on artificial intelligence triage according to claim 6, characterized in that, The intelligent triage module includes: The dynamic knowledge base update unit is used to periodically ingest external medical data and store it in a multi-source knowledge base through BGE model vectorization. A multimodal dialogue engine is used to parse text and voice input, analyze patient symptom descriptions, combine historical medical records to generate personalized queries, and output accurate department recommendations. The decision support unit is used to obtain similar cases, historical experience cases, and error reflection knowledge from multi-source knowledge bases to enhance the reasoning context and assist reasoning with the search results.

8. The remote consultation system based on artificial intelligence triage according to claim 6, characterized in that, The remote consultation module supports video and text communication and integrates an electronic medical record writing tool for recording consultation content.

9. The remote consultation system based on artificial intelligence triage according to claim 6, characterized in that, The payment and delivery module includes: The medical insurance interface call unit is used to calculate the reimbursement ratio; The drug delivery interface unit is used to generate delivery orders and provide feedback on logistics status.

10. The remote consultation system based on artificial intelligence triage according to claim 6, characterized in that, The architecture of the system includes: The user terminal layer provides the access interface for patients and doctors. The data processing layer is used to realize identity recognition, intelligent triage, remote consultation, payment processing and delivery control; The data storage layer is used to store electronic health records, medical knowledge bases, prescriptions, and consultation records. The support service layer is used to provide interface services for hospital information systems, medical insurance platforms, payment gateways, and logistics systems.

Images