Artificial intelligence navigation doctor seeing system and method based on real-time diagnosis and treatment data

Abstract

The invention discloses an artificial intelligence navigation doctor seeing system and method based on real-time diagnosis and treatment data, belongs to the technical field of medical informatization, and aims to solve the technical problems of complexity and low efficiency in the existing hospital doctor seeing process and improve the doctor seeing efficiency and patient satisfaction. Comprises: a multi-modal perception fusion module for performing intention understanding and OCR recognition through a large model, and establishing association between semantic data and space coordinates of a target department; the navigation planning module is used for optimizing the 3D navigation path through a Dijkstra algorithm and a neural network prediction algorithm based on the current position coordinates of the user and the treatment congestion state of the target department to generate an optimal navigation path; and the closed-loop management module is used for generating a payment and medicine taking plan and a 3D navigation path through a large model based on the real-time states of the payment window and the pharmacy, the diagnosis and treatment result and the payment mode, and generating a discharge traffic scheme for the user through the large model based on the target address and the travel mode.

Description

Technical Field

[0001] This invention relates to the field of medical information technology, specifically to an artificial intelligence-guided medical treatment system and method based on real-time medical data. Background Technology

[0002] In the existing hospital visit process, patients often face numerous inconveniences, such as spending long hours finding the location of departments, queuing for registration, payment, and medication pickup. This is especially true for middle-aged and elderly patients and first-time visitors, whose process is even more cumbersome and complex. Furthermore, congestion levels vary across departments, making it difficult for patients to obtain real-time information to choose the optimal route, resulting in low efficiency. Therefore, developing a system that intelligently guides patients is crucial. Currently, traditional 3D navigation only provides static path planning, which cannot adapt to the dynamically changing hospital environment; data isolation between the HIS system and the navigation system leads to poor information flow; and the lack of closed-loop management for the entire visit process prevents the provision of a continuous and seamless patient experience.

[0003] How to solve the problems of cumbersome and inefficient procedures in existing hospital visits and improve efficiency and patient satisfaction is a technical problem that needs to be addressed. Summary of the Invention

[0004] The technical objective of this invention is to address the above-mentioned shortcomings by providing an artificial intelligence-guided medical treatment system and method based on real-time medical data, in order to solve the technical problem of how to resolve the cumbersome and inefficient nature of existing hospital medical treatment processes and improve medical treatment efficiency and patient satisfaction.

[0005] In a first aspect, the present invention provides an artificial intelligence navigation medical treatment system based on real-time medical data, comprising a multimodal perception fusion module, a navigation planning module, and a closed-loop management module;

[0006] The multimodal perception fusion module interacts with users through a 3D digital human interactive interface. It is used to receive the user's guidance request, understand the intent through a large model and perform OCR recognition to obtain semantic data including guidance needs, medical information, registration information and pathology information. After verifying the semantic data, the semantic data is associated with the spatial coordinates of the target department to generate a 3D navigation path including path planning and floor switching prompts.

[0007] The navigation planning module is used to collect the user's current location coordinates in real time, and to call the hospital information system to identify the congestion status of the user's target department in real time. Based on the user's current location coordinates and the congestion status of the target department, the 3D navigation path is optimized through Dijkstra's algorithm and neural network prediction algorithm to generate the optimal navigation path. After the user completes the visit, a report detection mechanism is triggered to remind the user to have a follow-up visit and navigate to the target department.

[0008] The closed-loop management module is used to call the hospital information system to obtain the real-time status of the payment window and pharmacy. Based on the real-time status of the payment window and pharmacy, as well as the diagnosis and treatment results and payment methods, it generates a payment and medication collection plan and a 3D navigation path through a large model. After the user completes medication collection or does not need to collect medication, the system receives the user's selected destination and mode of transportation through the 3D digital human interaction interface, and generates a discharge transportation plan for the user based on the target address and mode of transportation through the large model.

[0009] Preferably, the multimodal perception fusion module is used to perform the following operations;

[0010] The system receives user-initiated triage requests through a 3D intelligent human interaction interface. The triage requests consist of voice data and image data.

[0011] Based on user-initiated triage requests, a large-scale model is used to analyze triage requests and determine user types, including users who have registered and paid and users who have not registered and paid.

[0012] For users who have not registered and paid, a 3D navigation path is generated for the user based on the user's current location coordinates and a large model. The user is then guided to the recommended window to register and pay using the 3D navigation path.

[0013] For users who have already registered and paid, a large model is invoked to convert the user's submitted voice data into text data using speech recognition methods. Deep learning algorithms are then used to perform deep semantic understanding on the text data to obtain medical information, including department location, doctor information, and symptom description. OCR recognition is used to extract information from the user's submitted image data to obtain registration information and pathology information. The registration information includes the patient ID, the registered department, and the registration time. Based on a predefined information verification rule base, the medical information, registration information, and pathology information are matched and verified according to rules. If the verification fails, the user is reminded to provide the information again or manual intervention is required through the 3D digital human interactive interface.

[0014] Based on a predefined spatial semantic mapping model, semantic data is associated with spatial coordinates to achieve inference results of symptoms-department-path.

[0015] Based on the reasoning results, a 3D navigation path is generated for the user through a large model, and a path preview is provided through a 3D digital human interactive interface.

[0016] Preferably, the navigation planning module is used for the following operations:

[0017] The user's current location is collected via GIS and CPS.

[0018] Access the hospital information system to collect real-time data on patient congestion in the target department;

[0019] Based on the user's current location and the congestion status of the target department, the 3D navigation path is optimized using the Dijkstra algorithm and the neural network prediction algorithm to generate 3D navigation guidance that includes direction indicators, turning prompts, and floor switching. The weight function for generating navigation guidance is W = α × d + (1 - α) × t, where d represents the path distance, α is an adjustable parameter, and t is the estimated waiting time. The Dijkstra algorithm is used to calculate the shortest path, and the neural network prediction algorithm is used to estimate the waiting time of each department. The influence of path distance and waiting time on the optimal path is balanced by adjusting the α parameter.

[0020] Based on 3D navigation path guidance, users are guided to their target department. The navigation methods include both voice navigation and visual navigation. When the user is detected to have deviated from the navigation path, the path adjustment mechanism will promptly replan the path.

[0021] Once a user completes their medical visit, a report detection mechanism is triggered. Based on a predefined report monitoring rule base, the user's examination report is automatically analyzed. When abnormal results are detected or a follow-up visit is required, a reminder message is promptly sent to the user, and a follow-up visit navigation service is provided.

[0022] Preferably, the closed-loop management module is used to perform the following operations:

[0023] Establish a data interface with the pharmacy system to obtain the user's medication collection status in real time, and send a successful medication collection notification to the user when the system detects that the user has completed the medication collection.

[0024] For users who have completed medication collection or do not need to collect medication, the 3D digital human interaction interface allows users to select their destination and mode of transportation. The large model generates in-hospital navigation and public map travel service options for the best discharge location.

[0025] As a preferred option, the 3D digital human interactive interface features a 3D digital human image. The digital human is driven by the user's expressions, actions, and lip movements in real time based on the user's questions. The 3D digital human interactive interface provides visual information for 3D navigation maps and medical process flowcharts.

[0026] Secondly, the present invention provides an artificial intelligence-guided medical visit method based on real-time medical data, comprising the following steps:

[0027] Multimodal perception fusion: Through the interaction with users via the 3D digital human interactive interface, the system receives the user's guidance request, performs intent understanding and OCR recognition through a large model, and obtains semantic data including guidance needs, medical information, registration information and pathology information. After verifying the semantic data, the system establishes a correlation between the semantic data and the spatial coordinates of the target department to generate a 3D navigation path including path planning and floor switching prompts.

[0028] Navigation planning: The system collects the user's current location coordinates in real time, calls the hospital information system to identify the congestion status of the user's target department in real time, and optimizes the 3D navigation path based on the user's current location coordinates and the congestion status of the target department through Dijkstra's algorithm and neural network prediction algorithm to generate the optimal navigation path. After the user completes the visit, the system triggers a report detection mechanism to remind the user to have a follow-up visit and navigate to the target department.

[0029] Closed-loop management: The system calls the hospital information system to obtain the real-time status of payment windows and pharmacies. Based on the real-time status of payment windows and pharmacies, as well as treatment results and payment methods, a payment and medication collection plan is generated through a large model, and a 3D navigation path is generated. After the user completes medication collection or does not need to collect medication, the system receives the user's selected destination and mode of transportation through a 3D intelligent human interaction interface. Based on the target address and mode of transportation, a discharge transportation plan is generated for the user through a large model.

[0030] Preferably, multimodal perception fusion includes the following operations;

[0031] The system receives user-initiated triage requests through a 3D intelligent human interaction interface. The triage requests consist of voice data and image data.

[0032] Based on user-initiated triage requests, a large-scale model is used to analyze triage requests and determine user types, including users who have registered and paid and users who have not registered and paid.

[0033] For users who have not registered and paid, a 3D navigation path is generated for the user based on the user's current location coordinates and a large model. The user is then guided to the recommended window to register and pay using the 3D navigation path.

[0034] For users who have already registered and paid, a large model is invoked to convert the user's submitted voice data into text data using speech recognition methods. Deep learning algorithms are then used to perform deep semantic understanding on the text data to obtain medical information, including department location, doctor information, and symptom description. OCR recognition is used to extract information from the user's submitted image data to obtain registration information and pathology information. The registration information includes the patient ID, the registered department, and the registration time. Based on a predefined information verification rule base, the medical information, registration information, and pathology information are matched and verified according to rules. If the verification fails, the user is reminded to provide the information again or manual intervention is required through the 3D digital human interactive interface.

[0035] Based on a predefined spatial semantic mapping model, semantic data is associated with spatial coordinates to achieve inference results of symptoms-department-path.

[0036] Based on the reasoning results, a 3D navigation path is generated for the user through a large model, and a path preview is provided through a 3D digital human interactive interface.

[0037] As a preferred method, navigation planning includes the following operations:

[0038] The user's current location is collected via GIS and CPS.

[0039] Access the hospital information system to collect real-time data on patient congestion in the target department;

[0040] Based on the user's current location and the congestion status of the target department, the 3D navigation path is optimized using the Dijkstra algorithm and the neural network prediction algorithm to generate 3D navigation guidance that includes direction indicators, turning prompts, and floor switching. The weight function for generating navigation guidance is W = α × d + (1 - α) × t, where d represents the path distance, α is an adjustable parameter, and t is the estimated waiting time. The Dijkstra algorithm is used to calculate the shortest path, and the neural network prediction algorithm is used to estimate the waiting time of each department. The influence of path distance and waiting time on the optimal path is balanced by adjusting the α parameter.

[0041] Based on 3D navigation path guidance, users are guided to their target department. The navigation methods include both voice navigation and visual navigation. When the user is detected to have deviated from the navigation path, the path adjustment mechanism will promptly replan the path.

[0042] Once a user completes their medical visit, a report detection mechanism is triggered. Based on a predefined report monitoring rule base, the user's examination report is automatically analyzed. When abnormal results are detected or a follow-up visit is required, a reminder message is promptly sent to the user, and a follow-up visit navigation service is provided.

[0043] As a preferred approach, closed-loop management includes the following operations:

[0044] Establish a data interface with the pharmacy system to obtain the user's medication collection status in real time, and send a successful medication collection notification to the user when the system detects that the user has completed the medication collection.

[0045] For users who have completed medication collection or do not need to collect medication, the 3D digital human interaction interface allows users to select their destination and mode of transportation. The large model generates in-hospital navigation and public map travel service options for the best discharge location.

[0046] As a preferred option, the 3D digital human interactive interface features a 3D digital human image. The digital human is driven by the user's expressions, actions, and lip movements in real time based on the user's questions. The 3D digital human interactive interface provides visual information for 3D navigation maps and medical process flowcharts.

[0047] The AI-based navigation system and method for medical visits based on real-time medical data of the present invention have the following advantages:

[0048] 1. Through highly intelligent technological means, the hospital's medical process has been comprehensively optimized, significantly improving efficiency and patient satisfaction;

[0049] 2. The application of multimodal perception fusion mechanism and real-time diagnosis and treatment data-driven navigation enables patients to easily obtain department locations, treatment routes, and congestion information, effectively shortening the consultation time.

[0050] 3. The implementation of payment, medication collection, and discharge navigation functions has further improved the patient's medical experience and provided the hospital with a more efficient and convenient management tool. Attached Figure Description

[0051] To more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0052] The invention will be further described below with reference to the accompanying drawings.

[0053] Figure 1 This is a flowchart of an AI-based navigation method for medical visits based on real-time medical data, as described in Example 2. Detailed Implementation

[0054] The present invention will be further described below with reference to the accompanying drawings and specific embodiments, so that those skilled in the art can better understand and implement the present invention. However, the embodiments are not intended to limit the present invention. In the absence of conflict, the embodiments of the present invention and the technical features in the embodiments can be combined with each other.

[0055] This invention provides an AI-based navigation system and method for medical visits based on real-time medical data, which addresses the technical problem of how to solve the cumbersome and inefficient nature of existing hospital medical processes and improve medical efficiency and patient satisfaction.

[0056] Example 1:

[0057] The present invention discloses an artificial intelligence navigation medical treatment system based on real-time medical data, comprising a multimodal perception fusion module, a navigation planning module, and a closed-loop management module.

[0058] The multimodal perception fusion module interacts with users through a 3D digital human interactive interface. It receives user-initiated triage requests, performs intent understanding and OCR recognition through a large model, and obtains semantic data including triage needs, medical information, registration information, and pathology information. After verifying the semantic data, it establishes a correlation between the semantic data and the spatial coordinates of the target department to generate a 3D navigation path that includes path planning and floor switching prompts.

[0059] As a specific implementation of the multimodal perception fusion module, this module is used to perform the following operations;

[0060] (1) Receive the triage request initiated by the user through the 3D digital human interactive interface, wherein the triage request is voice data and image data;

[0061] (2) Based on the user-initiated triage needs, triage needs analysis is conducted through a large model to determine user types, including users who have registered and paid and users who have not registered and paid.

[0062] (3) For users who have not registered and paid, a 3D navigation path is generated for the user based on the user's current location coordinates and through a large model. The user is then guided to the recommended window to register and pay through the 3D navigation path.

[0063] (4) For users who have registered and paid, the large model is called and the voice data submitted by the user is converted into text data through the speech recognition method. The text data is then subjected to deep semantic understanding through the deep learning algorithm to obtain the medical information, which includes the department location, doctor information and symptom description. The image data submitted by the user is extracted through OCR recognition to obtain the registration information and pathology information. The registration information includes the patient ID, the registered department and the registration time. The medical information, registration information and pathology information are matched and verified according to the predefined information verification rule base. If the verification fails, the user is reminded to provide the information again or to make manual intervention through the 3D digital human interactive interface.

[0064] (5) Based on a predefined spatial semantic mapping model, semantic data is associated with spatial coordinates to achieve the reasoning results of symptoms-department-path;

[0065] (6) Based on the reasoning results, generate 3D navigation paths for users through large models, and provide path previews for 3D navigation paths through the 3D digital human interactive interface.

[0066] In this embodiment, the workflow of the multimodal perception fusion module involves the user initiating a triage request, receiving the 3D digital human interactive interface, understanding the user's intent through a large model, recognizing medical records / registration slips / QR codes using OCR, extracting pathological features / registration information and verifying them, deep mapping of spatial coordinates and semantic data, and generating personalized 3D navigation paths.

[0067] Users initiate referral requests via a large interactive interface on a terminal device (such as a smartphone, tablet, or self-service terminal within the hospital). Requests can be made via voice or text input. Details: The interface design should be simple and clear, supporting one-click referral initiation, and providing voice-to-text and text editing functions to meet the needs of different users.

[0068] The 3D AI-powered human-computer interaction interface receives user voice or text input and transmits it to a large backend model for processing. For patients who have already registered and paid, they are directly navigated to the consultation room; for those who haven't, they are navigated to a recommended window for registration and payment. Details: The 3D AI-powered human-computer interface must possess highly realistic facial expressions, movements, and lip-syncing capabilities to enhance the user's interactive experience. Simultaneously, the interface must display the processing status in real time to improve user perception.

[0069] The large-scale model understands user intent by leveraging Natural Language Processing (NLP) technology to deeply understand users' referral needs, including department locations, doctor information, and symptom descriptions. Details include employing an advanced pre-trained language model, combined with a medical terminology database, and fine-tuning it to optimize the understanding of medical referral needs. Simultaneously, a context-aware mechanism is introduced to ensure the coherence of the dialogue.

[0070] OCR recognition of medical records / registration slips / QR codes: The system uses OCR (Optical Character Recognition) technology to recognize user-provided registration slips, medical records, or QR codes, extracting key registration information. Details: The OCR algorithm must possess high accuracy and robustness, capable of handling registration slips and medical records of different formats and qualities. Simultaneously, image preprocessing techniques are introduced to improve recognition accuracy.

[0071] Extracting and verifying pathological features / registration information: The system extracts key information from the identified text, such as patient ID, department, and registration time, and verifies it to ensure accuracy. Details: An information verification rule base is established to match and validate the extracted information according to rules. For information that fails verification, the system promptly prompts the user to resubmit or requires manual intervention.

[0072] Deep mapping between spatial coordinates and semantic data: This involves deeply mapping the spatial coordinate data of the navigation system with the semantic data of a large model to achieve joint reasoning based on "symptom-department-path". Details: A spatial semantic mapping model is constructed to associate semantic information such as department location and symptom description with spatial coordinates. Machine learning algorithms are used to continuously optimize the mapping relationship, improving navigation accuracy.

[0073] Personalized 3D navigation routes are generated. The system generates personalized 3D navigation routes for patients based on joint inference results, including route planning and floor switching prompts. Details: A precise map of the hospital's interior is constructed using 3D modeling technology, combined with a route planning algorithm to generate the optimal navigation route. A route preview function is also provided, allowing users to familiarize themselves with the navigation route in advance.

[0074] The navigation planning module is used to collect the user's current location coordinates in real time, and to call the hospital information system to identify the congestion status of the user's target department in real time. Based on the user's current location coordinates and the congestion status of the target department, the 3D navigation path is optimized through Dijkstra's algorithm and neural network prediction algorithm to generate the optimal navigation path. After the user completes the visit, a report detection mechanism is triggered to remind the user to have a follow-up visit and navigate to the target department.

[0075] As a specific implementation of the navigation planning module, this module is used for the following operations:

[0076] (1) Collect the user's current location through GIS and CPS;

[0077] (2) Call the hospital information system to collect real-time data on the congestion status of the target department;

[0078] (3) Based on the user's current location and the congestion status of the target department, the 3D navigation path is optimized using the Dijkstra algorithm and the neural network prediction algorithm to generate 3D navigation guidance that includes direction indication, turning prompts and floor switching. The weight function when generating navigation guidance is W=α×d+(1-α)×t, where d represents the path distance, α is an adjustable parameter, and t is the estimated waiting time. The Dijkstra algorithm is used to calculate the shortest path, and the neural network prediction algorithm is used to estimate the waiting time of each department. The influence of path distance and waiting time on the optimal path is balanced by adjusting the α parameter.

[0079] (4) Based on the 3D navigation path, guide the user to the target department. The navigation methods include voice navigation and visual navigation. When the user is detected to have deviated from the navigation path, the path adjustment mechanism will re-plan the path in a timely manner.

[0080] (5) After the user completes the medical visit, the report detection mechanism is triggered. The user's examination report is automatically analyzed based on the predefined report monitoring rule library. When abnormal results are detected or a follow-up visit is required, a reminder message is sent to the user in a timely manner, and a follow-up visit navigation service is provided.

[0081] In this embodiment, the navigation planning module workflow involves the GIS system locating the user's position, the large model backend being linked to the HIS system, generating the optimal medical route, the user following the navigation to the destination, and reporting, monitoring, and follow-up visit reminders.

[0082] The GIS system locates the user's position by utilizing GIS (Geographic Information System) and GPS (Global Positioning System) data to pinpoint the user's precise location within the hospital in real time. Details include: an integrated high-precision positioning module to ensure accurate location tracking even in complex hospital environments; and a location calibration function that allows users to manually adjust their location information.

[0083] The large-scale model backend is integrated with the hospital's HIS (Hospital Information System). This integration allows for real-time identification of the departments patients need to visit and dynamic analysis of congestion in each department. Detailed explanation: A standardized data interface is established to ensure real-time data synchronization between the large-scale model backend and the HIS system. Simultaneously, data cleaning and preprocessing mechanisms are introduced to improve data quality.

[0084] The optimal medical treatment route is generated by combining the patient's current location with the destination department information, using Dijkstra's algorithm and a neural network prediction algorithm. The weight function is W = α × d + (1 - α) × t, where d represents the path distance, α is an adjustable parameter (∈ [0.3, 0.7]), and t is the estimated waiting time. Details: Dijkstra's algorithm is used to calculate the shortest path, and the neural network prediction algorithm is used to estimate the waiting time for each department. By adjusting the α parameter, the influence of path distance and waiting time on the optimal route can be balanced.

[0085] Users follow the navigation to their destination. The system provides navigation prompts, including directional indicators, turn prompts, and floor level changes, allowing users to smoothly reach their destination. Details: Both voice and visual navigation are provided to meet the needs of different users. A real-time route adjustment mechanism is also implemented; when the system detects that the user has deviated from the navigation path, it promptly replans the route.

[0086] Report monitoring and follow-up visit reminders are implemented as follows: After a patient's visit, the system automatically triggers a report monitoring mechanism, reminding the patient to return for a follow-up visit and navigating them to the doctor's office. Details: A report monitoring rule base is established to automatically analyze patient examination reports. When abnormal results are detected or a follow-up visit is required, a timely reminder message is sent to the patient, along with follow-up visit navigation services.

[0087] The closed-loop management module is used to call the hospital information system to obtain the real-time status of the payment window and pharmacy. Based on the real-time status of the payment window and pharmacy, as well as the diagnosis and treatment results and payment methods, it generates a payment and medication collection plan and a 3D navigation path through a large model. After the user completes medication collection or does not need to collect medication, the system receives the user's selected destination and mode of transportation through the 3D digital human interaction interface, and generates a discharge transportation plan for the user based on the target address and mode of transportation through the large model.

[0088] As a specific implementation of the closed-loop management module, this module is used to perform the following operations:

[0089] (1) Establish a data interface with the pharmacy system to obtain the user's medication collection status in real time, and send a medication collection success prompt to the user when the user completes the medication collection.

[0090] (2) For users who have completed or do not need to pick up their medication, the 3D digital human interaction interface allows users to select their destination and mode of transportation. The large model generates the best in-hospital navigation and public map travel service options for the discharge location.

[0091] In this embodiment, the closed-loop management module's workflow involves intelligent navigation to the optimal payment window or pharmacy, prompts indicating whether medication pickup is complete or unnecessary, and personalized suggestions for discharge transportation.

[0092] Intelligent navigation to the optimal payment window or pharmacy: The system intelligently navigates to the optimal payment window or pharmacy based on the patient's treatment results and payment method, reducing patient waiting time. Details: A real-time status monitoring mechanism is established for payment windows and pharmacies, including the number of people queuing and waiting time. The optimal payment or medication pickup route is dynamically planned based on the patient's payment method and treatment results.

[0093] Medication pickup completion or no-pickup notification: The system automatically notifies the patient that medication pickup was successful upon completion; for patients who do not require medication pickup, a corresponding notification is also provided. Details: A data interface is established with the pharmacy system to obtain real-time medication pickup status. When medication pickup is detected as complete, a successful pickup notification is promptly sent to the patient.

[0094] Personalized discharge transportation suggestions are provided as follows: Patients who have completed medication pickup or do not require medication pickup can select their destination and mode of transportation through a large-scale interactive interface. The system then provides in-hospital navigation to the optimal discharge location and public map travel services. Details: The system integrates a public transportation API, offering multiple travel options and comparisons. Furthermore, it considers the patient's personal preferences and actual circumstances, such as whether they are carrying heavy items or require accessibility facilities, to provide personalized discharge transportation suggestions.

[0095] In this embodiment, a 3D digital human interactive interface is formed on the 3D digital human image. The digital human is driven by the user's expression, actions and lip movements in real time according to the user's questions. The 3D digital human interactive interface provides visual information for 3D navigation maps and medical treatment flowcharts.

[0096] This embodiment of the system deeply integrates DICT (Digital Information and Communication Technology) technology. Through a multimodal perception fusion mechanism and driven by real-time diagnostic data, it achieves a fully intelligent service, from accurate symptom description to intelligent department navigation, optimal department selection, dynamic route planning, convenient guidance for payment and medication collection, and personalized discharge transportation suggestions. This significantly improves hospital efficiency and greatly optimizes the patient experience.

[0097] Example 2:

[0098] This invention provides an AI-based navigation method for medical visits based on real-time medical data, comprising three steps: multimodal perception fusion, navigation planning, and closed-loop management.

[0099] Step S100 Multimodal Perception Fusion: Interact with the user through the 3D digital human interactive interface, receive the user's guidance request, perform intent understanding and OCR recognition through the large model, obtain semantic data including guidance needs, medical information, registration information and pathology information, after verifying the semantic data, establish the association between the semantic data and the spatial coordinates of the target department, and generate a 3D navigation path including path planning and floor switching prompts.

[0100] As a specific implementation of multimodal perception fusion, this step includes the following operations;

[0101] (1) Receive the triage request initiated by the user through the 3D digital human interactive interface, wherein the triage request is voice data and image data;

[0102] (2) Based on the user-initiated triage needs, triage needs analysis is conducted through a large model to determine user types, including users who have registered and paid and users who have not registered and paid.

[0103] (3) For users who have not registered and paid, a 3D navigation path is generated for the user based on the user's current location coordinates and through a large model. The user is then guided to the recommended window to register and pay through the 3D navigation path.

[0104] (4) For users who have registered and paid, the large model is called and the voice data submitted by the user is converted into text data through the speech recognition method. The text data is then subjected to deep semantic understanding through the deep learning algorithm to obtain the medical information, which includes the department location, doctor information and symptom description. The image data submitted by the user is extracted through OCR recognition to obtain the registration information and pathology information. The registration information includes the patient ID, the registered department and the registration time. The medical information, registration information and pathology information are matched and verified according to the predefined information verification rule base. If the verification fails, the user is reminded to provide the information again or to make manual intervention through the 3D digital human interactive interface.

[0105] (5) Based on a predefined spatial semantic mapping model, semantic data is associated with spatial coordinates to achieve the reasoning results of symptoms-department-path;

[0106] (6) Based on the reasoning results, generate 3D navigation paths for users through large models, and provide path previews for 3D navigation paths through the 3D digital human interactive interface.

[0107] In this embodiment, the multimodal perception fusion workflow involves the user initiating a triage request, receiving data from the 3D digital human interactive interface, understanding the user's intent through a large model, recognizing medical records / registration slips / QR codes using OCR, extracting pathological features / registration information and verifying them, deep mapping of spatial coordinates and semantic data, and generating personalized 3D navigation paths.

[0108] Users initiate referral requests via a large interactive interface on a terminal device (such as a smartphone, tablet, or self-service terminal within the hospital). Requests can be made via voice or text input. Details: The interface design should be simple and clear, supporting one-click referral initiation, and providing voice-to-text and text editing functions to meet the needs of different users.

[0109] The 3D AI-powered human-computer interaction interface receives user voice or text input and transmits it to a large backend model for processing. For patients who have already registered and paid, they are directly navigated to the consultation room; for those who haven't, they are navigated to a recommended window for registration and payment. Details: The 3D AI-powered human-computer interface must possess highly realistic facial expressions, movements, and lip-syncing capabilities to enhance the user's interactive experience. Simultaneously, the interface must display the processing status in real time to improve user perception.

[0110] The large-scale model understands user intent by leveraging Natural Language Processing (NLP) technology to deeply understand users' referral needs, including department locations, doctor information, and symptom descriptions. Details include employing an advanced pre-trained language model, combined with a medical terminology database, and fine-tuning it to optimize the understanding of medical referral needs. Simultaneously, a context-aware mechanism is introduced to ensure the coherence of the dialogue.

[0111] OCR recognition of medical records / registration slips / QR codes: The system uses OCR (Optical Character Recognition) technology to recognize user-provided registration slips, medical records, or QR codes, extracting key registration information. Details: The OCR algorithm must possess high accuracy and robustness, capable of handling registration slips and medical records of different formats and qualities. Simultaneously, image preprocessing techniques are introduced to improve recognition accuracy.

[0112] Extracting and verifying pathological features / registration information: The system extracts key information from the identified text, such as patient ID, department, and registration time, and verifies it to ensure accuracy. Details: An information verification rule base is established to match and validate the extracted information according to rules. For information that fails verification, the system promptly prompts the user to resubmit or requires manual intervention.

[0113] Deep mapping between spatial coordinates and semantic data: This involves deeply mapping the spatial coordinate data of the navigation system with the semantic data of a large model to achieve joint reasoning based on "symptom-department-path". Details: A spatial semantic mapping model is constructed to associate semantic information such as department location and symptom description with spatial coordinates. Machine learning algorithms are used to continuously optimize the mapping relationship, improving navigation accuracy.

[0114] Personalized 3D navigation routes are generated. The system generates personalized 3D navigation routes for patients based on joint inference results, including route planning and floor switching prompts. Details: A precise map of the hospital's interior is constructed using 3D modeling technology, combined with a route planning algorithm to generate the optimal navigation route. A route preview function is also provided, allowing users to familiarize themselves with the navigation route in advance.

[0115] Step S200 Navigation Planning: Real-time collection of the user's current location coordinates, calling the hospital information system to identify the congestion status of the user's target department in real time, and optimizing the 3D navigation path based on the user's current location coordinates and the congestion status of the target department through Dijkstra's algorithm and neural network prediction algorithm to generate the optimal navigation path. After the user completes the visit, a report detection mechanism is triggered to remind the user to have a follow-up visit and navigate to the target department.

[0116] As a specific implementation of navigation planning, the steps are as follows:

[0117] (1) Collect the user's current location through GIS and CPS;

[0118] (2) Call the hospital information system to collect real-time data on the congestion status of the target department;

[0119] (3) Based on the user's current location and the congestion status of the target department, the 3D navigation path is optimized using the Dijkstra algorithm and the neural network prediction algorithm to generate 3D navigation guidance that includes direction indication, turning prompts and floor switching. The weight function when generating navigation guidance is W=α×d+(1-α)×t, where d represents the path distance, α is an adjustable parameter, and t is the estimated waiting time. The Dijkstra algorithm is used to calculate the shortest path, and the neural network prediction algorithm is used to estimate the waiting time of each department. The influence of path distance and waiting time on the optimal path is balanced by adjusting the α parameter.

[0120] (4) Based on the 3D navigation path, guide the user to the target department. The navigation methods include voice navigation and visual navigation. When the user is detected to have deviated from the navigation path, the path adjustment mechanism will re-plan the path in a timely manner.

[0121] (5) After the user completes the medical visit, the report detection mechanism is triggered. The user's examination report is automatically analyzed based on the predefined report monitoring rule library. When abnormal results are detected or a follow-up visit is required, a reminder message is sent to the user in a timely manner, and a follow-up visit navigation service is provided.

[0122] In this embodiment, the navigation planning workflow involves the GIS system locating the user's position, the large model backend being linked with the HIS system, generating the optimal medical route, the user following the navigation to the destination, and reporting, monitoring, and follow-up visit reminders.

[0123] The GIS system locates the user's position by utilizing GIS (Geographic Information System) and GPS (Global Positioning System) data to pinpoint the user's precise location within the hospital in real time. Details include: an integrated high-precision positioning module to ensure accurate location tracking even in complex hospital environments; and a location calibration function that allows users to manually adjust their location information.

[0124] The large-scale model backend is integrated with the hospital's HIS (Hospital Information System). This integration allows for real-time identification of the departments patients need to visit and dynamic analysis of congestion in each department. Detailed explanation: A standardized data interface is established to ensure real-time data synchronization between the large-scale model backend and the HIS system. Simultaneously, data cleaning and preprocessing mechanisms are introduced to improve data quality.

[0125] The optimal medical treatment route is generated by combining the patient's current location with the destination department information, using Dijkstra's algorithm and a neural network prediction algorithm. The weight function is W = α × d + (1 - α) × t, where d represents the path distance, α is an adjustable parameter (∈ [0.3, 0.7]), and t is the estimated waiting time. Details: Dijkstra's algorithm is used to calculate the shortest path, and the neural network prediction algorithm is used to estimate the waiting time for each department. By adjusting the α parameter, the influence of path distance and waiting time on the optimal route can be balanced.

[0126] Users follow the navigation to their destination. The system provides navigation prompts, including directional indicators, turn prompts, and floor level changes, allowing users to smoothly reach their destination. Details: Both voice and visual navigation are provided to meet the needs of different users. A real-time route adjustment mechanism is also implemented; when the system detects that the user has deviated from the navigation path, it promptly replans the route.

[0127] Report monitoring and follow-up visit reminders are implemented as follows: After a patient's visit, the system automatically triggers a report monitoring mechanism, reminding the patient to return for a follow-up visit and navigating them to the doctor's office. Details: A report monitoring rule base is established to automatically analyze patient examination reports. When abnormal results are detected or a follow-up visit is required, a timely reminder message is sent to the patient, along with follow-up visit navigation services.

[0128] Step S300 Closed-loop Management: The hospital information system is called to obtain the real-time status of the payment window and pharmacy. Based on the real-time status of the payment window and pharmacy, as well as the diagnosis and treatment results and payment methods, a payment and medication collection plan is generated through a large model, and a 3D navigation path is generated. After the user completes medication collection or does not need to collect medication, the destination and mode of transportation selected by the user are received through the 3D digital human interaction interface. Based on the target address and mode of transportation, a discharge transportation plan is generated for the user through the large model.

[0129] As a specific implementation of closed-loop management, this step includes the following operations:

[0130] (1) Establish a data interface with the pharmacy system to obtain the user's medication collection status in real time, and send a medication collection success prompt to the user when the user completes the medication collection.

[0131] (2) For users who have completed or do not need to pick up their medication, the 3D digital human interaction interface allows users to select their destination and mode of transportation. The large model generates the best in-hospital navigation and public map travel service options for the discharge location.

[0132] In this embodiment, the closed-loop management workflow involves intelligent navigation to the optimal payment window or pharmacy, prompts indicating whether medication collection is complete or unnecessary, and personalized suggestions for discharge transportation.

[0133] Intelligent navigation to the optimal payment window or pharmacy: The system intelligently navigates to the optimal payment window or pharmacy based on the patient's treatment results and payment method, reducing patient waiting time. Details: A real-time status monitoring mechanism is established for payment windows and pharmacies, including the number of people queuing and waiting time. The optimal payment or medication pickup route is dynamically planned based on the patient's payment method and treatment results.

[0134] Medication pickup completion or no-pickup notification: The system automatically notifies the patient that medication pickup was successful upon completion; for patients who do not require medication pickup, a corresponding notification is also provided. Details: A data interface is established with the pharmacy system to obtain real-time medication pickup status. When medication pickup is detected as complete, a successful pickup notification is promptly sent to the patient.

[0135] Personalized discharge transportation suggestions are provided as follows: Patients who have completed medication pickup or do not require medication pickup can select their destination and mode of transportation through a large-scale interactive interface. The system then provides in-hospital navigation to the optimal discharge location and public map travel services. Details: The system integrates a public transportation API, offering multiple travel options and comparisons. Furthermore, it considers the patient's personal preferences and actual circumstances, such as whether they are carrying heavy items or require accessibility facilities, to provide personalized discharge transportation suggestions.

[0136] In this embodiment, a 3D digital human interactive interface is formed on the 3D digital human image. The digital human is driven by the user's expression, actions and lip movements in real time according to the user's questions. The 3D digital human interactive interface provides visual information for 3D navigation maps and medical treatment flowcharts.

[0137] The method in this embodiment is based on the system implementation disclosed in Embodiment 1.

[0138] The above provides a detailed description of the AI-guided medical treatment system and method based on real-time diagnostic data provided by this invention. Specific examples have been used to illustrate the principles and implementation methods of this invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of this invention. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this invention. Therefore, the content of this specification should not be construed as a limitation of this invention.

Claims

1. An AI-powered navigation system for medical visits based on real-time diagnostic data, characterized in that: It includes a multimodal perception fusion module, a navigation planning module, and a closed-loop management module; The multimodal perception fusion module interacts with users through a 3D digital human interactive interface. It is used to receive the user's guidance request, understand the intent through a large model and perform OCR recognition to obtain semantic data including guidance needs, medical information, registration information and pathology information. After verifying the semantic data, the semantic data is associated with the spatial coordinates of the target department to generate a 3D navigation path including path planning and floor switching prompts. The navigation planning module is used to collect the user's current location coordinates in real time, and to call the hospital information system to identify the congestion status of the user's target department in real time. Based on the user's current location coordinates and the congestion status of the target department, the 3D navigation path is optimized through Dijkstra's algorithm and neural network prediction algorithm to generate the optimal navigation path. After the user completes the visit, a report detection mechanism is triggered to remind the user to have a follow-up visit and navigate to the target department. The closed-loop management module is used to call the hospital information system to obtain the real-time status of the payment window and pharmacy. Based on the real-time status of the payment window and pharmacy, as well as the diagnosis and treatment results and payment methods, it generates a payment and medication collection plan and a 3D navigation path through a large model. After the user completes medication collection or does not need to collect medication, the system receives the user's selected destination and mode of transportation through the 3D digital human interaction interface, and generates a discharge transportation plan for the user based on the target address and mode of transportation through the large model.

2. The AI-powered navigation system for medical visits based on real-time diagnostic data according to claim 1, characterized in that, The multimodal perception fusion module is used to perform the following operations; The system receives user-initiated triage requests through a 3D intelligent human interaction interface. The triage requests consist of voice data and image data. Based on user-initiated triage requests, a large-scale model is used to analyze triage requests and determine user types, including users who have registered and paid and users who have not registered and paid. For users who have not registered and paid, a 3D navigation path is generated for the user based on the user's current location coordinates and a large model. The user is then guided to the recommended window to register and pay using the 3D navigation path. For users who have already registered and paid, a large model is invoked to convert the user's submitted voice data into text data using speech recognition methods. Deep learning algorithms are then used to perform deep semantic understanding on the text data to obtain medical information, including department location, doctor information, and symptom description. OCR recognition is used to extract information from the user's submitted image data to obtain registration information and pathology information. The registration information includes the patient ID, the registered department, and the registration time. Based on a predefined information verification rule base, the medical information, registration information, and pathology information are matched and verified according to rules. If the verification fails, the user is reminded to provide the information again or manual intervention is required through the 3D digital human interactive interface. Based on a predefined spatial semantic mapping model, semantic data is associated with spatial coordinates to achieve inference results of symptoms-department-path. Based on the reasoning results, a 3D navigation path is generated for the user through a large model, and a path preview is provided through a 3D digital human interactive interface.

3. The AI-powered navigation system for medical visits based on real-time diagnostic data according to claim 1, characterized in that, The navigation planning module is used for the following operations: The user's current location is collected via GIS and CPS. Access the hospital information system to collect real-time data on patient congestion in the target department; Based on the user's current location and the congestion status of the target department, the 3D navigation path is optimized using the Dijkstra algorithm and the neural network prediction algorithm to generate 3D navigation guidance that includes direction indicators, turning prompts, and floor switching. The weight function for generating navigation guidance is W = α × d + (1 - α) × t, where d represents the path distance, α is an adjustable parameter, and t is the estimated waiting time. The Dijkstra algorithm is used to calculate the shortest path, and the neural network prediction algorithm is used to estimate the waiting time of each department. The influence of path distance and waiting time on the optimal path is balanced by adjusting the α parameter. Based on 3D navigation path guidance, users are guided to their target department. The navigation methods include both voice navigation and visual navigation. When the user is detected to have deviated from the navigation path, the path adjustment mechanism will promptly replan the path. Once a user completes their medical visit, a report detection mechanism is triggered. Based on a predefined report monitoring rule base, the user's examination report is automatically analyzed. When abnormal results are detected or a follow-up visit is required, a reminder message is promptly sent to the user, and a follow-up visit navigation service is provided.

4. The AI-powered navigation system for medical visits based on real-time diagnostic data according to claim 1, characterized in that, The closed-loop management module is used to perform the following operations: Establish a data interface with the pharmacy system to obtain the user's medication collection status in real time, and send a successful medication collection notification to the user when the system detects that the user has completed the medication collection. For users who have completed medication collection or do not need to collect medication, the 3D digital human interaction interface allows users to select their destination and mode of transportation. The large model generates in-hospital navigation and public map travel service options for the best discharge location.

5. The AI-powered navigation system for medical visits based on real-time diagnostic data according to claim 1, characterized in that, The 3D AI human interactive interface features a 3D AI human figure. The AI ​​human figure responds to the user's questions with real-time facial expressions, actions, and lip movements. The 3D AI human interactive interface provides visual information for 3D navigation maps and medical consultation flowcharts.

6. An AI-guided medical appointment method based on real-time medical data, characterized in that, Includes the following steps: Multimodal perception fusion: Through the interaction with users via the 3D digital human interactive interface, the system receives the user's guidance request, performs intent understanding and OCR recognition through a large model, and obtains semantic data including guidance needs, medical information, registration information and pathology information. After verifying the semantic data, the system establishes a correlation between the semantic data and the spatial coordinates of the target department to generate a 3D navigation path including path planning and floor switching prompts. Navigation planning: The system collects the user's current location coordinates in real time, calls the hospital information system to identify the congestion status of the user's target department in real time, and optimizes the 3D navigation path based on the user's current location coordinates and the congestion status of the target department through Dijkstra's algorithm and neural network prediction algorithm to generate the optimal navigation path. After the user completes the visit, the system triggers a report detection mechanism to remind the user to have a follow-up visit and navigate to the target department. Closed-loop management: The system calls the hospital information system to obtain the real-time status of payment windows and pharmacies. Based on the real-time status of payment windows and pharmacies, as well as treatment results and payment methods, a payment and medication collection plan is generated through a large model, and a 3D navigation path is generated. After the user completes medication collection or does not need to collect medication, the system receives the user's selected destination and mode of transportation through a 3D intelligent human interaction interface. Based on the target address and mode of transportation, a discharge transportation plan is generated for the user through a large model.

7. The AI-guided medical appointment method based on real-time medical data according to claim 6, characterized in that, Multimodal sensing fusion includes the following operations; The system receives user-initiated triage requests through a 3D intelligent human interaction interface. The triage requests consist of voice data and image data. Based on user-initiated triage requests, a large-scale model is used to analyze triage requests and determine user types, including users who have registered and paid and users who have not registered and paid. For users who have not registered and paid, a 3D navigation path is generated for the user based on the user's current location coordinates and a large model. The user is then guided to the recommended window to register and pay using the 3D navigation path. For users who have already registered and paid, a large model is invoked to convert the user's submitted voice data into text data using speech recognition methods. Deep learning algorithms are then used to perform deep semantic understanding on the text data to obtain medical information, including department location, doctor information, and symptom description. OCR recognition is used to extract information from the user's submitted image data to obtain registration information and pathology information. The registration information includes the patient ID, the registered department, and the registration time. Based on a predefined information verification rule base, the medical information, registration information, and pathology information are matched and verified according to rules. If the verification fails, the user is reminded to provide the information again or manual intervention is required through the 3D digital human interactive interface. Based on a predefined spatial semantic mapping model, semantic data is associated with spatial coordinates to achieve inference results of symptoms-department-path. Based on the reasoning results, a 3D navigation path is generated for the user through a large model, and a path preview is provided through a 3D digital human interactive interface.

8. The AI-guided medical appointment method based on real-time medical data according to claim 6, characterized in that, Navigation planning includes the following operations: The user's current location is collected via GIS and CPS. Access the hospital information system to collect real-time data on patient congestion in the target department; Based on the user's current location and the congestion status of the target department, the 3D navigation path is optimized using the Dijkstra algorithm and the neural network prediction algorithm to generate 3D navigation guidance that includes direction indicators, turning prompts, and floor switching. The weight function for generating navigation guidance is W = α × d + (1 - α) × t, where d represents the path distance, α is an adjustable parameter, and t is the estimated waiting time. The Dijkstra algorithm is used to calculate the shortest path, and the neural network prediction algorithm is used to estimate the waiting time of each department. The influence of path distance and waiting time on the optimal path is balanced by adjusting the α parameter. Based on 3D navigation path guidance, users are guided to their target department. The navigation methods include both voice navigation and visual navigation. When the user is detected to have deviated from the navigation path, the path adjustment mechanism will promptly replan the path. Once a user completes their medical visit, a report detection mechanism is triggered. Based on a predefined report monitoring rule base, the user's examination report is automatically analyzed. When abnormal results are detected or a follow-up visit is required, a reminder message is promptly sent to the user, and a follow-up visit navigation service is provided.

9. The artificial intelligence-guided medical visit method based on real-time medical data according to claim 6, characterized in that, Closed-loop management includes the following operations: Establish a data interface with the pharmacy system to obtain the user's medication collection status in real time, and send a successful medication collection notification to the user when the system detects that the user has completed the medication collection. For users who have completed medication collection or do not need to collect medication, the 3D digital human interaction interface allows users to select their destination and mode of transportation. The large model generates in-hospital navigation and public map travel service options for the best discharge location.

10. The artificial intelligence-guided medical treatment method based on real-time medical data according to claim 6, characterized in that, The 3D AI human interactive interface features a 3D AI human figure. The AI ​​human figure responds to the user's questions with real-time facial expressions, actions, and lip movements. The 3D AI human interactive interface provides visual information for 3D navigation maps and medical consultation flowcharts.

Images