Acute critical symptom auxiliary data analysis method and system based on 5G and AR glasses

Abstract

The invention provides an acute critical disease auxiliary data analysis method and system based on 5G and AR glasses, and relates to the technical field of data analysis, 5G wireless connection and data transmission are carried out on a base node, a transfer node and an expert node, first transfer analysis data are acquired through an AR glasses all-in-one machine of the base node, feature extraction and feature critical analysis are carried out, and the first transfer analysis data are transmitted to the expert node; the extracted features are uploaded, and then a first feedback instruction output by the expert node is obtained; the base layer node carries out operation collection according to the marking point information output by the expert node, carries out second feature extraction and critical analysis on the obtained marking point operation collection data, then uploads second feature extraction data, and obtains a second feedback instruction output by the expert node; and performing instruction contradiction analysis, performing operation acquisition abnormity judgment and early warning according to instruction contradiction analysis data, and obtaining expert node guidance information, thereby realizing remote communication and guidance for acute and critical diseases.

Description

Technical Field

[0001] This invention proposes a method and system for auxiliary data analysis of critical and emergency patients based on 5G and AR glasses, which relates to the field of data analysis technology, specifically to the field of auxiliary data analysis of critical and emergency patients based on 5G and AR glasses. Background Technology

[0002] The diagnosis and treatment of acute and critical illnesses demands extremely high levels of response speed, accuracy, and resource coordination, but the current system suffers from multi-dimensional bottlenecks. Primary healthcare resources are scarce, with high-quality resources concentrated in cities. Existing telemedicine relies on 4G and ordinary video equipment, resulting in high latency, incomplete information transmission, and susceptibility to misinterpretation of expert guidance. Data transmission and analysis are inefficient, lacking priority scheduling and urgency stratification; critical data is prone to buffering and loss, requiring experts to sift through massive amounts of data, delaying decision-making. Remote command interaction lacks a closed loop, the one-way mode is difficult to dynamically adjust, there is no conflict identification and early warning, and a lack of visual guidance, leading to high operational risks. Summary of the Invention

[0003] This invention provides a method and system for auxiliary data analysis in acute and critical care based on 5G and AR glasses, in order to solve the above-mentioned problems: The present invention proposes a method and system for auxiliary data analysis of acute and critical illnesses based on 5G and AR glasses, the method comprising: S1. Establish 5G wireless connection and data transmission for grassroots nodes, transfer nodes and expert nodes to obtain remote connectivity networks; S2. Obtain the first transfer analysis data through the AR glasses all-in-one machine at the grassroots node, perform feature extraction and feature urgency analysis, upload the extracted features based on the feature urgency analysis results, and then obtain the first feedback instruction output by the expert node. S3. The grassroots node performs operation collection based on the annotation point information output by the expert node, performs second feature extraction and crisis analysis on the obtained annotation point operation collection data, and then uploads the second feature extraction data to obtain the second feedback instruction output by the expert node. S4. Perform instruction conflict analysis based on the first and second feedback instructions, determine and warn of operational anomalies based on the instruction conflict analysis data, and obtain expert node guidance information.

[0004] Further, S1 includes: A remote emergency medical network is constructed by establishing grassroots nodes, transfer nodes, and expert nodes. The system transmits information and instructions through the remote emergency medical network, triggers network node tasks based on the information and instructions transmitted, performs node task operations based on the network node tasks, and obtains node operation information. The remote emergency network is connected based on the node operation information to obtain a remote connection network.

[0005] Furthermore, the construction of a remote emergency medical network through grassroots nodes, transfer nodes, and expert nodes includes: Designate the emergency scene or primary healthcare institution as the primary care node; Hospitals that provide remote guidance capabilities will be designated as expert nodes; The communication service device used in the transfer process between grassroots nodes and expert nodes is defined as the transfer node; The communication service equipment includes an emergency cloud, AR glasses resource nodes, and a CDA acceleration network; Wireless communication connections are established between grassroots nodes, expert nodes, and transfer nodes to create a remote emergency medical network.

[0006] Furthermore, the process of transmitting information commands based on the remote emergency medical network, triggering network node tasks based on the information command transmission, performing node task operations based on the network node tasks, and obtaining node operation information includes: The grassroots nodes send remote collaboration request commands to the transfer nodes via a 5G wireless network; The transfer node generates a collaboration feedback instruction based on the remote collaboration request instruction; Establish a connection between the grassroots node and the transit node according to the collaborative feedback instruction, and obtain the optimal network path; According to the optimal network path, the audio and video information is accessed, and the audio and video transmission is accelerated through the CDA acceleration network of the transfer node, and then transmitted to the glasses management backend and the emergency cloud business platform. The CDA acceleration network will also accelerate the transmission of audio and video information to expert nodes via 5G wireless network; The transfer node and the expert node have a two-way communication connection.

[0007] Further, S2 includes: Audio and video data are collected using an AR glasses all-in-one device, and the collected audio and video data is then uploaded to a grassroots node and subsequently to a remote emergency medical network. By analyzing the vital signs data from the audio and video data collected through the emergency cloud platform, the first transport analysis data can be obtained. Data features are extracted from the first transport analysis data to obtain the first feature-extracted data; The first feature extraction data is subjected to a criticality coefficient analysis to obtain the first feature criticality coefficient; The first feature extraction data is uploaded based on the first feature urgency coefficient. The expert node acquires audio and video acquisition data and first feature extraction data through the remote emergency rescue network, and generates a first feedback command based on the audio and video acquisition data and the first feature extraction data.

[0008] Further, S3 includes: Expert nodes perform AR point selection and frozen screen annotation based on audio and video acquisition data and first feature extraction data to determine annotation point information; The grassroots nodes perform annotation point data operations and collection based on the annotation point information to obtain annotation point operation and collection data; Based on the data collected from the marked points, a second feature analysis is performed to generate a second feedback instruction.

[0009] Furthermore, the step of performing secondary feature analysis and feedback based on the data collected from the marked points to generate a second feedback instruction includes: Data feature extraction is performed on the data collected from the marked points to obtain second feature extraction data; The criticality coefficient of the second feature is obtained by performing a criticality coefficient analysis on the extracted second feature data. Data extracted from the second feature is uploaded based on the second feature's criticality coefficient. The expert node acquires audio and video data and second feature extraction data through the remote emergency rescue network, and generates a second feedback command based on the audio and video data and the second feature extraction data.

[0010] Further, S4 includes: Obtain the first instruction characteristic information of the first feedback instruction and the second instruction characteristic information of the second feedback instruction; The number of contradictory features and the number of non-contradictory features between the first instruction feature information and the second instruction feature information are obtained by semantic similarity judgment method; Obtain the ratio of the number of contradictory features to the number of non-contradictory features to obtain the contradiction ratio coefficient; The contradiction ratio coefficient is compared with the preset contradiction ratio threshold to obtain the contradiction comparison result; Based on the results of the contradiction comparison, anomaly detection is performed on the data collected by the annotation point operation to obtain the anomaly detection results; Based on the anomaly determination result, an early warning is issued for the data collected during the annotation point operation, and annotation point early warning information is obtained; Expert nodes provide guidance on correcting data collected for annotation point operations based on the early warning information of the annotation points.

[0011] Further, the step of comparing the contradiction ratio coefficient with a preset contradiction ratio threshold to obtain the contradiction comparison result includes: When the contradiction ratio coefficient is greater than the preset contradiction ratio threshold, the marked point is judged as abnormal; When the contradiction ratio coefficient is less than or equal to the preset contradiction ratio threshold, the marked point is judged normally.

[0012] Furthermore, the system includes: The connection module is used to enable 5G wireless connection and data transmission between grassroots nodes, transit nodes and expert nodes, and to build a remote connectivity network. The first feedback module is used to obtain the first transfer analysis data through the AR glasses all-in-one machine of the grassroots node, perform feature extraction and feature urgency analysis, upload the extracted features according to the feature urgency analysis results, and then obtain the first feedback instruction output by the expert node. The second feedback module is used by the grassroots nodes to perform operation collection based on the annotation point information output by the expert nodes, perform second feature extraction and crisis analysis on the obtained annotation point operation collection data, and then upload the second feature extraction data to obtain the second feedback instructions output by the expert nodes. The contradiction analysis module is used to perform instruction contradiction analysis based on the first and second feedback instructions, determine and warn of operational anomalies based on the instruction contradiction analysis data, and obtain expert node guidance information.

[0013] The beneficial effects of this invention are as follows: Real-time collaboration between on-site and remote experts is achieved through 5G and AR, addressing the problem of insufficient medical resources at the grassroots level. For example, emergency care in remote areas can quickly obtain guidance from experts at tertiary hospitals, shortening the decision-making time for diagnosis and treatment; phased feature extraction and critical analysis prioritize the uploading of high-risk data (such as sudden drops in heart rate), avoiding bandwidth waste and ensuring priority processing of critical information; a conflicting instruction analysis mechanism can promptly identify conflicting instructions (such as injecting medication versus prohibiting injection), reducing the risk of misoperation and meeting the zero-error requirement for the diagnosis and treatment of acute and critical illnesses; multiple rounds of adjustments from initial diagnosis to practical feedback adapt to the dynamic changes in the patient's condition, avoiding the limitations of single instructions and improving the accuracy of treatment plans. Attached Figure Description

[0014] Figure 1 This is a schematic diagram of an auxiliary data analysis method for acute and critical illnesses based on 5G and AR glasses. Detailed Implementation

[0015] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0016] In one embodiment of the present invention, the present invention proposes a method and system for auxiliary data analysis of acute and critical illnesses based on 5G and AR glasses, the method comprising: S1. Establish 5G wireless connection and data transmission for grassroots nodes, transfer nodes and expert nodes to obtain remote connectivity networks; S2. Obtain the first transfer analysis data through the AR glasses all-in-one machine at the grassroots node, perform feature extraction and feature urgency analysis, upload the extracted features based on the feature urgency analysis results, and then obtain the first feedback instruction output by the expert node. S3. The grassroots node performs operation collection based on the annotation point information output by the expert node, performs second feature extraction and crisis analysis on the obtained annotation point operation collection data, and then uploads the second feature extraction data to obtain the second feedback instruction output by the expert node. S4. Perform instruction conflict analysis based on the first and second feedback instructions, determine and warn of operational anomalies based on the instruction conflict analysis data, and obtain expert node guidance information.

[0017] The working principle and technical effects of the above technical solution are as follows: A cross-space remote communication network is constructed by connecting grassroots nodes (on-site / grassroots institutions), transfer nodes (communication service devices), and expert nodes (remote hospitals) through 5G wireless technology; AR glasses at grassroots nodes collect on-site audio and video data, which is then uploaded after feature extraction and crisis analysis; expert nodes generate the first feedback instruction based on the data; grassroots nodes execute operations and collect data according to expert-annotated point information (such as lesion annotations and operation step annotations), and upload the data after a second feature extraction and crisis analysis; experts generate a second feedback instruction (such as operation adjustment suggestions) based on the new data; semantic similarity analysis is used to compare the contradictory features of the two feedback instructions, calculate the contradiction ratio coefficient and compare it with a threshold, determine operational anomalies and issue warnings, obtain expert correction guidance, and form a complete closed loop of collection, analysis, feedback, and verification.

[0018] By leveraging 5G and AR, real-time collaboration between on-site and remote experts can be achieved, addressing the shortage of primary healthcare resources. For example, emergency care in remote areas can quickly access guidance from experts at top-tier hospitals, shortening the decision-making time for diagnosis and treatment. Phased feature extraction and critical analysis prioritize the uploading of high-risk data (such as sudden drops in heart rate) to avoid bandwidth waste and ensure that critical information is processed first. A conflicting instruction analysis mechanism can promptly identify conflicting instructions (such as injecting medication versus prohibiting injection), reducing the risk of misoperation and meeting the zero-error requirement for the diagnosis and treatment of acute and critical illnesses. Multiple rounds of adjustments from initial diagnosis to practical feedback adapt to the dynamic changes in the patient's condition, avoiding the limitations of single instructions and improving the accuracy of treatment plans.

[0019] In one embodiment of the present invention, S1 includes: A remote emergency medical network is constructed by establishing grassroots nodes, transfer nodes, and expert nodes. The system transmits information and instructions through the remote emergency medical network, triggers network node tasks based on the information and instructions transmitted, performs node task operations based on the network node tasks, and obtains node operation information. The remote emergency network is connected based on the node operation information to obtain a remote connection network.

[0020] The working principle and technical effects of the above technical solution are as follows: Integrating three types of nodes—grassroots, transport, and expert—and clarifying the functional boundaries of each node (e.g., the transport node is responsible for communication services), a remote emergency rescue network infrastructure is formed; based on network transmission information instructions (e.g., collaboration requests from grassroots nodes), corresponding node tasks are triggered (e.g., route planning tasks from transport nodes and instruction generation tasks from expert nodes); after each node executes its task, it generates operation information (e.g., route results from transport nodes and instruction feedback from expert nodes); the system verifies network connectivity based on the operation information (e.g., whether data is transmitted smoothly and whether instructions are received effectively), and determines that the remote connection network is available.

[0021] By implementing task operation and connectivity verification processes, subsequent medical treatment is prevented from being hindered due to network interruptions or node failures, ensuring network stability during emergency care. Clear node task triggering logic avoids resource idleness or task conflicts (e.g., transfer nodes prioritize emergency data processing and do not occupy non-emergency bandwidth), improving network resource utilization. Compared to ordinary network setups, the addition of task triggering and connectivity verification closed loops allows for rapid identification of network problems (e.g., a fault is indicated if a node does not provide operational feedback), reducing network troubleshooting time during emergency care.

[0022] In one embodiment of the present invention, the construction of a remote emergency medical network through grassroots nodes, transfer nodes, and expert nodes includes: Designate the emergency scene or primary healthcare institution as the primary care node; Hospitals that provide remote guidance capabilities will be designated as expert nodes; The communication service device used in the transfer process between grassroots nodes and expert nodes is defined as the transfer node; The communication service equipment includes an emergency cloud, AR glasses resource nodes, and a CDA acceleration network; Wireless communication connections are established between grassroots nodes, expert nodes, and transfer nodes to create a remote emergency medical network.

[0023] The working principle and technical effects of the above technical solution are as follows: Based on the emergency and critical care process, the on-site execution end is defined as the grassroots node (emergency site / grassroots institution), the remote decision-making end is defined as the expert node (hospital providing guidance), and the communication relay end is defined as the transfer node (communication service device between the grassroots and experts); the transfer node includes the emergency cloud (data storage and preliminary analysis), the AR glasses resource node (AR device management and data adaptation), and the CDA acceleration network (data transmission acceleration), forming a communication service system of storage, adaptation, and acceleration; the three types of nodes are connected through 5G wireless technology, and the data transmission link (such as data from grassroots AR glasses to the CDA acceleration network to the emergency cloud to the expert node) is tested to ensure that data can be transmitted smoothly in both directions between each node, thus completing the construction of the remote emergency network. To avoid overlapping node functions (e.g., transport nodes do not participate in diagnosis and treatment decisions, but are only responsible for communication), reduce coordination chaos, and improve the processing efficiency of each link; the combination of CDA acceleration network and emergency cloud can reduce data transmission latency, and AR glasses resource nodes ensure device compatibility (e.g., data from different brands of AR glasses can be uniformly adapted), solving the pain points of slow data transmission and device incompatibility; node definition and component design (e.g., transport nodes can add edge computing modules) facilitate the access of more devices (e.g., portable ultrasound machines) or the expansion of functions (e.g., AI-assisted diagnosis).

[0024] In one embodiment of the present invention, the step of transmitting information commands according to a remote emergency rescue network, triggering network node tasks according to the information command transmission, performing node task operations according to the network node tasks, and obtaining node operation information includes: The grassroots node (AR glasses all-in-one device) sends a remote collaboration request command to the transfer node (emergency cloud) via a 5G wireless network; The transfer node (emergency cloud) generates a collaborative feedback instruction based on the remote collaboration request instruction; Based on the collaborative feedback instructions, a connection is established between the base node (AR glasses all-in-one device) and the transfer node (AR glasses resource node) to obtain the optimal network path; the optimal network path is the path with the best average transmission speed and quality.

[0025] According to the optimal network path, the audio and video information is accessed and the audio and video transmission is accelerated through the CDA acceleration network of the transfer node (AR glasses resource node) and transmitted to the glasses management backend and the emergency cloud business platform. The CDA acceleration network also accelerates the transmission of audio and video information to expert nodes (electronic terminals) via 5G wireless networks; the content in parentheses is one of the components of the corresponding node.

[0026] The transfer node (the management backend of the glasses and the emergency cloud business platform) and the expert node (electronic terminal) have a two-way communication connection.

[0027] The working principle and technical effects of the above technical solution are as follows: The AR glasses all-in-one device at the grassroots node sends a remote collaboration request (such as requiring expert guidance for patient hemostasis) to the emergency cloud at the transfer node via the 5G network; after receiving the request, the emergency cloud analyzes the current network status (such as the bandwidth and latency of each transmission path), generates a collaboration feedback instruction, guides the AR glasses all-in-one device to establish a connection with the AR glasses resource node at the transfer node, and determines the path with the optimal average transmission speed and quality; the AR glasses all-in-one device accesses audio and video data through the optimal path, which is processed by the CDA acceleration network of the AR glasses resource node (such as data compression and multi-link concurrent transmission), and synchronously transmitted to the glasses end management backend (device control) and the emergency cloud business platform (data storage and analysis). At the same time, the CDA acceleration network transmits the data to the electronic terminal (such as expert computer or tablet) at the expert node via 5G; the management backend and business platform of the transfer node establish two-way communication with the electronic terminal of the expert node (such as expert instructions to the transfer node to the grassroots node, and grassroots operation data to the transfer node to the expert node), ensuring real-time information exchange. Optimal path selection and CDA acceleration can control the latency of audio and video data transmission within a clinically acceptable range, avoiding misjudgments by experts due to data lag (such as not being able to see lesion details); average optimal path planning can balance the load of each path and avoid congestion of a single path. For example, when transmitting ECG data and imaging data simultaneously, priority is given to ensuring the bandwidth of ECG data (high priority), improving resource utilization; a two-way communication mechanism ensures that expert instructions are implemented quickly (such as adjusting the intubation angle can be immediately transmitted to the primary care unit), and the primary care unit's operational data is fed back in real time, meeting the time-sensitive needs of critical and severe cases; data is synchronized to the management backend, business platform, and expert terminal, realizing multi-terminal collaboration in equipment control, data storage, and expert decision-making, avoiding data silos (such as experts not being able to see the equipment status, resulting in instructions not being able to be executed).

[0028] In one embodiment of the present invention, S2 includes: Audio and video data are collected using an AR glasses all-in-one device, and the collected audio and video data is then uploaded to a grassroots node and subsequently to a remote emergency medical network. By analyzing the vital signs data from the audio and video data collected through the emergency cloud platform, the first transport analysis data can be obtained. Data features are extracted from the first transport analysis data to obtain the first feature-extracted data; The first feature extraction data is subjected to urgency coefficient analysis to obtain the first feature urgency coefficient; the urgency coefficient is determined by the feature preset priority. The first feature extraction data is uploaded based on the first feature urgency coefficient. The expert node acquires audio and video acquisition data and first feature extraction data through the remote emergency rescue network, and generates a first feedback command based on the audio and video acquisition data and the first feature extraction data.

[0029] The feedback instructions are transmitted to the glasses management backend of the transfer node via the remote emergency rescue network; The glasses management backend controls the functions of the AR glasses all-in-one device through a remote emergency network and obtains function control information.

[0030] The working principle and technical effects of the above technical solution are as follows: The AR glasses all-in-one device at the grassroots node collects patient audio-visual data (such as facial expressions, wound condition, and vital sign monitoring images), uploads it to local storage at the grassroots node, and then synchronizes it to the remote emergency network; the emergency cloud platform receives the audio-visual data, performs vital sign analysis using AI algorithms (such as image recognition to extract wound area and heart rate monitoring data), and generates first transport analysis data (such as patient heart rate 120 beats / min, open wound on the left upper limb); key features (such as abnormal heart rate, open wound) are extracted from the first transport analysis data, and based on... The system calculates the urgency coefficient based on preset priorities (e.g., abnormal vital signs > trauma characteristics). For example, if the heart rate exceeds the normal range by 30%, the coefficient is 0.8, indicating high urgency. High-risk characteristics (e.g., abnormal heart rate) are uploaded first. Expert nodes combine audio and video data with high-risk characteristics to generate the first feedback instruction (e.g., administer oxygen and monitor heart rate changes). The instruction is transmitted remotely to the management backend of the glasses at the transport node. The management backend adjusts the AR glasses' functions according to the instruction (e.g., enable the real-time heart rate display module and zoom in on the wound monitoring screen) and generates function control information to ensure that primary healthcare workers operate efficiently according to the instructions. The AR glasses all-in-one device integrates audio and video acquisition functions, eliminating the need to carry multiple additional devices (such as cameras and monitors), making it suitable for emergency scenarios with limited space and tight schedules. Emergency cloud AI performs preliminary analysis, replacing manual screening and quickly extracting key data (such as identifying abnormal heart rates within 10 seconds), preventing primary care medical staff from wasting time on data processing during rescue efforts. Instructions are generated based on high-risk and urgency characteristics, allowing experts to focus on core issues (such as prioritizing life-threatening abnormal heart rates), avoiding redundant instructions (such as mentioning multiple low-risk issues simultaneously). The AR glasses' functions dynamically adjust according to instructions (such as automatically activating the corresponding module if the instruction requires heart rate monitoring), reducing the number of steps required by medical staff (such as eliminating the need to manually switch function interfaces) and improving operational efficiency.

[0031] In one embodiment of the present invention, S3 includes: Expert nodes perform AR point selection and frozen screen annotation based on audio and video acquisition data and first feature extraction data to determine annotation point information; The grassroots nodes perform annotation point data operations and collection based on the annotation point information to obtain annotation point operation and collection data; Based on the data collected from the marked points, a second feature analysis is performed to generate a second feedback instruction.

[0032] The step of performing secondary feature analysis and feedback based on the data collected from the marked points to generate a second feedback instruction includes: Data feature extraction is performed on the data collected from the marked points to obtain second feature extraction data; The criticality coefficient of the second feature is obtained by performing a criticality coefficient analysis on the extracted second feature data. Data extracted from the second feature is uploaded based on the second feature's criticality coefficient. The expert node acquires audio and video data and second feature extraction data through the remote emergency rescue network, and generates a second feedback command based on the audio and video data and the second feature extraction data.

[0033] The working principle and technical effect of the above technical solution are as follows: Based on the audio and video data of S2 and the first feature extraction data (such as lesion location, abnormal physiological indicators), the expert node performs point selection and annotation (such as marking wound hemostasis points) and freeze-screen annotation (such as annotating operation steps on static images) through the AR system to generate annotation point information; the primary care node medical staff performs operations according to the annotation point information (such as pressing to stop bleeding at the marked hemostasis points), and at the same time collects operation process data through AR glasses (such as pressure feedback, wound hemostasis effect image), forming annotation point operation collection data; the second feature is extracted from the operation collection data (such as wound bleeding decreases after pressing, patient blood pressure rises), and the second feature criticality coefficient is calculated (such as the criticality coefficient drops from 0.9 to 0.3 if bleeding decreases); the second feature extraction data is uploaded to the expert node, and the expert combines the initial data, operation data, and secondary features to judge the operation effect (such as whether hemostasis is effective) and generates a second feedback instruction (such as continuing to press for 5 minutes, and adjusting the pressing pressure if blood pressure is stable). AR annotation points provide visual guidance for primary healthcare workers, avoiding operational errors caused by lack of experience (e.g., enabling novice healthcare workers to accurately locate hemostasis points) and reducing the error rate. Through a two-stage cycle of annotation, practice, and feedback, expert guidance is transformed into verifiable operational results, avoiding the problem of no follow-up after instructions are issued and ensuring the effective implementation of treatment plans. Secondary feature extraction can reflect the impact of operations on the patient's condition in real time (e.g., changes in blood pressure after hemostasis), avoiding decision-making bias based on static data. During the execution of annotation operations, primary healthcare workers can learn from the expert annotation logic (e.g., why this hemostasis point was chosen), indirectly improving primary healthcare capabilities and alleviating resource shortages in the long term.

[0034] In one embodiment of the present invention, S4 includes: Obtain the first instruction characteristic information of the first feedback instruction and the second instruction characteristic information of the second feedback instruction; The number of contradictory features and the number of non-contradictory features between the first instruction feature information and the second instruction feature information are obtained by semantic similarity judgment method; Obtain the ratio of the number of contradictory features to the number of non-contradictory features (make it 1 when the denominator is 0) to obtain the contradiction ratio coefficient; The contradiction ratio coefficient is compared with the preset contradiction ratio threshold to obtain the contradiction comparison result; Based on the results of the contradiction comparison, anomaly detection is performed on the data collected by the annotation point operation to obtain the anomaly detection results; Based on the anomaly determination result, an early warning is issued for the data collected during the annotation point operation, and annotation point early warning information is obtained; Expert nodes provide guidance on correcting data collected for annotation point operations based on the early warning information of the annotation points.

[0035] The step of comparing the contradiction ratio coefficient with a preset contradiction ratio threshold to obtain the contradiction comparison result includes: When the contradiction ratio coefficient is greater than the preset contradiction ratio threshold, the marked point is judged as abnormal; When the contradiction ratio coefficient is less than or equal to the preset contradiction ratio threshold, the marked point is judged normally.

[0036] The threshold was determined based on expert experience.

[0037] The working principle and technical effect of the above technical solution are as follows: Extract the first instruction feature information of the first feedback instruction (e.g., oxygen inhalation, heart rate monitoring) and the second instruction feature information of the second feedback instruction (e.g., continue compression to stop bleeding, adjust compression pressure), and decompose them into comparable feature items (e.g., operation type, parameter requirements, contraindications); use a semantic similarity algorithm (e.g., BERT model) to compare the feature items, and count the number of contradictory features (e.g., stopping compression and continuing compression are contradictory features) and the number of non-contradictory features (e.g., monitoring heart rate and monitoring blood pressure are non-contradictory features); rank the number of contradictory features by the number of non-contradictory features. The system calculates the coefficient for the number of shield features (the coefficient is 1 when the denominator is 0, i.e., the coefficient = 1 if there are no consistent features); the coefficient is compared with the preset contradiction ratio threshold (e.g., 0.3, which can be adjusted according to clinical risk). If the coefficient > the threshold (e.g., 0.4 > 0.3), the operation is judged as abnormal and an early warning is triggered (e.g., a red warning prompt pops up on the AR glasses); the system pushes the abnormal judgment result and contradictory features (e.g., conflicting press commands) to the expert node. The expert analyzes the cause of the conflict (e.g., the operation effect did not meet expectations, leading to command adjustment) and generates correction guidance information (e.g., continue pressing according to the second command, while strengthening heart rate monitoring). The system proactively identifies conflicting instructions (such as injecting drug A and prohibiting drug A), preventing medical staff from misoperating due to unawareness of conflicts. This is particularly suitable for scenarios involving multiple rounds of instruction adjustments, reducing safety risks. By quantifying the conflict ratio coefficient and comparing it with a threshold, it replaces subjective human judgment, avoiding omissions in conflict identification due to differences in experience (such as novice medical staff failing to detect hidden conflicting instructions). Shield analysis and early warning are completed in real time (e.g., analysis is completed within 1 second after instruction generation), meeting the needs of rapid response in acute and critical illnesses and avoiding delays in optimal treatment. The system synchronously pushes conflict characteristics, allowing experts to quickly locate conflict points (e.g., conflict only in pressure intensity instructions) without needing to re-examine all instructions, shortening correction guidance time and improving efficiency.

[0038] According to one embodiment of the present invention, the system includes: The connection module is used to enable 5G wireless connection and data transmission between grassroots nodes, transit nodes and expert nodes, and to build a remote connectivity network. The first feedback module is used to obtain the first transfer analysis data through the AR glasses all-in-one machine of the grassroots node, perform feature extraction and feature urgency analysis, upload the extracted features according to the feature urgency analysis results, and then obtain the first feedback instruction output by the expert node. The second feedback module is used by the grassroots nodes to perform operation collection based on the annotation point information output by the expert nodes, perform second feature extraction and crisis analysis on the obtained annotation point operation collection data, and then upload the second feature extraction data to obtain the second feedback instructions output by the expert nodes. The contradiction analysis module is used to perform instruction contradiction analysis based on the first and second feedback instructions, determine and warn of operational anomalies based on the instruction contradiction analysis data, and obtain expert node guidance information.

[0039] The working principle and technical effects of the above technical solution are as follows: It is responsible for configuring the 5G wireless connection of three types of nodes (such as network parameter settings and node permission allocation), testing data transmission links, building and maintaining a remote connection network to provide a communication foundation for subsequent modules; controlling the grassroots AR glasses to collect audio and video data, calling the emergency cloud AI algorithm for feature extraction and crisis analysis, filtering and uploading data according to the crisis coefficient, and receiving and forwarding the first feedback instructions from expert nodes to the grassroots nodes; receiving the annotation point information from expert nodes and pushing it to the grassroots level, controlling the AR glasses to collect operation process data, performing secondary feature extraction and crisis analysis, uploading data and receiving the second feedback instructions from experts; calling the semantic similarity model to compare the features of the two instructions, calculating the contradiction ratio coefficient, comparing it with a preset threshold to determine an operational anomaly, triggering an early warning mechanism, and simultaneously pushing the anomaly information to expert nodes, receiving and forwarding expert correction guidance information, thus completing the closed loop. Each module is functionally independent with clear interfaces, allowing for individual upgrades (e.g., adding 6G compatibility to the connectivity module and optimizing the algorithm in the conflict analysis module), reducing overall system upgrade costs and adapting to technological iterations. In the event of a module failure (e.g., the first feedback module is temporarily unable to upload data), other modules can be temporarily adjusted (e.g., the connectivity module prioritizes basic communication, and the conflict analysis module suspends judgment), preventing system-wide paralysis and meeting the high reliability requirements of emergency scenarios. Modules are designed with reserved interfaces (e.g., the first feedback module can access blood glucose monitoring data), allowing for the addition of new functions (e.g., an AI-assisted diagnostic module) without system reconstruction, adapting to different acute and critical care scenarios (e.g., trauma emergency care, cardiovascular emergency care). Module collaboration is highly automated (e.g., feature extraction is automatically triggered after data collection), requiring no complex operations from primary care physicians; they only need to follow AR prompts, lowering the barrier to entry and making it suitable for the tense environment of emergency scenes.

[0040] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A method for auxiliary data analysis of acute and critical illnesses based on 5G and AR glasses, characterized in that, The method includes: S1. Establish 5G wireless connection and data transmission for grassroots nodes, transfer nodes and expert nodes to obtain remote connectivity networks; S2. Obtain the first transfer analysis data through the AR glasses all-in-one machine at the grassroots node, perform feature extraction and feature urgency analysis, upload the extracted features based on the feature urgency analysis results, and then obtain the first feedback instruction output by the expert node. S3. The grassroots node performs operation collection based on the annotation point information output by the expert node, performs second feature extraction and crisis analysis on the obtained annotation point operation collection data, and then uploads the second feature extraction data to obtain the second feedback instruction output by the expert node. S4. Perform instruction conflict analysis based on the first and second feedback instructions, determine and warn of operational anomalies based on the instruction conflict analysis data, and obtain expert node guidance information.

2. The method for auxiliary data analysis of acute and critical illnesses based on 5G and AR glasses according to claim 1, characterized in that, S1 includes: A remote emergency medical network is constructed by establishing grassroots nodes, transfer nodes, and expert nodes. The system transmits information and instructions through the remote emergency medical network, triggers network node tasks based on the information and instructions transmitted, performs node task operations based on the network node tasks, and obtains node operation information. The remote emergency network is connected based on the node operation information to obtain a remote connection network.

3. The method for auxiliary data analysis of acute and critical illnesses based on 5G and AR glasses according to claim 1, characterized in that, The construction of a remote emergency medical network through grassroots nodes, transfer nodes, and expert nodes includes: Designate the emergency scene or primary healthcare institution as the primary care node; Hospitals that provide remote guidance capabilities will be designated as expert nodes; The communication service device used in the transfer process between grassroots nodes and expert nodes is defined as the transfer node; The communication service equipment includes an emergency cloud, AR glasses resource nodes, and a CDA acceleration network; Wireless communication connections are established between grassroots nodes, expert nodes, and transfer nodes to create a remote emergency medical network.

4. The method for auxiliary data analysis of acute and critical illnesses based on 5G and AR glasses according to claim 1, characterized in that, The process of transmitting information commands according to the remote emergency medical network, triggering network node tasks based on the information command transmission, performing node task operations based on the network node tasks, and obtaining node operation information includes: The grassroots nodes send remote collaboration request commands to the transfer nodes via a 5G wireless network; The transfer node generates a collaboration feedback instruction based on the remote collaboration request instruction; Establish a connection between the grassroots node and the transit node according to the collaborative feedback instruction, and obtain the optimal network path; According to the optimal network path, the audio and video information is accessed, and the audio and video transmission is accelerated through the CDA acceleration network of the transfer node, and then transmitted to the glasses management backend and the emergency cloud business platform. The CDA acceleration network will also accelerate the transmission of audio and video information to expert nodes via 5G wireless network; The transfer node and the expert node have a two-way communication connection.

5. The method for auxiliary data analysis of acute and critical illnesses based on 5G and AR glasses according to claim 1, characterized in that, S2 includes: Audio and video data are collected using an AR glasses all-in-one device, and the collected audio and video data is then uploaded to a grassroots node and subsequently to a remote emergency medical network. By analyzing the vital signs data from the audio and video data collected through the emergency cloud platform, the first transport analysis data can be obtained. Data features are extracted from the first transport analysis data to obtain the first feature-extracted data; The first feature extraction data is subjected to a criticality coefficient analysis to obtain the first feature criticality coefficient; The first feature extraction data is uploaded based on the first feature urgency coefficient. The expert node acquires audio and video acquisition data and first feature extraction data through the remote emergency rescue network, and generates a first feedback command based on the audio and video acquisition data and the first feature extraction data.

6. The method for auxiliary data analysis of acute and critical illnesses based on 5G and AR glasses according to claim 1, characterized in that, S3 includes: Expert nodes perform AR point selection and frozen screen annotation based on audio and video acquisition data and first feature extraction data to determine annotation point information; The grassroots nodes perform annotation point data operations and collection based on the annotation point information to obtain annotation point operation and collection data; Based on the data collected from the marked points, a second feature analysis is performed to generate a second feedback instruction.

7. The method for auxiliary data analysis of acute and critical illness based on 5G and AR glasses according to claim 6, characterized in that, The step of performing secondary feature analysis and feedback based on the data collected from the marked points to generate a second feedback instruction includes: Data feature extraction is performed on the data collected from the marked points to obtain second feature extraction data; The criticality coefficient of the second feature is obtained by performing a criticality coefficient analysis on the extracted second feature data. Data extracted from the second feature is uploaded based on the second feature's criticality coefficient. The expert node acquires audio and video data and second feature extraction data through the remote emergency rescue network, and generates a second feedback command based on the audio and video data and the second feature extraction data.

8. The method for auxiliary data analysis of acute and critical illnesses based on 5G and AR glasses according to claim 1, characterized in that, S4 includes: Obtain the first instruction characteristic information of the first feedback instruction and the second instruction characteristic information of the second feedback instruction; The number of contradictory features and the number of non-contradictory features between the first instruction feature information and the second instruction feature information are obtained by semantic similarity judgment method; Obtain the ratio of the number of contradictory features to the number of non-contradictory features to obtain the contradiction ratio coefficient; The contradiction ratio coefficient is compared with the preset contradiction ratio threshold to obtain the contradiction comparison result; Based on the results of the contradiction comparison, anomaly detection is performed on the data collected by the annotation point operation to obtain the anomaly detection results; Based on the anomaly determination result, an early warning is issued for the data collected during the annotation point operation, and annotation point early warning information is obtained; Expert nodes provide guidance on correcting data collected for annotation point operations based on the early warning information of the annotation points.

9. The method for auxiliary data analysis of acute and critical illness based on 5G and AR glasses according to claim 8, characterized in that, The step of comparing the contradiction ratio coefficient with a preset contradiction ratio threshold to obtain the contradiction comparison result includes: When the contradiction ratio coefficient is greater than the preset contradiction ratio threshold, the marked point is judged as abnormal; When the contradiction ratio coefficient is less than or equal to the preset contradiction ratio threshold, the marked point is judged normally.

10. A critical care auxiliary data analysis system based on 5G and AR glasses, characterized in that, The system includes: The connection module is used to enable 5G wireless connection and data transmission between grassroots nodes, transit nodes and expert nodes, and to build a remote connectivity network. The first feedback module is used to obtain the first transfer analysis data through the AR glasses all-in-one machine of the grassroots node, perform feature extraction and feature urgency analysis, upload the extracted features according to the feature urgency analysis results, and then obtain the first feedback instruction output by the expert node. The second feedback module is used by the grassroots nodes to perform operation collection based on the annotation point information output by the expert nodes, perform second feature extraction and crisis analysis on the obtained annotation point operation collection data, and then upload the second feature extraction data to obtain the second feedback instructions output by the expert nodes. The contradiction analysis module is used to perform instruction contradiction analysis based on the first and second feedback instructions, determine and warn of operational anomalies based on the instruction contradiction analysis data, and obtain expert node guidance information.

Images