Intelligent first-aid grading method and system based on large-scale language model

Abstract

The invention provides an intelligent first-aid grading method and system based on a large-scale language model, and relates to the technical field of large-scale language models.The intelligent first-aid grading method comprises the steps that first-level multi-dimensional data of a patient is obtained, a three-party description language model is constructed, primary feature processing is conducted according to the three-party description language model, and primary feature multi-dimensional data is obtained; performing comprehensive analysis on the primary feature multi-dimensional data to obtain a route planning matching degree and a primary reasoning disease severity index, and generating a pre-estimation treatment decision according to the route planning matching degree and the primary reasoning disease severity index; acquiring secondary multi-dimensional data of the patient, constructing an autonomous description language model, and performing secondary feature processing according to the autonomous description language model to obtain secondary feature multi-dimensional data; the secondary feature multi-dimensional data is subjected to feature processing, the real-time route planning matching degree and the secondary reasoning disease severity index are obtained, the final treatment decision is selected, the first-aid preparation time is greatly shortened, and the first-aid efficiency is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of large-scale language models, and in particular to an intelligent first aid grading method and system based on large-scale language models. Background Art

[0002] Current mainstream emergency triage systems mostly use decision tree models based on rule engines. This type of technology has significant limitations: first, it relies on structured data input and cannot effectively parse complex symptoms described by patients in natural language; second, it lacks dynamic learning capabilities and has difficulty adapting to the rapid evolution of symptoms of new diseases; third, the existing system exposes insufficient multimodal data processing capabilities and cannot effectively integrate heterogeneous data sources such as voice descriptions and imaging data.

[0003] Therefore, an intelligent first aid classification method and system based on a large-scale language model is provided. Summary of the Invention

[0004] In order to solve the above technical problems, the purpose of the present invention is to provide an intelligent first aid classification method and system based on a large-scale language model.

[0005] To achieve the above objectives, the present invention provides the following technical solution: an intelligent first aid classification method based on a large-scale language model, comprising: Acquire the primary multidimensional data of the patient, construct a three-party description language model, perform feature processing based on the three-party description language model, and obtain feature multidimensional data; Performing a comprehensive analysis on the primary feature multi-dimensional data to obtain a route planning matching degree and a first-level reasoning severity index, and generating an estimated treatment decision based on the route planning matching degree and the first-level reasoning severity index; Acquiring secondary multidimensional data of the patient, constructing an autonomous description language model, performing secondary feature processing according to the autonomous description language model, and obtaining secondary feature multidimensional data; Perform feature processing on the secondary feature multidimensional data to obtain the real-time route planning matching degree and the secondary reasoning disease severity index, and select the final treatment decision.

[0006] Furthermore, the process of obtaining the patient's primary multidimensional data and secondary multidimensional data includes: Set up the data acquisition unit; The data acquisition unit is used to receive conversations between medical staff and third-party personnel or emergency patients; based on natural language processing technology, the data of conversations between medical staff and third-party personnel are recorded as first-level multidimensional data; and the data of conversations between medical staff and emergency patients are recorded as second-level multidimensional data.

[0007] Furthermore, the process of building a tripartite description language model includes: Based on convolutional network neural networks, a standard description language model is constructed; Obtain the first-level multidimensional data; group and label the sample first-level multidimensional data, and record them as is a natural number; Will The sample level one multidimensional data is grouped as sample data, and is less than The natural number of is recorded as the sample set; the remaining group of sample level one multidimensional data is used as the test set; The sample level multidimensional data The corresponding preset result is recorded as , the verification result corresponding to the standard description language model is , according to the corresponding preset result And verification results , and obtain the verification error rate ; Preset error rate ; like , then the verification result of the standard description language model meets the standard, there is no need to adjust the parameters of the standard description language model, and it is recorded as a three-party description language model; like , then the verification result of the neural network model does not meet the standard, and the standard description language model needs to be adjusted, and the model after the standard description language model is adjusted is recorded as the three-party description language model.

[0008] Furthermore, a feature processing is performed based on the tripartite description language model to obtain feature multi-dimensional data, including the following process: Decomposing the patient's consciousness into three sub-items based on the Glasgow Coma Scale: eye opening response index, language response index, and motor response index, performing normalization processing on each sub-item, and mapping the three sub-items to a standardized interval of 0-1; Wound characteristics are encoded based on the injury location and wound size, and the injury location is mapped to the three-dimensional coordinates of the wound based on the anatomical coordinate system; the wound size is converted into area, aspect ratio and depth index; Based on the ID number, patient age, gender, and place of origin, the patient's family history, allergy history, and historical medical records are obtained. Based on the tripartite description language model, it is linked to the SNOMED CT knowledge graph to obtain the patient's genetic risk coefficient, drug cross-allergy matrix, and historical medical record time decay factor; Calculate the accessibility score of medical resources based on the Logistic function; The eye opening reaction index, language reaction index, motor reaction index, wound three-dimensional coordinates, area, aspect ratio, depth index, genetic risk coefficient, drug cross-allergy matrix, historical medical history time attenuation factor and accessibility score are recorded as primary feature multidimensional data.

[0009] Furthermore, the process of comprehensive analysis of the multi-dimensional data of a feature includes: The eye opening reaction index, language reaction index, and motor reaction index are evaluated to obtain a neurological status assessment; the three-dimensional coordinates, area, aspect ratio, and depth index of the wound are evaluated to obtain a wound characteristic assessment; based on the drug cross-allergy matrix, the maximum drug cross-allergy risk is obtained; Obtaining a first-level inference severity index based on the neurological status assessment, wound characteristic assessment, genetic risk coefficient, maximum drug cross-allergy risk, and historical medical history time attenuation factor; A route planning matching degree is obtained according to the accessibility score.

[0010] Furthermore, based on the route planning matching degree and the first-level inference severity index, the specific process of generating an estimated treatment decision includes: Preset severity index grading interval , , , ; If the first-level inference severity index , by contacting the local community hospital in advance; If the first-level inference severity index , by contacting the local secondary hospital or specialist center in advance; If the first-level inference severity index , by contacting the local comprehensive tertiary hospital in advance; If the first-level inference severity index , then immediately complete the preparations for the first-level emergency response; Generate several routes, obtain the route planning matching degree of each route; and select the route with the largest route planning matching degree as the best route.

[0011] Furthermore, the process of constructing an autonomous description language model for secondary feature processing includes: Preset error rate ; like , the verification error rate of the standard description language model meets the standard, the verification result of the standard description language model meets the standard, and there is no need to adjust the parameters of the standard description language model; like , the verification error rate of the standard description language model does not meet the standard, the verification result of the neural network model does not meet the standard, and the standard description language model needs to be adjusted; the standard description language model is recorded as an autonomous description language model; Obtain secondary multidimensional data of patients; Filter and interpolate abnormal values ​​in the patient's body temperature, heart rate, respiratory rate, blood pressure, and blood oxygen saturation, and obtain circulatory failure risk and respiratory compensation index based on the autonomous description language model; Real-time traffic conditions are obtained through the API, and the congestion level is mapped to a time penalty term, recorded as the traffic impact factor; weather impact is obtained as a weight to reduce the accessibility score, recorded as the environmental impact factor; the slope is calculated through the elevation API, and the time taken by the ambulance to climb the slope is corrected, recorded as the terrain complexity; A real-time dynamic accessibility correction index is obtained based on the traffic impact factor, environmental impact factor and terrain complexity; the real-time dynamic accessibility correction index, circulatory failure risk and respiratory compensation index are recorded as secondary feature multidimensional data.

[0012] Furthermore, a comprehensive analysis of the secondary feature multidimensional data is performed to obtain the real-time route planning matching degree and the secondary reasoning severity index. The process of selecting the final treatment decision includes: Obtain the corresponding first-level reasoning severity index; The secondary feature multidimensional data is introduced as a weight item into the primary reasoning disease severity index to obtain the secondary reasoning disease severity index ; The real-time dynamic accessibility correction index is introduced into the route planning matching degree to generate the final route planning matching degree. ; Preset severity index grading interval , , , ; Preset optimal route planning matching degree ; If the secondary inference severity index and , the patient will be sent directly to the local community hospital for treatment; If the secondary inference severity index and , the patient will be sent directly to a local secondary hospital or specialist center for treatment; If the secondary inference severity index and , the patient will be sent directly to the local comprehensive tertiary hospital for treatment; If the secondary inference severity index , the patient will be given first aid immediately and sent to the comprehensive tertiary hospital with the best matching degree of the final route planning for treatment.

[0013] In order to achieve at least one of the above-mentioned objects, the present invention further provides an intelligent first aid classification system based on a large-scale language model, wherein the system implements the above-mentioned intelligent first aid classification method based on a large-scale language model, and comprises a data acquisition module, a data processing module, a data analysis module, and an intelligent first aid module; Data acquisition module, used to obtain the patient's primary multidimensional data and secondary multidimensional data; A data processing module is used to perform feature processing on the primary multidimensional data and the secondary multidimensional data to obtain primary feature multidimensional data and secondary feature multidimensional data; Data analysis module, used for analyzing and processing primary feature multidimensional data and secondary feature multidimensional data; The intelligent management module is used to obtain estimated treatment decisions and final treatment decisions based on the results of analysis and processing.

[0014] The present invention further provides a computer-readable storage medium storing a computer program, which can be executed by a processor to implement the above-mentioned intelligent first aid classification method based on a large-scale language model.

[0015] Compared with the prior art, the present invention has the following beneficial effects: 1. By acquiring the patient's primary and secondary multidimensional data, and constructing a tripartite description language model and an autonomous description language model for feature processing, the system can quickly and systematically analyze patient information, evaluate the condition from multiple dimensions, generate estimated treatment decisions, and select final treatment decisions, thus reducing the time spent on information analysis and decision-making, and greatly shortening the preparation time for emergency treatment.

[0016] 2. Comprehensively analyze the primary and secondary feature multidimensional data to obtain key information such as route planning matching degree and condition severity index. This information provides a scientific basis for the generation and selection of treatment decisions, enabling more reasonable allocation and utilization of emergency resources. Based on accurate assessment of the patient's condition, it improves the efficiency of the entire emergency process. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, a brief introduction to the drawings required for use in the embodiments will be given below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.

[0018] Figure 1Schematic diagram of an intelligent first aid classification method based on a large-scale language model.

[0019] Figure 2 A schematic diagram of an intelligent first aid classification process based on a large-scale language model.

[0020] Figure 3 A module diagram of an intelligent first aid classification system based on a large-scale language model. DETAILED DESCRIPTION

[0021] The following description is intended to disclose the present invention so that those skilled in the art can implement the present invention. The preferred embodiments described below are for illustrative purposes only, and those skilled in the art will readily appreciate other obvious variations. The basic principles of the present invention defined in the following description may be applied to other embodiments, variations, improvements, equivalents, and other technical solutions that do not depart from the spirit and scope of the present invention.

[0022] It is to be understood that the term "one" should be understood as "at least one" or "one or more", that is, in one embodiment, the number of an element may be one, while in another embodiment, the number of the elements may be multiple, and the term "one" should not be understood as a limitation on the quantity.

[0023] like Figure 1 As shown, an intelligent first aid classification method based on a large-scale language model includes the following steps: Acquire the primary multidimensional data of the patient, construct a three-party description language model, perform feature processing based on the three-party description language model, and obtain feature multidimensional data; Performing a comprehensive analysis on the primary feature multi-dimensional data to obtain a route planning matching degree and a first-level reasoning severity index, and selecting an estimated treatment decision based on the route planning matching degree and the first-level reasoning severity index; Acquiring secondary multidimensional data of the patient, constructing an autonomous description language model, performing secondary feature processing according to the autonomous description language model, and obtaining secondary feature multidimensional data; Comprehensively analyze the secondary feature multidimensional data to obtain the real-time route planning matching degree and the secondary reasoning disease severity index, and select the final treatment decision.

[0024] It should be further explained that, in the specific implementation process, the specific process of obtaining the patient's primary multidimensional data and secondary multidimensional data includes: Set up the data acquisition unit; The data collection unit is used to receive conversations between medical personnel and third-party personnel or emergency patients; based on natural language processing (NLP) technology, extract data that has a substantial effect in the conversations between medical personnel and third-party personnel or emergency patients, and record the data that has a substantial effect in the conversations between medical personnel and third-party personnel as first-level multidimensional data; and record the data that has a substantial effect in the conversations between medical personnel and emergency patients as second-level multidimensional data; The first-level multi-dimensional data includes vital sign data, identity information data, and geographic location data; the second-level multi-dimensional data includes real-time vital sign data and real-time geographic location data; The vital signs data include the patient's consciousness, injury location, wound size, etc.; the identity information data include the patient's age, gender, place of origin, etc.; the real-time vital signs data include the patient's body temperature, heart rate, respiratory rate, blood pressure, and blood oxygen saturation, etc.

[0025] It should be further explained that, in the specific implementation process, the specific process of building the tripartite description language model includes: Based on convolutional network neural networks, a standard description language model is constructed; It should be further explained that the number of iterations of the standard description language model is set to N, and the number of hidden layers is set to A; the number of iterations is set to 900<N<4000, and the number of hidden layers is set to 10<A<30; It should be further explained that if the number of iterations and the number of hidden layers of the neural network model are set too large, the model will become complex, the calculation data will be huge, the calculation efficiency will be low, and even the generalization ability of the model will be poor. If the number of iterations of the neural network model is too many or too few and the number of hidden layers is set too large or too small, the model will have a large prediction error and overfitting will occur. Setting the number of iterations to 900<N<4000 and the number of hidden layers to 10<A<30 ensures that the neural network model has good prediction accuracy and avoids overfitting. The specific process of setting the number of iterations and the number of hidden layers includes: Obtaining publicly available, varying degrees of completeness, first-level multidimensional data of conversations between medical staff and third-party personnel; it should be further clarified that the first-level multidimensional data includes sample vital signs data, sample identity information data, and sample geographic location data; The first-level multidimensional data of the samples with different degrees of completeness are grouped and labeled as is a natural number; Will The sample level one multidimensional data is grouped as sample data, and is less than The natural number of is recorded as the sample set; the remaining group of sample level one multidimensional data is used as the test set; The sample level multidimensional data The corresponding preset result is recorded as , the verification result corresponding to the standard description language model is , according to the corresponding preset result is recorded as And verification results , calculate the verification error rate of the standard description language model , the verification error rate for: ; Preset error rate ; like , the verification error rate of the standard description language model meets the standard, the verification result of the standard description language model meets the standard, and there is no need to adjust the parameters of the standard description language model; like , the verification error rate of the standard description language model does not meet the standard, the verification result of the neural network model does not meet the standard, and the standard description language model needs to be adjusted; The model after adjusting the parameters of the standard description language model is recorded as the three-party description language model.

[0026] It should be further explained that, in a specific implementation process, a feature processing is performed based on the three-party description language model to obtain a feature multi-dimensional data. The specific process includes: Obtaining the patient's first-level multidimensional data; The first-level multidimensional data is input into the three-party description language model for feature processing. The specific process of feature processing includes: Perform feature processing on vital signs data. The specific process includes: Based on the Glasgow Coma Scale (GCS), the patient's consciousness is decomposed into three sub-items: eye opening reaction index, language reaction index, and motor reaction index, each of which is normalized, and the three sub-items are mapped to a standardized interval of 0-1; Wound characteristics are encoded based on the injury location and size. The injury location is mapped to three-dimensional wound coordinates based on the anatomical coordinate system (ATC). It should be further explained that the three-dimensional wound coordinates are normalized relative to a standard human body model. The wound size is converted into area, aspect ratio, and depth index. It should be further explained that the depth index is estimated based on a color recognition algorithm. Perform feature processing on identity information data. The specific process includes: Based on data such as ID number, patient age, gender, and place of origin, the patient's family history, allergy history, and historical medical history are obtained. Based on the tripartite description language model, the model is linked to the SNOMED CT knowledge graph to obtain the patient's genetic risk coefficient, drug cross-allergy matrix, and historical medical history time decay factor. It should be further explained that the historical medical history time decay factor reflects the strength of the association between the historical medical history and the current condition. The specific process of processing geographic location data features includes: Calculating the accessibility score of medical resources based on the Logistic function , the accessibility score for: ; in is the actual distance, k is the distance sensitivity parameter, and k 0; The eye opening reaction index, language reaction index, motor reaction index, wound three-dimensional coordinates, area, aspect ratio, depth index, genetic risk coefficient, drug cross-allergy matrix, historical medical history time attenuation factor and accessibility score are recorded as primary feature multidimensional data.

[0027] It should be further explained that, in the specific implementation process, the specific process of comprehensive analysis of the multi-dimensional data of a feature includes: The eye opening reaction index, language reaction index, and motor reaction index are evaluated to obtain the neurological status assessment GCS, which is: ;in, Eye opening response index, The language response index, is the motor response index; The three-dimensional coordinates, area, aspect ratio, and depth index of the wound are evaluated to obtain a wound characteristic evaluation SS, which is: ;in, The three-dimensional coordinates of the wound, For area, is the aspect ratio, is the depth index; Based on the drug cross-allergy matrix, obtaining the maximum drug cross-allergy risk; The first-level inference severity index is obtained based on the neurological status assessment GCS, wound characteristic assessment SS, genetic risk coefficient, maximum drug cross-allergy risk and historical medical history time attenuation factor. , the first-level inference severity index for: ; in, is the genetic risk factor, The maximum risk of drug cross-allergy, is the historical medical record time decay factor, 、 、 、 、 is the weight coefficient; According to the accessibility score , get the route planning matching degree , the route planning matching degree for: ;in, For hospital matching, is the estimated travel time; 、 、 is the weight coefficient.

[0028] It should be further explained that, in the specific implementation process, the specific process of generating an estimated treatment decision based on the route planning matching degree and the first-level reasoning severity index includes: Preset severity index grading interval , , , ; If the first-level inference severity index , by contacting the local community hospital in advance; If the first-level inference severity index , by contacting the local secondary hospital or specialist center in advance; If the first-level inference severity index , by contacting the local comprehensive tertiary hospital in advance; If the first-level inference severity index , immediately complete the preparations for the first-level emergency response (such as cardiopulmonary resuscitation and intubation); Generate several routes, obtain the route planning matching degree of each route; and select the route with the largest route planning matching degree as the best route.

[0029] It should be further explained that, in the specific implementation process, the specific process of building an autonomous description language model includes: describing the language model according to the standard; Preset error rate ; like , the verification error rate of the standard description language model meets the standard, the verification result of the standard description language model meets the standard, and there is no need to adjust the parameters of the standard description language model; like , the verification error rate of the standard description language model does not meet the standard, the verification result of the neural network model does not meet the standard, and the standard description language model needs to be adjusted; The standard description language model is recorded as an autonomous description language model.

[0030] It should be further explained that, in the specific implementation process, the specific process of performing secondary feature processing according to the autonomous description language model includes: Obtain secondary multidimensional data of patients; The secondary multi-dimensional data is input into the autonomous description language model for feature processing. The specific process of feature processing includes: Perform feature processing on real-time vital signs data. The specific process includes: Filter and interpolate abnormal values ​​in the patient's body temperature, heart rate, respiratory rate, blood pressure, and blood oxygen saturation. It should be further explained that the filtering reduces the risk of data anomalies or true criticality; the interpolation is based on linear interpolation, and temporarily missing data is filled with the mean of the previous and next values ​​to improve accuracy. Obtain circulatory failure risk based on autonomous description language model , the risk of circulatory failure for: ; Obtain respiratory compensation index , the respiratory compensation index for: ; Perform feature processing on real-time geographic location data. The specific process includes: Obtain real-time traffic conditions through the API and map the congestion level (0-1) to a time penalty term, which is recorded as the traffic impact factor; Obtain weather impacts, such as extreme weather such as rain / blizzard, as a weight to reduce accessibility scores and record them as environmental impact factors; The slope is calculated through the elevation API, and the time taken by the ambulance to climb the slope is corrected and recorded as terrain complexity; A real-time dynamic accessibility correction index is obtained based on the traffic impact factor, environmental impact factor and terrain complexity; the real-time dynamic accessibility correction index, circulatory failure risk and respiratory compensation index are recorded as secondary feature multidimensional data.

[0031] It should be further explained that, during the specific implementation process, the comprehensive analysis of the secondary feature multidimensional data, the acquisition of the real-time route planning matching degree and the secondary reasoning severity index, and the selection of the final treatment decision are as follows: Obtain the corresponding first-level reasoning severity index; The secondary feature multidimensional data is introduced as a weight item into the primary reasoning disease severity index to obtain the secondary reasoning disease severity index ; The secondary reasoning severity index for: ;in, 、 is the weight coefficient; Introducing the real-time dynamic accessibility correction index into the route planning matching degree , generate the final route planning matching degree for: ;in, It is the real-time dynamic accessibility correction index; Preset severity index grading interval , , , ; Preset optimal route planning matching degree ; If the secondary inference severity index and , the patient will be sent directly to the local community hospital for treatment; If the secondary inference severity index and , the patient will be sent directly to a local secondary hospital or specialist center for treatment; If the secondary inference severity index and , the patient will be sent directly to the local comprehensive tertiary hospital for treatment; If the secondary inference severity index , the patient will be given first aid immediately and sent to the comprehensive tertiary hospital with the best matching degree of the final route planning for treatment.

[0032] like Figure 2 As shown, an intelligent first aid classification system based on a large-scale language model includes a data acquisition module, a data processing module, a data analysis module and an intelligent first aid module; The data acquisition module is used to obtain the primary multidimensional data and the secondary multidimensional data of the patient; The data processing module is used to perform feature processing on the primary multidimensional data and the secondary multidimensional data to obtain primary feature multidimensional data and secondary feature multidimensional data; The data analysis module is used to analyze and process the primary feature multidimensional data and the secondary feature multidimensional data; The intelligent management module is used to obtain an estimated treatment decision and a final treatment decision based on the results of the analysis and processing.

[0033] The embodiments disclosed in the present invention, the processes described above with reference to the flowcharts can be implemented as computer software programs. The embodiments disclosed in the present invention include a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes a program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from the network through the communication part, and / or installed from a removable medium. When the computer program is executed by the central processing unit (CPU), the above-mentioned functions defined in the method of the present application are executed. It should be noted that the computer-readable medium mentioned above in the present application can be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. The computer-readable storage medium can be, for example, but not limited to, a system, device or component of an electrical, magnetic, optical, electromagnetic, infrared segment, or semiconductor, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to, an electrical connection having one or more wire segments, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In this application, a computer-readable storage medium may be any tangible medium that contains or stores a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. In this application, a computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, which carries computer-readable program code. Such a propagated data signal may take a variety of forms, including, but not limited to, an electromagnetic signal, an optical signal, or any suitable combination thereof. A computer-readable signal medium may also be any computer-readable medium other than a computer-readable storage medium that can transmit, propagate, or transfer a program for use by or in conjunction with an instruction execution system, apparatus, or device. Program code embodied on a computer-readable medium may be transmitted using any appropriate medium, including but not limited to wireless, electrical wire, optical fiber cable, RF, etc., or any suitable combination thereof.

[0034] The flow charts and block diagrams in the accompanying drawings illustrate the possible implementation architecture, functions and operations of the system, method and computer program product according to various embodiments of the present invention. In this regard, each box in the flow chart or block diagram can represent a module, program segment, or a part of code, and the module, program segment, or a part of code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order than that marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or flow chart, and the combination of the boxes in the block diagram and / or flow chart, can be implemented with a dedicated hardware-based system that performs the specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0035] Those skilled in the art should understand that the embodiments of the present invention described above and shown in the accompanying drawings are only examples and do not limit the present invention. The objectives of the present invention have been fully and effectively achieved. The functional and structural principles of the present invention have been demonstrated and explained in the embodiments. Without departing from the principles, the implementation methods of the present invention may be subject to any deformation or modification.

Claims

1. An intelligent first aid classification method based on a large-scale language model, characterized in that: include: Acquire the primary multidimensional data of the patient, construct a three-party description language model, perform feature processing based on the three-party description language model, and obtain feature multidimensional data; Performing a comprehensive analysis on the primary feature multi-dimensional data to obtain a route planning matching degree and a first-level reasoning severity index, and generating an estimated treatment decision based on the route planning matching degree and the first-level reasoning severity index; Acquiring secondary multidimensional data of the patient, constructing an autonomous description language model, performing secondary feature processing according to the autonomous description language model, and obtaining secondary feature multidimensional data; Perform feature processing on the secondary feature multidimensional data to obtain the real-time route planning matching degree and the secondary reasoning disease severity index, and select the final treatment decision.

2. The intelligent first aid classification method based on a large-scale language model according to claim 1, characterized in that: The process of obtaining the patient's primary multidimensional data and secondary multidimensional data includes: Set up the data acquisition unit; The data acquisition unit is used to receive conversations between medical staff and third-party personnel or emergency patients; based on natural language processing technology, the data of conversations between medical staff and third-party personnel are recorded as first-level multidimensional data; and the data of conversations between medical staff and emergency patients are recorded as second-level multidimensional data.

3. The intelligent first aid classification method based on a large-scale language model according to claim 2, characterized in that: The process of building a tripartite description language model includes: Based on convolutional network neural networks, a standard description language model is constructed; Obtain the first-level multidimensional data; group and label the sample first-level multidimensional data, and record them as is a natural number; Will The sample level one multidimensional data is grouped as sample data, and is less than The natural number of is recorded as the sample set; the remaining group of sample level one multidimensional data is used as the test set; The sample level multidimensional data The corresponding preset result is recorded as , the verification result corresponding to the standard description language model is , according to the corresponding preset result And verification results , and obtain the verification error rate ; Preset error rate ; like , then the verification result of the standard description language model meets the standard, there is no need to adjust the parameters of the standard description language model, and it is recorded as a three-party description language model; like , then the verification result of the neural network model does not meet the standard, and the standard description language model needs to be adjusted, and the model after the standard description language model is adjusted is recorded as the three-party description language model.

4. The intelligent first aid classification method based on a large-scale language model according to claim 3, characterized in that: The process of performing feature processing based on the tripartite description language model to obtain feature multidimensional data includes: Decomposing the patient's consciousness into three sub-items based on the Glasgow Coma Scale: eye opening response index, language response index, and motor response index, performing normalization processing on each sub-item, and mapping the three sub-items to a standardized interval of 0-1; Wound characteristics are encoded based on the injury location and wound size, and the injury location is mapped to the three-dimensional coordinates of the wound based on the anatomical coordinate system; the wound size is converted into area, aspect ratio and depth index; Based on the ID number, patient age, gender, and place of origin, the patient's family history, allergy history, and historical medical records are obtained. Based on the tripartite description language model, it is linked to the SNOMED CT knowledge graph to obtain the patient's genetic risk coefficient, drug cross-allergy matrix, and historical medical record time decay factor; Calculate the accessibility score of medical resources based on the Logistic function; The eye opening reaction index, language reaction index, motor reaction index, wound three-dimensional coordinates, area, aspect ratio, depth index, genetic risk coefficient, drug cross-allergy matrix, historical medical history time attenuation factor and accessibility score are recorded as primary feature multidimensional data.

5. The intelligent first aid classification method based on a large-scale language model according to claim 4, characterized in that: The process of comprehensive analysis of multidimensional data of a feature includes: The eye opening reaction index, language reaction index, and motor reaction index are evaluated to obtain a neurological status assessment; the three-dimensional coordinates, area, aspect ratio, and depth index of the wound are evaluated to obtain a wound characteristic assessment; based on the drug cross-allergy matrix, the maximum drug cross-allergy risk is obtained; Obtaining a first-level inference severity index based on the neurological status assessment, wound characteristic assessment, genetic risk coefficient, maximum drug cross-allergy risk, and historical medical history time attenuation factor; A route planning matching degree is obtained according to the accessibility score.

6. The intelligent first aid classification method based on a large-scale language model according to claim 5, characterized in that: The specific process of generating an estimated treatment decision based on the route planning matching degree and the first-level inference severity index includes: Preset severity index grading interval , , , ; If the first-level inference severity index , by contacting the local community hospital in advance; If the first-level inference severity index , by contacting the local secondary hospital or specialist center in advance; If the first-level inference severity index , by contacting the local comprehensive tertiary hospital in advance; If the first-level inference severity index , then immediately complete the preparations for the first-level emergency response; Generate several routes, obtain the route planning matching degree of each route; and select the route with the largest route planning matching degree as the best route.

7. The intelligent first aid classification method based on a large-scale language model according to claim 6, characterized in that: The process of building an autonomous description language model for secondary feature processing includes: Preset error rate ; like , the verification error rate of the standard description language model meets the standard, the verification result of the standard description language model meets the standard, and there is no need to adjust the parameters of the standard description language model; like , the verification error rate of the standard description language model does not meet the standard, the verification result of the neural network model does not meet the standard, and the standard description language model needs to be adjusted; the standard description language model is recorded as an autonomous description language model; Obtain secondary multidimensional data of patients; Filter and interpolate abnormal values ​​in the patient's body temperature, heart rate, respiratory rate, blood pressure, and blood oxygen saturation, and obtain circulatory failure risk and respiratory compensation index based on the autonomous description language model; Real-time traffic conditions are obtained through the API, and the congestion level is mapped to a time penalty term, recorded as the traffic impact factor; weather impact is obtained as a weight to reduce the accessibility score, recorded as the environmental impact factor; the slope is calculated through the elevation API, and the time taken by the ambulance to climb the slope is corrected, recorded as the terrain complexity; A real-time dynamic accessibility correction index is obtained based on the traffic impact factor, environmental impact factor and terrain complexity; the real-time dynamic accessibility correction index, circulatory failure risk and respiratory compensation index are recorded as secondary feature multidimensional data.

8. The intelligent first aid classification method based on large-scale language model according to claim 7 is characterized in that: Comprehensively analyzing the secondary feature multidimensional data to obtain the real-time route planning matching degree and the secondary reasoning severity index, and selecting the final treatment decision process includes: Obtain the corresponding first-level reasoning severity index; The secondary feature multidimensional data is introduced as a weight item into the primary reasoning disease severity index to obtain the secondary reasoning disease severity index ; The real-time dynamic accessibility correction index is introduced into the route planning matching degree to generate the final route planning matching degree. ; Preset severity index grading interval , , , ; Preset optimal route planning matching degree ; If the secondary inference severity index and , the patient will be sent directly to the local community hospital for treatment; If the secondary inference severity index and , the patient will be sent directly to a local secondary hospital or specialist center for treatment; If the secondary inference severity index and , the patient will be sent directly to the local comprehensive tertiary hospital for treatment; If the secondary inference severity index , the patient will be given first aid immediately and sent to the comprehensive tertiary hospital with the best matching degree of the final route planning for treatment.

9. An intelligent first aid classification system based on a large-scale language model, the system implementing the intelligent first aid classification method based on a large-scale language model according to any one of claims 1 to 8, comprising a data acquisition module, a data processing module, a data analysis module, and an intelligent first aid module; Data acquisition module, used to obtain the patient's primary multidimensional data and secondary multidimensional data; A data processing module is used to perform feature processing on the primary multidimensional data and the secondary multidimensional data to obtain primary feature multidimensional data and secondary feature multidimensional data; Data analysis module, used for analyzing and processing primary feature multidimensional data and secondary feature multidimensional data; The intelligent management module is used to obtain estimated treatment decisions and final treatment decisions based on the results of analysis and processing.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a computer program, and the computer program can be executed by a processor to implement the intelligent first aid classification method based on a large-scale language model as described in any one of claims 1 to 8.

Images