Rapid triage classification method for sudden group events and medium

Abstract

The invention relates to a rapid triage classification method for sudden group events and a medium, and relates to the technical field of medical information. Collecting vital sign data of each patient based on the intelligent wearable device, collecting external trauma information of the patient with the external trauma based on the image collection device, and associating and uploading the collected vital sign data and external trauma information with identity information; for any patient, based on the collected vital sign data or the vital sign data and the external trauma information, performing injury condition analysis to obtain an injury condition level; performing priority ranking treatment based on the injury level of each patient; for any patient with multiple external trauma, the external trauma severity order is obtained based on the external trauma information, and priority order treatment is performed based on the external trauma severity order, so that the injury severity of each patient in the sudden group event can be more accurately judged, and the treatment efficiency is improved. And the condition of illness delay caused by lack of timely treatment is reduced.

Description

Technical Field

[0001] The present invention relates to the field of medical information technology, and in particular to a rapid injury classification method and medium for sudden mass incidents. Background Art

[0002] In sudden mass incidents (such as natural disasters, large-scale accidents, etc.), emergency departments are faced with the urgent treatment of a large number of injured people, whose types and injuries are complex and changeable. Traditional injury assessment and classification usually rely on manual judgment and subjective experience, which is prone to problems such as delayed judgment, misclassification, and uneven resource allocation, affecting the efficiency of treatment.

[0003] Existing injury classification methods mostly rely on simple manual assessments or data based on biosensors, but these methods are single, fail to effectively combine multiple data sources, and lack intelligent and automated decision support.

[0004] Therefore, there is an urgent need for an intelligent system that can efficiently and accurately classify the casualties and provide real-time decision support in sudden mass incidents, especially in the emergency handling of sudden mass incidents, which can improve the efficiency of casualty treatment, reduce the waste of medical resources, and increase the success rate of rescue. Summary of the invention

[0005] The technical problem to be solved by the present application is to provide a method and medium for rapid injury classification in sudden mass incidents, which has the characteristics of being able to more accurately judge the severity of the injuries of each patient in a sudden mass incident and reduce the delay in the condition due to lack of timely treatment.

[0006] In a first aspect, an embodiment provides a method for rapid triage of sudden mass incidents, including: Collect the identity information of each patient; Collect each patient's vital signs data based on smart wearable devices, and associate the collected vital signs data with identity information and upload them; Collecting external trauma information of patients with external trauma based on an image acquisition device, and associating the collected external trauma information with identity information and uploading the information; the external trauma information includes trauma image information and the human body location where the trauma is located; For any patient, based on the collected vital signs data, or vital signs data and external trauma information, injury analysis is performed to obtain the injury level; priority sorting and treatment are performed based on the injury level of each patient; For any patient with multiple external injuries, the severity ranking of the external injuries is obtained based on the external trauma information, and priority treatment is performed based on the severity ranking of the external injuries.

[0007] In one embodiment, the step of performing injury analysis to obtain the injury level of any patient based on the collected vital sign data, or the vital sign data and external trauma information, includes: For any patient, when the collected data only includes vital signs data, the injury level of the patient is identified based on the vital signs data through the first injury level identification model; the first injury level identification model is a neural network model trained in combination with the first loss function, which identifies the injury level based on the characteristics of the vital signs data.

[0008] In one embodiment, the step of performing injury analysis to obtain the injury level of any patient based on the collected vital sign data, or the vital sign data and external trauma information, includes: For any patient, when the collected data includes vital signs data and external trauma information, the injury level of the patient is identified based on the vital signs data and the external trauma information through a second injury level identification model; the second injury level identification model is a neural network model trained in combination with the second loss function, which identifies the injury level based on the characteristics of the vital signs data and the external trauma information.

[0009] In one embodiment, the identification of the injury level of any one of the patients based on the vital sign data and the external trauma information by using the second injury level identification model includes: Identify the wound image information in the external wound information to obtain the wound type, wound depth and wound range; Based on the identified wound type, wound depth and trauma range, combined with the human body location of the trauma and vital signs data, the injury level of any patient is identified through the second injury level identification model.

[0010] In one embodiment, the injury level includes red, yellow, green and black from high to low, where red indicates that there is a risk of life and emergency treatment is required immediately; yellow indicates that the injury is serious but not life-threatening; green indicates that the injury is relatively minor and vital signs are stable; black indicates death or incurable.

[0011] In one embodiment, for patients whose injury level has not reached the preset injury level, the patient's vital signs data are continuously collected within a preset time period and the injury level is obtained. If the preset injury level is still not reached when the collection time meets the preset time period, the highest injury level obtained within the preset time period is used as the highest injury level for treatment.

[0012] In one embodiment, an alarm is issued to patients whose injury level reaches a preset injury level.

[0013] In one embodiment, for any patient with multiple external injuries, obtaining the external injury severity ranking based on the external injury information includes: Based on the trauma image information and the human body location information of the trauma, the trauma severity of each trauma is identified through a trauma severity recognition model; the trauma severity recognition model is a neural network model trained in combination with the third loss function, which identifies the severity of the trauma based on the trauma image information and the human body location information characteristics.

[0014] In one embodiment, the identification of the severity of each injury based on the injury image information and the human body location information of the injury by using an injury severity identification model includes: For any wound, the wound image information in the external wound information is identified to obtain the wound type, wound depth and wound range; Based on the identified wound type, wound depth and wound range, combined with the human body location of the wound, the wound severity of any wound is identified through a wound severity identification model; Based on the severity of each identified injury, the severity of external injuries is ranked.

[0015] In a second aspect, an embodiment provides a computer-readable storage medium, wherein the medium stores a program, and the program can be loaded by a processor to execute the method for rapid triage of sudden mass incidents described in any one of the above embodiments.

[0016] The beneficial effects of the present invention are: Since the vital signs data of each patient is collected based on the smart wearable device, and the external trauma information of patients with external trauma is collected based on the image acquisition device, for any patient, based on the collected vital signs data, or, vital signs data and external trauma information, the injury level is obtained by performing injury analysis, so that the injury level of each patient can be obtained more accurately, so that patients with high injury levels can be given priority treatment, and the situation that people with more serious conditions cannot receive timely treatment and delay their condition can be reduced. For patients with lower injury levels who do not meet the needs of timely treatment or are sent to the hospital for treatment, the waste of medical resources can be reduced. For any patient with multiple external traumas, the severity ranking of the external trauma is obtained based on the external trauma information, and priority treatment is performed based on the severity ranking of the external trauma, so that when a patient has multiple diseases or traumas, the parts that need priority treatment can be more accurately judged, and the situation that more serious traumas cannot receive timely treatment and delay their condition can be reduced. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 This is a flow chart of a rapid injury triage method for sudden mass incidents according to an embodiment of the present application; Figure 2 This is a flow chart of a method for identifying the injury level of any patient according to an embodiment of the present application; Figure 3 It is a flowchart of a method for identifying the severity of various injuries according to an embodiment of the present application. DETAILED DESCRIPTION

[0018] The present invention is further described in detail below by specific embodiments in conjunction with the accompanying drawings. Wherein similar elements in different embodiments adopt associated similar element numbers. In the following embodiments, many detailed descriptions are for making the present application better understood. However, those skilled in the art can easily recognize that some features can be omitted in different situations, or can be replaced by other elements, materials, methods. In some cases, some operations related to the present application are not shown or described in the specification, this is to avoid the core part of the present application being overwhelmed by too much description, and for those skilled in the art, it is not necessary to describe these related operations in detail, and they can fully understand the related operations according to the description in the specification and the general technical knowledge in the art.

[0019] In addition, the features, operations or characteristics described in the specification can be combined in any appropriate manner to form various implementations. At the same time, the steps or actions in the method description can also be interchanged or adjusted in a manner that is obvious to those skilled in the art. Therefore, the various sequences in the specification and the drawings are only for the purpose of clearly describing a certain embodiment and are not meant to be a required sequence, unless otherwise specified that a certain sequence must be followed.

[0020] The serial numbers assigned to the components in this article, such as "first", "second", etc., are only used to distinguish the objects described and do not have any order or technical meaning.

[0021] To facilitate the description of the inventive concept of the present application, the first aid techniques in sudden mass incidents are briefly described below.

[0022] If a sudden mass incident occurs, such as a natural disaster or a large-scale accident, there are often a large number of injuries. In the current sudden mass incidents, all the injured people are generally sent to the hospital for examination and treatment. However, the applicant found in the study that, on the one hand, due to insufficient medical resources, the people who are given priority for treatment are often those with more serious injuries on the surface of the body, or those whose physical condition is seriously affected by visual observation. However, the severity is often difficult to accurately judge from the surface condition. Therefore, people with more serious conditions often cannot receive timely treatment and their condition is delayed. On the other hand, some people with injuries may not be in serious condition and do not need to be sent to the hospital for emergency treatment, but medical resources are wasted due to the inability to accurately and quickly judge. On the third hand, when a patient has multiple diseases or injuries, which disease or injury should be pre-treated first? In the case of insufficient experience of medical staff, it is often impossible to accurately judge, which will also cause more serious injuries to be unable to receive timely treatment and delay the condition.

[0023] In view of this, the present application provides a method and medium for rapid injury triage of sudden mass events, which collects vital sign data of each patient based on smart wearable devices, collects external trauma information of patients with external trauma based on image acquisition devices, and performs injury analysis to obtain the injury level for any patient based on the collected vital sign data, or vital sign data and external trauma information. In this way, the injury level of each patient can be obtained more accurately, so that patients with high injury levels can be given priority treatment, reducing the situation where people with more serious conditions cannot receive timely treatment and delay the condition. For patients with lower injury levels who do not meet the needs of timely treatment or are sent to the hospital for treatment, the waste of medical resources can be reduced. In addition, for any patient with multiple external traumas, the severity ranking of the external trauma is obtained based on the external trauma information, and priority ranking treatment is performed based on the severity ranking of the external trauma. In this way, when a patient has multiple diseases or traumas, the parts that need priority treatment can be more accurately judged, reducing the situation where more serious traumas cannot receive timely treatment and delay the condition.

[0024] In one embodiment, the present application provides a method for rapid triage of sudden mass incidents, please refer to Figure 1 ,include: Step S10, collecting the identity information of each patient.

[0025] In one embodiment, the identity information may include any combination of name, ID number, and telephone number, which may represent the patient's identity.

[0026] Step S20: collect vital sign data of each patient based on the smart wearable device, and associate the collected vital sign data with the identity information and upload it.

[0027] Those skilled in the art can understand that the smart wearable device here can be any existing smart wearable device that can collect vital sign data, such as smart bracelets, portable heart rate monitors, blood oximeters, etc. The collected vital sign data may include heart rate, blood pressure, blood oxygen saturation, body temperature, etc. The collected vital sign data is associated with the identity information and uploaded, so as to obtain the vital sign data of each patient with the same identity.

[0028] Step S30: collect external trauma information of the patient with external trauma based on the image acquisition device, and associate the collected external trauma information with the identity information and upload it. The external trauma information includes trauma image information and the human body location where the trauma is located.

[0029] In one embodiment, the image acquisition device may be an intelligent image acquisition terminal with a camera such as a mobile phone or a tablet, or may be a dedicated device.

[0030] Based on the collected trauma image information, the wound type, wound depth and trauma range can be obtained through image feature recognition, where the wound type may include lacerations, hemorrhagic wounds and open wounds.

[0031] Step S40: For any patient, based on the collected vital sign data, or the vital sign data and external trauma information, an injury analysis is performed to obtain the injury level; and priority sorting and treatment are performed based on the injury level of each patient.

[0032] During the study, the applicant found that in sudden mass incidents, some of the individuals involved had obvious surface trauma, but some did not. Some even showed neither any trauma nor any symptoms affecting vital signs in the early stages. Therefore, situations requiring emergency treatment or intervention were not discovered in time, resulting in irreparable situations without timely treatment.

[0033] In view of this, the vital signs data of each individual subject in the sudden group event is collected. For example, for a dozen people involved in a car accident, the vital signs data of each person is collected. For any of them, if there is no obvious external trauma, only the vital signs data is collected. If there is obvious external trauma, the external trauma information is also collected.

[0034] In one embodiment, for any patient, when the collected data only includes vital sign data, the injury level of the patient is identified based on the vital sign data through a first injury level identification model. The first injury level identification model is a neural network model trained in combination with the first loss function, which identifies the injury level based on the characteristics of the vital sign data.

[0035] In one embodiment, a training method for the first injury level recognition model includes: Step S110, collecting vital sign data of multiple patients and marking the vital sign data to clarify the injury level.

[0036] Extensive collection of vital sign data corresponding to various injury levels, such as heart rate, blood pressure, blood oxygen saturation, body temperature, etc. Data sources can be the hospital's electronic medical record system, emergency center records, and related medical research databases, etc., to ensure that the data is sufficiently diverse and representative.

[0037] The collected data is labeled with the injury level according to professional medical standards and clinical experience. The labeling process must be participated in by professional medical staff or medical experts to ensure the accuracy of the labeling.

[0038] In one embodiment, the injury level may include red, yellow, green and black from high to low, where red indicates that there is a danger to life and emergency treatment is required immediately; yellow indicates that the injury is serious but not life-threatening; green indicates that the injury is relatively minor and vital signs are stable; black indicates death or cannot be treated.

[0039] For the above-mentioned red, yellow, green and black injury levels, those skilled in the art can understand that they can be graded based on actual demand standards. For example, for the red injury level, the applicable situations usually include cardiac and respiratory arrest, severe airway obstruction, massive and uncontrollable bleeding, severe craniocerebral injury accompanied by coma, severe chest trauma causing breathing difficulties, etc. If these situations are not handled in time, the injured may die in a short time. For the yellow injury level, the applicable situations may include multiple fractures, large-area burns without respiratory injury, severe soft tissue injury, etc. Such injured persons need to be treated within a certain period of time, but they can wait a little and let the red-level injured persons receive treatment first. For the green injury level, common ones include mild abrasions, sprains, small-area superficial burns, etc. These injured persons can usually be simply treated on the spot, or further treated after the red and yellow injured persons are treated. For the black injury level, the applicable situations include when the injured person has obvious signs of death, such as dilated and fixed pupils, the appearance of livor mortis, and severe body damage that cannot be effectively treated, etc., and will be marked as black. For this type of injured people, the main focus is on identity confirmation and subsequent treatment, rather than medical treatment.

[0040] Step S210, pre-processing the labeled data, converting it into a format suitable for input of the first injury level recognition model, and obtaining a data set.

[0041] Step S210 may include: cleaning the original data to remove duplicates, errors and missing values, and then performing normalization or standardization to map data with different features to the same scale range to improve the training efficiency and stability of the model.

[0042] Step S310, dividing the data set into a training set, a validation set and a test set.

[0043] Step S410, constructing the primary architecture of the first injury level recognition model, based on preset training parameters, using the training set as the input of the primary architecture of the first injury level recognition model for training, evaluating the performance of the first injury level recognition model based on the validation set during the training process and adjusting the hyperparameters, and evaluating the generalization ability of the first injury level recognition model based on the test set, thereby obtaining a trained first injury level recognition model.

[0044] In one embodiment, the primary architecture of the first injury level recognition model can be constructed based on a multi-layer perceptron (MLP) and a long short-term memory network (LSTM), so that a more accurate injury level can be obtained based on continuous vital sign data collected in real time.

[0045] In one embodiment, the first loss function may adopt a cross entropy loss function.

[0046] In one embodiment, for any patient, when the collected data includes vital sign data and external trauma information, the injury level of the patient is identified based on the vital sign data and the external trauma information through a second injury level identification model. The second injury level identification model is a neural network model trained in combination with the second loss function, which identifies the injury level based on the characteristics of the vital sign data and the characteristics of the external trauma information.

[0047] In one embodiment, please refer to Figure 2 Based on the vital signs data and external trauma information, the second injury level identification model is used to identify the injury level of any patient, including: Step S100, identifying the wound image information in the external wound information to obtain the wound type, wound depth and wound range.

[0048] Based on the wound image information, image recognition can be performed to obtain the wound type, wound depth and wound range. Among them, the wound type can include laceration, hemorrhagic wound and open wound.

[0049] Step S200, based on the identified wound type, wound depth and wound range, combined with the human body location of the wound and vital signs data, the injury level of any patient is identified through a second injury level identification model.

[0050] In one embodiment, the training method of the second injury level recognition model includes: Step S120 , collecting wound type, wound depth, wound range, human body location of the wound and vital sign data of multiple patients, and marking them to clarify the injury level.

[0051] Step S220, pre-processing the labeled data, converting it into a format suitable for input into the second injury level recognition model, and obtaining a data set.

[0052] Step S220 may include: cleaning the original data to remove duplicates, errors and missing values, and then performing normalization or standardization to map data with different features to the same scale range to improve the training efficiency and stability of the model.

[0053] Step S320, dividing the data set into a training set, a validation set and a test set.

[0054] Step S420, construct the primary architecture of the second injury level recognition model, and based on preset training parameters, use the training set as the input of the primary architecture of the second injury level recognition model for training, evaluate the performance of the second injury level recognition model based on the validation set during the training process and adjust the hyperparameters, and evaluate the generalization ability of the second injury level recognition model based on the test set, so as to obtain a trained second injury level recognition model.

[0055] In one embodiment, the primary architecture of the second injury level recognition model can be constructed based on a multi-branch convolutional neural network (CNN) and a recurrent neural network (RNN).

[0056] In one embodiment, the second loss function can be obtained by weighted summing the classification loss and various auxiliary losses. Among them, the classification loss function can adopt the cross entropy loss function, and the auxiliary loss function can adopt the multimodal consistency loss function and the position loss function. Auxiliary losses for different modal data are introduced into the model to promote the model's fusion of multimodal information. For example, for wound features, trauma location features, and vital signs features, their consistency losses in the feature space are calculated so that the model can learn the association between different modalities. According to the potential relationship between the trauma location and the injury level, a position constraint loss is designed. For example, trauma in certain key locations may have a greater impact on the injury level, and the loss function is used to constrain the model's learning of these position information.

[0057] In this way, since the vital signs data of each patient is collected based on the smart wearable device, and the external trauma information of patients with external trauma is collected based on the image acquisition device, for any patient, an injury analysis is performed to obtain the injury level based on the collected vital signs data, or the vital signs data and external trauma information, so that the injury level of each patient can be obtained more accurately, so that patients with high injury levels can be given priority, reducing the situation where people with more serious conditions cannot receive timely treatment and their condition is delayed. For patients with lower injury levels who do not meet the needs of timely treatment or need to be sent to the hospital for treatment, the waste of medical resources can be reduced.

[0058] The applicant also found in the study that in sudden mass incidents, especially car accidents, some of the individual subjects involved initially have stable vital signs data and it is difficult to find trauma on the body surface. However, as time goes by, their lives will be in danger, but they are misled by the initial appearances and are ignored, resulting in an irreversible situation.

[0059] In view of this, for patients whose injury level has not reached the preset injury level, the patient's vital signs data will be continuously collected within the preset time period and the injury level will be obtained. If the preset injury level is still not reached when the collection time meets the preset time period, the highest injury level obtained within the preset time period will be used as the highest injury level for treatment.

[0060] For example, under normal circumstances, yellow can be used as the preset injury level. For any patient, if the injury level obtained through analysis based on the uploaded vital signs data, or the vital signs data and external trauma information reaches yellow, appropriate treatment measures can be taken immediately (such as immediately performing appropriate first aid on the spot or sending the patient to the hospital for treatment). If the injury level obtained through analysis does not reach yellow, vital signs data will continue to be collected within a preset time period, and further analyzed in real time to obtain the injury level. Once the preset injury level is reached, treatment measures will be taken to reduce the situation where the patient is misled by the initial appearance and is ignored.

[0061] Those skilled in the art will appreciate that the preset time period may be set based on actual needs, for example, may be set to half an hour, one hour, etc.

[0062] In one embodiment, for patients whose injury level reaches a preset injury level, an alarm is issued to remind medical staff to take timely treatment measures.

[0063] Those skilled in the art will appreciate that, when medical resources are extremely limited, red may be used as the preset injury level.

[0064] Step S50: for any patient with multiple external injuries, obtain the external injury severity ranking based on the external injury information, and prioritize the treatment based on the external injury severity ranking.

[0065] In one embodiment, for any patient with multiple external injuries, the external injury severity ranking is obtained based on the external injury information, including: based on the injury image information and the human body location information where the injury is located, the injury severity recognition model is used to identify the injury severity of each injury. The injury severity recognition model is a neural network model obtained by training with the third loss function, which identifies the injury severity based on the characteristics of the injury image information and the human body location information where the injury is located.

[0066] In one embodiment, please refer to Figure 3 Based on the trauma image information and the human body location information of the trauma, the trauma severity recognition model is used to identify the severity of each trauma, including: Step S1000: for any wound, the wound image information in the external wound information is identified to obtain the wound type, wound depth and wound range.

[0067] Based on the wound image information, image recognition can be performed to obtain the wound type, wound depth and wound range. Among them, the wound type can include laceration, hemorrhagic wound and open wound.

[0068] Step S2000, based on the identified wound type, wound depth and wound range, combined with the human body location of the wound, the wound severity of any wound is identified through a wound severity identification model.

[0069] In one embodiment, a method for training a trauma severity recognition model may include: Step S130 , collecting wound type, wound depth, wound range, human body location of the wound and vital sign data of multiple patients, and marking them to clarify the severity level of the wound.

[0070] Professional medical personnel classify the severity of trauma into mild, moderate, or severe categories based on clinical standards and experience.

[0071] Step S230 , preprocessing the labeled data and converting it into a format suitable for input into a trauma severity recognition model to obtain a data set.

[0072] In one embodiment, the categorical data such as wound type and wound location are converted into numerical codes to facilitate model processing. The numerical data such as wound depth and wound range are normalized, such as using minimum-maximum normalization to scale the data to the [0, 1] interval.

[0073] Step S230 may include: cleaning the original data to remove duplicates, errors and missing values, and then performing normalization or standardization to map data with different features to the same scale range to improve the training efficiency and stability of the model.

[0074] Step S330, dividing the data set into a training set, a validation set and a test set.

[0075] Step S430, constructing a primary architecture of the trauma severity recognition model, based on preset training parameters, using the training set as the input of the primary architecture of the trauma severity recognition model for training, evaluating the performance of the trauma severity recognition model during the training process based on the validation set and adjusting the hyperparameters, and evaluating the generalization ability of the trauma severity recognition model based on the test set, thereby obtaining a trained trauma severity recognition model.

[0076] Among them, the primary architecture of the trauma severity recognition model can be constructed based on the fully connected network and the convolutional neural network. The feature vectors of wound type, depth, range and wound location are spliced ​​in the input layer or hidden layer, allowing the model to learn multiple features at the same time and capture spatial features based on the convolutional neural network.

[0077] In one embodiment, the third loss function may adopt a cross entropy loss function.

[0078] Step S3000, sorting the severity of external injuries based on the identified severity of each injury.

[0079] In one embodiment, injuries of the same severity level but in different locations are ranked according to their impact, for example, head injuries are more severe than injuries near organs, and injuries near organs are more severe than bone injuries.

[0080] In this way, for any patient with multiple external traumas, the severity ranking of the external traumas is obtained based on the external trauma information, and priority treatment is performed based on the severity ranking of the external traumas. This allows for a more accurate determination of the areas that require priority treatment when a patient has multiple diseases or traumas, thereby reducing the situation where more serious traumas cannot be treated in time and the condition is delayed.

[0081] In one embodiment of the present application, a computer-readable storage medium is provided, on which a program is stored. The stored program includes a method that can be loaded by a processor and process any of the above embodiments.

[0082] Those skilled in the art will appreciate that all or part of the functions of the various methods in the above-mentioned embodiments can be implemented by hardware or by computer programs. When all or part of the functions in the above-mentioned embodiments are implemented by computer programs, the program can be stored in a computer-readable storage medium, and the storage medium can include: read-only memory, random access memory, disk, optical disk, hard disk, etc., and the program is executed by a computer to implement the above-mentioned functions. For example, the program is stored in the memory of the device, and when the program in the memory is executed by the processor, all or part of the above-mentioned functions can be implemented. In addition, when all or part of the functions in the above-mentioned embodiments are implemented by computer programs, the program can also be stored in a storage medium such as a server, another computer, disk, optical disk, flash disk or mobile hard disk, and can be downloaded or copied and saved in the memory of the local device, or the system of the local device is updated, and when the program in the memory is executed by the processor, all or part of the functions in the above-mentioned embodiments can be implemented.

[0083] The above specific examples are used to illustrate the present invention, which is only used to help understand the present invention and is not intended to limit the present invention. For those skilled in the art, according to the concept of the present invention, some simple deductions, modifications or substitutions can be made.

Claims

1. A rapid triage method for sudden mass incidents, characterized in that: include: Collect the identity information of each patient; Collect each patient's vital signs data based on smart wearable devices, and associate the collected vital signs data with identity information and upload them; Collecting external trauma information of patients with external trauma based on an image acquisition device, and associating the collected external trauma information with identity information and uploading the information; the external trauma information includes trauma image information and the human body location where the trauma is located; For any patient, based on the collected vital signs data, or vital signs data and external trauma information, injury analysis is performed to obtain the injury level; priority sorting and treatment are performed based on the injury level of each patient; For any patient with multiple external injuries, the severity ranking of the external injuries is obtained based on the external trauma information, and priority treatment is performed based on the severity ranking of the external injuries.

2. The rapid injury classification method for sudden mass incidents according to claim 1 is characterized in that: The injury analysis for any patient based on the collected vital sign data, or the vital sign data and external trauma information, to obtain the injury level includes: For any patient, when the collected data only includes vital signs data, the injury level of the patient is identified based on the vital signs data through the first injury level identification model; the first injury level identification model is a neural network model trained in combination with the first loss function, which identifies the injury level based on the characteristics of the vital signs data.

3. The rapid injury classification method for sudden mass incidents according to claim 1 is characterized in that: The injury analysis for any patient based on the collected vital sign data, or the vital sign data and external trauma information, to obtain the injury level includes: For any patient, when the collected data includes vital signs data and external trauma information, the injury level of the patient is identified based on the vital signs data and the external trauma information through a second injury level identification model; the second injury level identification model is a neural network model trained in combination with the second loss function, which identifies the injury level based on the characteristics of the vital signs data and the external trauma information.

4. The rapid injury classification method for sudden mass incidents as claimed in claim 3 is characterized in that: The identification of the injury level of any patient based on the vital sign data and external trauma information through the second injury level identification model includes: Identify the wound image information in the external wound information to obtain the wound type, wound depth and wound range; Based on the identified wound type, wound depth and trauma range, combined with the human body location of the trauma and vital signs data, the injury level of any patient is identified through the second injury level identification model.

5. The rapid injury classification method for sudden mass incidents according to any one of claims 1 to 4, characterized in that: The injury levels are from high to low, including red, yellow, green and black, where red indicates life-threatening and requires immediate emergency treatment; yellow indicates serious injuries but not life-threatening; green indicates minor injuries and stable vital signs; black indicates death or inability to treat.

6. The rapid injury classification method for sudden mass incidents according to any one of claims 1 to 4, characterized in that: For patients whose injury level has not reached the preset injury level, the patient's vital signs data will be continuously collected within the preset time period and the injury level will be obtained. If the preset injury level is still not reached when the collection time meets the preset time period, the highest injury level obtained within the preset time period will be used as the highest injury level for treatment.

7. The rapid injury classification method for sudden mass incidents according to claim 6 is characterized in that: For patients whose injury level reaches the preset injury level, an alarm prompt will be issued.

8. The rapid injury classification method for sudden mass incidents as claimed in claim 1 is characterized in that: For any patient with multiple external injuries, obtaining the external injury severity ranking based on the external injury information includes: Based on the trauma image information and the human body location information of the trauma, the trauma severity of each trauma is identified through a trauma severity recognition model; the trauma severity recognition model is a neural network model trained in combination with the third loss function, which identifies the severity of the trauma based on the trauma image information and the human body location information characteristics.

9. The rapid injury classification method for sudden mass incidents as claimed in claim 8, characterized in that: The trauma severity identification model based on the trauma image information and the human body location information of the trauma is used to identify the severity of the trauma at each location, including: For any wound, the wound image information in the external wound information is identified to obtain the wound type, wound depth and wound range; Based on the identified wound type, wound depth and wound range, combined with the human body location of the wound, the wound severity of any wound is identified through a wound severity identification model; Based on the severity of each identified injury, the severity of external injuries is ranked.

10. A computer-readable storage medium, characterized in that: The medium stores a program, which can be loaded by a processor and execute the method for rapid triage of sudden mass incidents as claimed in any one of claims 1 to 9.

Images