Method and device for determining injury level, computer equipment and storage medium

Abstract

The embodiment of the invention provides a method and device for determining the injury condition grade, computer equipment and a storage medium, and the method for determining the injury condition grade comprises the steps: obtaining vital sign data and injury condition information of a wounded person; the injury condition information is used for describing the injury condition of the wounded person; based on the vital sign data and the injury condition information, generating an injury condition text; and analyzing the injury condition text by using a pre-trained triage classification model to obtain an injury condition level and an injury type of the wounded, and outputting the injury condition level and / or the injury type.

Description

Technical Field

[0001] The present application relates to but is not limited to the field of artificial intelligence, and in particular to a method, apparatus, computer equipment and storage medium for determining an injury level. Background Art

[0002] At the scene of a large-scale disaster, there are many injured people with complex injuries, but emergency resources such as doctors, equipment, medicines, and materials often cannot meet the needs of treatment at the same time, and there is a contradiction between treatment needs and resources. Therefore, it is necessary to skillfully conduct a preliminary assessment of the injured in a short period of time, determine the order of treatment and what type of rescue is needed, so as to shorten the emergency time and give priority to the injured who need emergency treatment the most. Therefore, how to determine the injury level and type of injury of the injured has become a key issue that needs to be solved. Summary of the invention

[0003] In view of this, embodiments of the present application provide a method, apparatus, computer device, and storage medium for determining an injury level.

[0004] The technical solution of this application is implemented as follows:

[0005] In a first aspect, an embodiment of the present application provides a method for determining an injury level, the method comprising: obtaining vital sign data and injury information of an injured person; the injury information is used to describe the injury of the injured person; based on the vital sign data and the injury information, an injury text is generated; using a pre-trained injury classification model to analyze the injury text to obtain the injury level and injury type of the injured person, and outputting the injury level and / or the injury type.

[0006] In a second aspect, an embodiment of the present application provides a device for determining an injury level, the device comprising:

[0007] The first acquisition module is used to acquire the vital sign data and injury information of the injured person; the injury information is used to describe the injury of the injured person;

[0008] A generating module, used for generating an injury text based on the vital sign data and the injury information;

[0009] The analysis module is used to analyze the injury text using a pre-trained injury classification model to obtain the injury level and injury type of the injured person, and output the injury level and / or the injury type.

[0010] In a third aspect, an embodiment of the present application provides a computer device, including a memory and a processor, wherein the memory stores a computer program that can be executed on the processor, and when the processor executes the program, some or all of the steps in the above method are implemented.

[0011] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program implements some or all of the steps in the above method when executed by a processor.

[0012] In an embodiment of the present application, on the one hand, the vital signs data and injury information of the injured person are integrated to generate an injury text, thereby improving the standardization and conciseness of the injury text containing the vital signs data and injury information of the injured person; on the other hand, the injury text is analyzed using a pre-trained injury detection and classification model to obtain the injury level and injury type of the injured person, and the injury level and / or the injury type are output, thereby improving the accuracy of analyzing the injury level and injury type.

[0013] It should be understood that the above general description and the following detailed description are merely exemplary and explanatory, and are not intended to limit the technical solutions of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] The drawings herein are incorporated into the specification and constitute a part of the specification. These drawings illustrate embodiments consistent with the present application and are used together with the specification to illustrate the technical solution of the present application.

[0015] Figure 1 A schematic diagram of an implementation flow of a method for determining an injury level provided in an embodiment of the present application;

[0016] Figure 2 A schematic diagram of the implementation flow of another method for determining the injury level provided in an embodiment of the present application;

[0017] Figure 3 A schematic diagram of a triage rule provided in an embodiment of the present application;

[0018] Figure 4 A schematic diagram of the structure of a device for determining an injury level provided in an embodiment of the present application;

[0019] Figure 5 A schematic diagram of determining the injury level and injury type provided in an embodiment of the present application Figure 1 ;

[0020] Figure 6 A schematic diagram of determining the injury level and injury type provided in an embodiment of the present application Figure 2 ;

[0021] Figure 7 A schematic diagram of the composition structure of another device for determining the injury level provided in an embodiment of the present application;

[0022] Figure 8 A hardware entity diagram of a computer device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0023] In order to make the purpose, technical solutions and advantages of the present application clearer, the technical solutions of the present application are further elaborated in detail below in conjunction with the drawings and embodiments. The described embodiments should not be regarded as limiting the present application. All other embodiments obtained by ordinary technicians in the field without making creative work are within the scope of protection of the present application.

[0024] In the following description, reference is made to “some embodiments”, which describe a subset of all possible embodiments, but it will be understood that “some embodiments” may be the same subset or different subsets of all possible embodiments and may be combined with each other without conflict.

[0025] The terms "first / second / third" involved are merely used to distinguish similar objects and do not represent a specific ordering of the objects. It is understandable that "first / second / third" can be interchanged with a specific order or sequence where permitted so that the embodiments of the present application described herein can be implemented in an order other than that illustrated or described herein.

[0026] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art to which this application belongs. The terms used herein are only for the purpose of describing this application and are not intended to limit this application.

[0027] In order to facilitate understanding of the embodiments of the present application, the technical solutions related to the embodiments of the present application and the shortcomings of the technical solutions are briefly introduced below.

[0028] The World Health Organization defines mass casualty incidents as disasters and major events characterized by the large number of casualties, high severity and complexity of injuries, which may rapidly overwhelm the capacity of local health resources to provide comprehensive medical care. In recent decades, the frequency of mass casualty incidents has gradually increased and has affected the socioeconomic development of all countries. A clear prehospital triage system is essential to save lives and optimize the activation of appropriate resource allocation when disasters occur.

[0029] One of the characteristics of disasters is that the direct medical needs of the affected groups exceed the existing medical and health resources. Therefore, how to use limited resources to provide the best medical treatment for the people is a key issue that needs to be solved. The core essence of triage is to allocate limited medical resources among the disaster accident sites. Triage is one of the key principles for effectively managing major emergencies.

[0030] The Simple Triage And Rapid Treatment (START) method in the prior art is applicable to mass casualty incidents. Emergency personnel use the START method to quickly identify and classify the wounded, thereby achieving preliminary triage of a large number of wounded in a short period of time.

[0031] Initial triage is the process of classifying, categorizing and prioritizing the casualties according to their severity. During the triage process, the casualties are not given enough medical resources to ensure that the maximum number of casualties receive medical services. Initial triage is performed on multiple casualties at the first opportunity, and different triage categories are given according to the severity of the casualties' injuries. Different color labels are used to indicate the triage results of the casualties. Red labels indicate that the casualties are in a critical condition, yellow labels indicate that the casualties are slightly less severe, green labels indicate that the casualties are generally slightly injured and can move independently, and black labels indicate that the casualties have died or are seriously injured and are about to die.

[0032] The simplified triage method has the following disadvantages: it only determines the basic conditions of the injured person such as breathing, pulse, and consciousness, and cannot comprehensively collect the injured person's injuries and triage. Triage is a dynamic process, and the injured person's injuries will change, which may change the triage results. However, in an emergency, it is difficult to repeatedly check and record the injured on a large scale, and it is impossible to compare the dynamic changes of the results of the previous and subsequent examinations, and re-evaluate and re-identify the injured person's injuries.

[0033] Based on this, the present application embodiment provides a method for determining the injury level, such as Figure 1 As shown, the method may include the following steps S101 to S103, wherein:

[0034] Step S101, obtaining vital sign data and injury information of the injured person; the injury information is used to describe the injury of the injured person;

[0035] Here, the vital signs data of the injured person refers to: data obtained after validity checking of the collected heart rate data, blood oxygen data and respiratory rate data of the injured person; the injury information of the injured person refers to: information obtained after a series of processing of the voice about the injured person's injury.

[0036] In some embodiments, the implementation of “obtaining vital sign data of the injured person” in step S101 may include the following steps S111 to S113, wherein:

[0037] Step S111, collecting initial vital sign data of the injured person;

[0038] Here, the initial vital sign data of the injured person includes at least one of heart rate data, blood oxygen data and respiratory rate data.

[0039] During implementation, the vital signs data of the injured are collected in real time through the basic vital signs detection module. For example, the green light photoelectric measurement method is used to detect the heart rate data of the injured; the reflective photoelectric detection method is used to detect the blood oxygen data of the injured; because the ratio of respiratory rate to heart rate is 5:1, that is, one breath is equal to five heartbeats, so when the heart rate data is known, the respiratory rate data of the injured is determined by using the relationship between respiratory rate and heart rate.

[0040] Step S112, when the change rate of two adjacent data in the same type of initial vital sign data is greater than a preset value, a section of the same type of initial vital sign data is selected for averaging to obtain an average value of the same type of initial vital sign data;

[0041] During implementation, if the rate of change of two consecutive heart rate data of the injured person collected is greater than a preset value, a segment of real-time heart rate data is selected for averaging to obtain the average value of the heart rate data. In some embodiments, by real-time monitoring of the heart rate data, N heart rate data at the positions before and after the heart rate data corresponding to the preset value whose rate of change of the heart rate data is greater than that of the preset value are selected for averaging to obtain the average value of the heart rate data. For example, N can be 5 or 6, and the embodiment of the present application does not limit the number of selected heart rate data.

[0042] Step S113: determining the average value of each type of the initial vital sign data as the vital sign data.

[0043] During implementation, if the change rates of two consecutive heart rate data, blood oxygen data and respiratory rate data collected from the injured person are greater than the preset values, the average value of the heart rate data, the average value of the blood oxygen data and the average value of the respiratory rate data will be used as vital sign data.

[0044] In some embodiments, the implementation of "obtaining injury information of the injured person" in step S101 may include the following steps S121 and S122, wherein:

[0045] Step S121, collecting and converting the voice describing the injury of the injured person into text to obtain a voice text;

[0046] Here, the voice describing the injured person's condition may be spoken by the injured person or by someone near the injured person.

[0047] During implementation, since the description of the injured person's condition is collected in the form of voice, the voice needs to be converted into text first, so as to obtain the injured person's injury information by correcting errors and standardizing terminology on the voice text in turn.

[0048] Step S122, performing error correction and terminology standardization processing on the voice text in sequence to obtain the injury information of the injured person.

[0049] During implementation, in order to ensure the accuracy of the voice text, the voice text needs to be corrected. Since the injured may not be a medical professional, the voice text may be relatively colloquial. Medical terminology is used to standardize the corrected voice text to obtain the injured person's injury information.

[0050] In some embodiments, the implementation of step S122 "performing error correction and terminology standardization processing on the voice text in sequence to obtain the injury information of the injured person" may include the following steps S131 and S132, wherein:

[0051] Step S131, using the candidate words corresponding to the words in the speech text to correct the speech text, to obtain a speech text after error correction;

[0052] Here, the candidate words refer to homophones and synonyms of words in the speech text obtained through the recall strategy.

[0053] In some embodiments, the implementation of step S131 "correcting the speech text using candidate words corresponding to the words in the speech text to obtain the corrected speech text" may include the following steps S141 and S144, wherein:

[0054] Step S141, determining candidate words corresponding to the words in the speech text based on a preset recall strategy;

[0055] Here, the preset recall strategy includes a homophone recall strategy and / or a front and back nasal recall strategy.

[0056] During implementation, the speech text is processed using a homophone recall strategy and / or a front and back nasal recall strategy to obtain homophones and synonyms of words in the speech text.

[0057] Step S142, using each of the candidate words to replace the words in the wrong position in the speech text, and obtaining the replaced speech text;

[0058] During implementation, if there are multiple homophones of a word in the speech text, each homophone is used to replace the word in the wrong position in the speech text, and the speech text after each homophone is replaced is obtained.

[0059] In some embodiments, determining the error location includes the following steps S151 to S153, wherein:

[0060] Step S151, segmenting the speech text to obtain a word list;

[0061] Here, the word list includes all words after the speech text is segmented.

[0062] During implementation, the speech text is segmented to obtain a word list containing independent words.

[0063] Step S152, determining the score of at least two adjacent words in the word list after being concatenated based on the preset word scores;

[0064] Here, the preset word scores include the scores of all words in the word list.

[0065] During implementation, at least two adjacent words in the word list are concatenated into a larger word, and the score of the concatenated word is determined using the word scores in the word list.

[0066] Step S153: determining the position corresponding to the word with a low score after concatenation as an incorrect position.

[0067] During implementation, the scores of all concatenated words are compared, and the positions corresponding to the words with low concatenated scores are determined as incorrect positions.

[0068] Step S143, determining the perplexity of each replaced speech text based on each replaced speech text;

[0069] In implementation, since there are more than one candidate words, a replaced speech text will be generated after each candidate word is replaced. In this way, the correct candidate word can be determined by calculating the perplexity of each replaced speech text.

[0070] Step S144: determine the speech text with the lowest confusion as the speech text after error correction.

[0071] During implementation, the perplexity of each replaced speech text is compared, and the speech text with the lowest perplexity is determined as the speech text after error correction.

[0072] Step S132, using the preset correspondence between medical terms and vernacular terms, term standardization processing is performed on the corrected speech text to obtain the injury information of the injured person.

[0073] During implementation, the existing medical knowledge graph was utilized to add the correspondence between medical terminology and vernacular terminology in disaster rescue scenarios. For example, the vernacular term for being crushed by a stone corresponds to the medical term for being compressed by a foreign object. Through the correspondence between medical terminology and vernacular terminology, the corrected speech text was processed for terminology standardization to determine the injury information of the injured.

[0074] Step S102, generating an injury text based on the vital sign data and the injury information;

[0075] Here, the injury text may be in the format of an electronic medical record. For example, the electronic medical record format may be: age X years old, gender X, consciousness X, heart rate X, blood oxygen X, respiratory rate X, and injury information.

[0076] During implementation, the vital signs data and injury information of the injured person are integrated according to the electronic medical record format to generate an injury text in the electronic case format corresponding to the injured person.

[0077] Step S103, using a pre-trained injury classification model to analyze the injury text, obtain the injury level and injury type of the injured person, and output the injury level and / or the injury type.

[0078] Here, the pre-trained triage classification model refers to: using the pre-trained medical model to train the training data of previous injured persons to obtain a model that can classify the injury level and injury type of the injured persons; the injury level includes four categories from light to heavy, namely green, yellow, red and black; the injury types include burns, poisoning, trunk trauma, limb trauma and head and neck trauma, etc.

[0079] During implementation, the injury text is analyzed using a pre-trained injury classification model to obtain the injury level and injury type of the injured person, and the injury level and / or the injury type are displayed in real time through a display module. Since the vital sign data of the injured person is collected in real time, the vital sign data of the injured person in each time period may be different, so the injury level of the injured person may also change, thereby dynamically changing the injury level displayed by the display module.

[0080] In an embodiment of the present application, on the one hand, the vital signs data and injury information of the injured person are integrated to generate an injury text, thereby improving the standardization and conciseness of the injury text containing the vital signs data and injury information of the injured person; on the other hand, the injury text is analyzed using a pre-trained injury detection and classification model to obtain the injury level and injury type of the injured person, and the injury level and / or the injury type are output, thereby improving the accuracy of analyzing the injury level and injury type.

[0081] In some embodiments, the method further includes the following steps S161 and S162, wherein:

[0082] Step S161, when the injury level of the injured person is low, based on the injury type of the injured person, extracting self-rescue measures corresponding to the injury type from a rescue database;

[0083] Here, the injury level includes green, yellow, red, and black from light to heavy, so the injury level of the injured person is low and can be green. In some embodiments, if the injury level of the injured person is green and the injury type is limb trauma, the intelligent analysis module extracts the self-help measures corresponding to the limb trauma from the rescue database.

[0084] Step S162: Based on the self-rescue measures, instruct the injured person to perform self-rescue through voice.

[0085] During implementation, the self-rescue measures corresponding to limb injuries are extracted from the rescue database, and the injured are guided to perform self-rescue through voice. For example, if the injured is still moving after being injured, the voice playback module will tell the injured that he should stop the activity immediately and stay calm as much as possible to prevent excessive blood loss.

[0086] In some embodiments, the injury level, the injury type, and the acquired location of the injured are uploaded to a rescue center; the rescue center is used to provide rescue to the injured according to the number of the injured, the location of the injured, the injury level and the injury type.

[0087] During implementation, the injury level and injury type of the wounded are uploaded to the rescue center through the data communication module. The rescue center allocates rescue resources to rescue the wounded according to the number of the wounded, the location of the wounded, the injury level and the injury type. This enables the rescue center to fully grasp the injuries of the wounded and maximize the rescue role of rescue resources within a limited time, so that the overall rescue achieves the greatest effect.

[0088] In some embodiments, the method further includes the following steps S171 and S172, wherein:

[0089] Step S171, obtaining a training set; the training set includes training samples of each category; the training samples include the first vital sign data, first injury information, first injury level and first injury type of the injured person;

[0090] Here, the training samples refer to: the vital signs data, injury information, injury level and injury type of the injured collected in previous disaster relief situations.

[0091] Step S172, using the pre-trained medical model to train part or all of the training samples in the training set to obtain the pre-trained triage classification model.

[0092] Here, the pre-trained medical model is trained on a large amount of Chinese language and has been strengthened to a certain extent in the medical field.

[0093] During implementation, the pre-trained medical model is used to train part or all of the training samples in the training set to obtain a pre-trained triage model that can analyze the injury level and injury type of the injured.

[0094] The embodiment of the present application provides another method for determining the level of injury. The method can actively and continuously detect the heart rate, blood oxygen and respiratory rate of the injured, while collecting descriptions of the injured person's injuries from the injured person or people near the injured person, and using a pre-trained injury classification model to intelligently classify the injured person's injuries in more detail, and guide the injured person to perform actionable self-rescue according to the injured person's injury level. Dynamically adjust the injury level indicator based on the injured person's vital signs data and the description of the injured person's injury. In the case of limited resources, as many injured as possible can obtain the best treatment effect to improve the efficiency of emergency first aid and rescue.

[0095] Currently, four colors of identification are used: red, yellow, green, and black, which respectively indicate different levels of injuries and the order of urgency of rescue. A red identification indicates that critical injuries are the first priority, a yellow identification indicates that serious injuries are the second priority, a green identification indicates that minor injuries will be treated in a deferred manner, and a black identification indicates fatal injuries and death. A black identification may change into a red identification, a red identification may change into a black identification, a yellow identification may change into a red identification, and a green identification may change into a yellow identification. In view of the deficiencies of the prior art, the embodiments of the present application provide another method for determining the level of injuries, such as Figure 2 As shown, the method comprises the following steps:

[0096] Step S201: At the disaster site, medical rescue personnel classify the injured according to the simplified triage method.

[0097] Step S202, continuously detecting the vital sign data of the injured person and collecting descriptions of the injured person's injuries.

[0098] Step S203, through intelligent analysis of the wounded's injury information and vital signs data, recommend and guide the wounded to find the best self-rescue plan.

[0099] Step S204, dynamically changing the mark of the injury level according to the continuous detection of the vital sign data of the injured person.

[0100] Step S205, upload the injury level and injury type of the injured to the rescue center, so that rescue professionals can timely grasp the situation at the disaster site, deploy rescue resources, maximize the role of rescue resources, and achieve the maximum effect of overall rescue.

[0101] The above steps S201 to S205 are explained as follows.

[0102] In step S201, the concise triage method is applicable to mass casualty incidents, and the preliminary triage of a large number of injured persons in a short period of time at the scene is performed by the first emergency personnel to quickly identify and classify the injured and sick. Triage can maximize the survival rate, reduce the degree of disability as much as possible, and safely and promptly transfer patients to qualified hospitals for treatment. In sudden large-scale accident health events or safety events, such as stampedes, explosions, sudden safety accidents in campuses, shopping malls and public places, traffic accidents (serial car accidents), industrial enterprise fires, forest fires, earthquakes, mudslides, landslides and other natural or non-natural disasters, it is very important to quickly identify critically ill patients. Due to the suddenness of the incident, the medical resources and rescue forces at the scene are not sufficient. At this time, experienced emergency rescue personnel or emergency medical rescue personnel are needed to quickly organize on-site to allow as many injured persons as possible to obtain the best treatment effect under limited resources. Triage of trauma is the primary link of on-site medical first aid. One of the characteristics of modern society, group injuries, and on-site first aid is to perform triage. It should be noted that casualty classification is a dynamic process, as the casualty's condition is constantly changing, and the casualty needs to be repeatedly reassessed and appropriate adjustments made.

[0103] Further, the details of triage, such as Figure 3 As shown, the method includes:

[0104] Step 301, determining whether the injured person can walk.

[0105] Here, if yes, that is, the injured person can walk, the injured person is determined to be a green injured person 31; if no, that is, the injured person cannot walk, the process proceeds to step 302;

[0106] Step 302, determining whether the injured person can breathe.

[0107] Here, if yes, that is, the injured person can breathe, then the process proceeds to step 303 ; if no, that is, the injured person cannot breathe, then the process proceeds to step 304 .

[0108] Step 303, determine whether the injured person's breathing rate is greater than 30 times / minute.

[0109] Here, if yes, that is, the patient's breathing is greater than 30 times / minute, the patient is determined to be a red patient 32; if no, that is, the patient's breathing is less than or equal to 30 times / minute, the process proceeds to step 305.

[0110] Step 304, open the airway of the injured person and determine whether he can breathe autonomously.

[0111] Here, when it is yes, that is, when the injured person can breathe on his own, the injured person is determined to be a red injured person 32; when it is no, that is, when the injured person cannot breathe on his own, the injured person is determined to be a black injured person 33.

[0112] Step 305, determining whether the radial artery pulse of the injured person is palpable or whether the capillary filling time is greater than 2 seconds.

[0113] Here, if yes, that is, the radial artery pulsation of the injured person cannot be touched or the capillary refill time is greater than 2 seconds, the injured person is determined to be a red injured person 32; if no, that is, the radial artery pulsation of the injured person can be touched or the capillary refill time is less than or equal to 2 seconds, the process proceeds to step 306.

[0114] Step 306, determining whether the injured person is conscious.

[0115] Here, if yes, that is, the injured person is conscious, the injured person is determined to be a yellow injured person 34 ; if no, the injured person is determined to be a red injured person 32 .

[0116] Red casualty 32: Critically ill patients, first priority, red logo, indicating emergency treatment. Critically ill patients need immediate medical treatment, and priority should be given to nursing and transfer. They have severe life-threatening injuries, and can be rescued with simple methods, shorter time and fewer resources, and the rescue can lead to better results. Respiration is greater than 30 times / minute; respiratory circulation is stable, radial artery pulsation cannot be touched or capillary filling time is greater than 2 seconds; they can breathe spontaneously after opening the airway or relieving airway obstruction. For example: controllable massive bleeding in the limb arteries.

[0117] Yellow casualty 34: Severely ill patients, second priority, yellow mark, indicating delayed treatment. Patients with stable injuries but unable to walk, stable breathing and circulation, palpable radial artery pulse or capillary refill of less than 2 seconds, clear consciousness, major trauma, but can still wait for a short time without endangering life or causing physical disability, serious injuries, but after emergency treatment, vital signs or injuries are temporarily stable, and treatment and transfer are allowed to be delayed for a certain period of time. For example: simple femoral or humeral fractures.

[0118] Green casualty 31: Lightly injured, third priority, green logo, slightly injured, can walk on their own, no serious injuries, their injuries can be appropriately delayed for transfer and treatment. The injured will be guided to the lightly injured receiving station first, and the lightly injured can wait for treatment after the serious injured are treated, or they can rescue themselves under the guidance of rescuers, and even assist rescuers in first aid for other injured when manpower is insufficient. For example: body surface abrasions, contusions, wounds with less bleeding, and joint sprains, etc.

[0119] Black casualty 33: dead or dying casualties, fourth priority, black mark, indicating that the injuries are too serious and there is little hope of survival even with strong treatment, parked in a specific area. The dying state caused by the serious injuries, death or irreparable fatal trauma. For example: respiratory arrest, severe head, chest, and abdominal trauma, and cardiopulmonary resuscitation cannot be performed. For example: severe craniocerebral injury and 95% body surface area third-degree burns, even if the airway is opened, the wounded still cannot breathe independently. When resources are scarce or the on-site environment is unsafe, it is necessary to give up the treatment of dead or dying casualties, otherwise the occupation of medical resources will cause a large number of deaths of the wounded marked in red. Active treatment can also be given when the rescue force is sufficient.

[0120] In step S202, a device that can actively and continuously detect the heart rate, blood oxygen and respiratory rate of the injured person and collect descriptions of the injured person's condition is fixed on the skin of the injured person, for example, the arm. The device is tied to the arm with a soft material, with the side with the photoelectric sensor close to the skin. Figure 4 As shown, the device for determining the injury level includes but is not limited to the following modules: a voice acquisition module 41, a basic vital signs detection module 42, a global positioning system (GPS) module 43, an intelligent analysis module 44, a data communication module 45, a voice playback module 46, a display module 47, a power module 48 and a clock module 49, etc.

[0121] Furthermore, the basic vital signs detection module is used to continuously detect the heart rate, blood oxygen and respiratory rate of the injured. The blood oxygen detection adopts a reflective photoelectric detection scheme, the heart rate detection uses a green light photoelectric measurement method, and the respiratory rate is obtained by simple calculation. Generally speaking, the ratio of respiratory rate to heart rate is 5:1, that is, five heartbeats are required for each breath. The voice collection module is used to collect descriptions of the injured person's injuries and convert the voice of the injured person's injuries into text and store it.

[0122] In step S203, the intelligent analysis module is used to perform intelligent analysis on the injury information and vital signs of the injured, and recommend and guide the best self-rescue plan for the injured. Figure 5 As shown, the intelligent analysis module combines the injury information 52 and vital signs data 51 of the injured person to generate an injury text 53 in the electronic medical record format, and then uses the pre-trained injury classification model 54 to analyze the injury text in the electronic medical record format, intelligently determine the injury level 55 and injury type 56 of the injured person, and guide the injured person to perform self-rescue according to the specific injury situation.

[0123] The specific implementation process of step S203 includes the following contents:

[0124] Process the vital sign data of the wounded: Real-time detect the vital sign data (heart rate, blood oxygen, and respiratory rate) of the wounded, and perform validity checks on the vital sign data (such as whether the numerical range is valid, whether the necessary data is non-empty, etc. For example, the heart rate range is [0, 200], the blood oxygen range is [0%, 100%], and the respiratory rate is [0, 100]). If the change rate of adjacent two blood oxygen data is greater than 80%, then the check fails. A segment of real-time blood oxygen data will be averaged, and the average value of the blood oxygen data will be used as the final blood oxygen data. For example, if the detected blood oxygen data suddenly drops from 98% to 10%, then 5 adjacent blood oxygen data will be selected for average calculation, and the average value of the blood oxygen data will be used as the final blood oxygen data.

[0125] Process the voice about the wounded's injury: Convert the voice about the wounded's injury into text to obtain the voice text containing dialogue data. In addition, the wounded may not be medical professionals, and the content of the voice text may be relatively colloquial. To ensure the accuracy and conciseness of the voice text, error correction and standardization of medical terms need to be performed on the voice text. The voice text is first corrected for errors, and then entity extraction is performed to extract key entities in the voice text, such as the head, abdomen, and pain. Then, based on the medical knowledge graph, term standardization processing is carried out. For each extracted entity, find its synonymous relationship standard medical entity description in the medical knowledge graph and replace the original word in the voice text with this description.

[0126] Among them, the process of error correction of the voice text specifically includes:

[0127] 1) Recall strategy: Establish multiple recall strategies and determine the candidate words of the words in the voice text through the recall strategy; for example, homophone recall (such as "识", "十", "史", and "实") and front-back nasal sound recall (such as "an" and "ang");

[0128] 2) Word segmentation: Segment the voice text to obtain a list of words;

[0129] 3) Locate the position of the wrong word: Based on the scores of the preset words, determine the scores of at least 2 adjacent words spliced in the word list, and determine the position with an extremely low score as the wrong position;

[0130] 4) Replace all candidate words at the wrong position and calculate the perplexity of the voice text after replacement. Determine the voice text with the lowest perplexity as the voice text after error correction.

[0131] The voice text after error correction is as follows:

[0132] Q: May I ask your age and gender?

[0133] A: 56 years old, male.

[0134] Q: Please describe your current condition?

[0135] Answer: My calf has been pressed by a stone for a long time, it is bleeding and feels numb.

[0136] Among them, terminology standardization specifically includes:

[0137] 1) Entity extraction: Use named entity recognition method to identify entities;

[0138] 2) Medical knowledge graph: Manually extract medical entities from medical literature and synonym dictionaries to construct triples (for example, the vernacular text of being pressed by a stone corresponds to the medical term of being compressed by a foreign body), and use the extracted triples to construct a medical knowledge graph using a graph database.

[0139] Furthermore, after the speech text has been corrected and terminology standardized, it is fused with the vital signs data in a standard electronic medical record format. The fusion process includes:

[0140] 1) Construction template: age X years old, gender X, consciousness X, heart rate X, blood oxygen X, respiratory rate X, description of the injured person's injury information.

[0141] 2) Extract key information, including:

[0142] ① The age range is [0-150], and the gender range is [male, female]. For example, in the first answer in the example, age and gender are extracted, and the results are matched and filled into the corresponding X.

[0143] ②Heart rate, blood oxygen, and respiratory rate are all detected through the basic vital signs detection module.

[0144] ③ Consciousness is determined by the description in the conversation. If the person answers that he is conscious, if not, the operator will call people near the injured person to ask whether he is conscious. The range of consciousness is [yes, no].

[0145] ④ Take the second answer in the example as the voice text describing the patient’s injury information, first perform word segmentation to obtain a word list; then send all the words in the word list to the medical knowledge graph for synonym replacement. If there is no synonym, retain the original word.

[0146] The final injury text in the electronic medical record format is: age 56, male, conscious, heart rate 110, blood oxygen 95%, respiratory rate 25, the calf was compressed by a foreign object for a long time, bleeding occurred, and the limbs were numb.

[0147] Input the injury text in the electronic medical record format into the pre-trained injury classification model (BidirectionalEncoder Representation from Transformers, BERT) with a dimension of 512, such as Figure 6 As shown, the pre-trained triage model 54 outputs the injury level 55 and injury type 56 of the injured, and guides the injured to perform self-rescue according to the specific injury level and injury type. A [CLS] symbol is inserted before and after the injury text in the electronic medical record format, and the output vector corresponding to the symbol is used as the semantic representation of the entire text for text classification. This symbol without obvious semantic information will more fairly integrate the semantic information of each word and phrase in the text. The injury text is judged by the pre-trained triage model, and four types of injury levels and injury types are output: red, yellow, green and black.

[0148] At the same time, according to the level of injury and type of injury of the injured, the voice playback module guides the injured to self-rescue. For example, if the level of injury of the injured is green and the type of injury is limb trauma, the self-rescue measures corresponding to the trauma are extracted from the rescue database to guide the injured to self-rescue: 1) Stop activities: If the injured is still doing activities after being injured, they should stop immediately and stay as calm as possible to prevent excessive blood loss.

[0149] 2) Massage the limbs: You can gently massage the injured part to promote blood circulation and relieve muscle tension and numbness. 3) If conditions permit, bandage the wound with clean gauze or bandage to stop bleeding and prevent infection. Among them, the guidance and rescue database refers to a two-dimensional table of injury types and guidance and rescue. Injury types include trauma, burns, crush injuries, etc. Guidance and rescue refers to the corresponding rescue methods under the injury type.

[0150] The process of establishing a pre-trained triage model for outputting the injury level and injury type of the injured includes:

[0151] 1) Obtain a training set, which includes all training samples. The training samples are composed of age data, gender data, consciousness data, heart rate data, blood oxygen data, respiratory rate data, and a description of the injuries of the injured. Each training sample corresponds to an injury level and injury type. The injury level categories include red, yellow, green and black; the injury types include burns, poisoning, trunk trauma, limb trauma, and head and neck trauma.

[0152] 2) Use the pre-trained medical model (ernie-health-zh) to perform secondary training on part or all of the training samples in the training set to obtain a pre-trained triage classification model that can classify injury levels and injury types.

[0153] In step S204, according to the analysis result of the injury level by the pre-trained triage model, the injury level of the injured person is dynamically changed in real time through the display module. The change of the basic vital sign detection module or the voice collection information can trigger the pre-trained triage model in the intelligent analysis module to re-analyze the injury level of the injured person.

[0154] In step S205, the injury level and injury type of the injured are uploaded to the rescue center, which can fully grasp the number of injured people of various types, the location of the injured people, and the injuries of the injured people. According to the on-site situation, rescue resources are deployed. For example, the site needs to increase resources such as fire brigades, ambulances, and medical equipment. At the same time, the rescue center needs to formulate rescue methods to ensure that the lives of the injured are saved to the maximum extent within a limited time. In addition, in terms of resource allocation, the GPS positioning module can be used to collect the location of the injured, so as to plan the rescue action path according to the location of the injured, and then better coordinate various rescue measures. In addition, emerging technologies such as drones can also be used to detect the situation at the disaster site in real time, providing more detailed and comprehensive information collection and analysis for rescue work.

[0155] Compared with the prior art, the embodiments of the present application have the following advantages:

[0156] 1. The embodiment of the present application realizes real-time dynamic judgment of the injury level of the injured person by collecting the vital signs data of the injured person in real time, and dynamically adjusts the injury level indicator according to the injury level of the injured person.

[0157] 2. The embodiment of the present application utilizes a pre-trained injury classification model to classify the injury level and injury type of the injured person's injury text.

[0158] 3. The embodiment of the present application regularly uploads data on the number of injured persons, their locations, and their injuries to the rescue center, which deploys rescue resources based on the on-site situation.

[0159] Based on the foregoing embodiments, the embodiments of the present application provide a device for determining the level of injury, which includes the modules included, the sub-modules included in each module, the units included in each sub-module, and the sub-units included in each unit, which can be implemented by a processor in a computer device; of course, it can also be implemented by a specific logic circuit; in the implementation process, the computer device can be a service platform and a fixed-line terminal, and the processor can be a central processing unit (CPU), a microprocessor (MPU), a digital signal processor (DSP) or a field programmable gate array (FPGA), etc.

[0160] Figure 7 A schematic diagram of the structure of a device for determining an injury level provided in an embodiment of the present application is shown in FIG. Figure 7As shown, the device 700 for determining the injury level includes: a first acquisition module 710, used to obtain the vital signs data and injury information of the injured person; the injury information is used to describe the injury of the injured person; a generation module 720, used to generate an injury text based on the vital signs data and the injury information; an analysis module 730, used to analyze the injury text using a pre-trained injury classification model to obtain the injury level and injury type of the injured person, and output the injury level and / or the injury type.

[0161] In some embodiments, the device for determining the injury level also includes: an extraction module, which is used to extract self-rescue measures corresponding to the injury type from the rescue database based on the injury type of the injured person when the injury level of the injured person is low; and a guidance module, which is used to guide the injured person to perform self-rescue through voice based on the self-rescue measures.

[0162] In some embodiments, the device for determining the injury level also includes: an uploading module, used to upload the injury level, the injury type, and the acquired location of the injured person to a rescue center; the rescue center is used to provide rescue to the injured person according to the number of the injured persons, the location of the injured persons, the injury level and the injury type.

[0163] In some embodiments, the first acquisition module includes: a first acquisition submodule, used to collect the initial vital signs data of the injured person; the initial vital signs data includes at least one of heart rate data, blood oxygen data and respiratory rate data; a selection submodule, used to select a section of the same type of initial vital signs data for averaging when the rate of change of two adjacent data in the same type of initial vital signs data is greater than a preset value, so as to obtain an average value of the same type of initial vital signs data; a determination submodule, used to determine the average value of each type of initial vital signs data as the vital signs data.

[0164] In some embodiments, the first acquisition module includes: a second acquisition submodule, used to collect and convert the voice describing the injury of the injured person into text to obtain voice text; a processing submodule, used to perform error correction and terminology standardization on the voice text in turn to obtain the injury information of the injured person.

[0165] In some embodiments, the processing submodule includes: an error correction unit, which is used to correct the speech text using candidate words corresponding to the words in the speech text to obtain a corrected speech text; a processing unit, which is used to perform terminology standardization processing on the corrected speech text using the correspondence between preset medical terms and vernacular terms to obtain the injury information of the injured person.

[0166] In some embodiments, the error correction unit includes: a first determination subunit, used to determine the candidate word corresponding to the word in the speech text based on a preset recall strategy; the preset recall strategy includes a homophone recall strategy and / or a front and back nasal recall strategy; a replacement subunit, used to replace the word in the wrong position in the speech text with each of the candidate words, and obtain the corresponding replaced speech text; a second determination subunit, used to determine the perplexity of each replaced speech text based on each replaced speech text; a third determination subunit, used to determine the speech text with the lowest perplexity as the corrected speech text.

[0167] In some embodiments, the device for determining the injury level also includes: a word segmentation module, which is used to segment the speech text to obtain a word list; the word list includes the words after the speech text is segmented; a first determination module, which is used to determine the score of at least two adjacent words in the word list after splicing based on a preset word score; and a second determination module, which is used to determine the position corresponding to the word with a low score after splicing as an error position.

[0168] In some embodiments, the device for determining the injury level also includes: a second acquisition module, used to acquire a training set; the training set includes training samples of each category; the training samples include the first vital sign data of the injured person, the first injury information, the first injury level and the first injury type; a training module, used to use a pre-trained medical model to train part or all of the training samples in the training set to obtain the pre-trained injury classification model.

[0169] The description of the above device embodiment is similar to the description of the above method embodiment, and has similar beneficial effects as the method embodiment. In some embodiments, the functions or modules included in the device provided in the embodiment of the present application can be used to execute the method described in the above method embodiment. For technical details not disclosed in the device embodiment of the present application, please refer to the description of the method embodiment of the present application for understanding.

[0170] It should be noted that in the embodiment of the present application, if the above-mentioned method for determining the injury level is implemented in the form of a software function module and sold or used as an independent product, it can also be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the embodiment of the present application can be essentially or partly reflected in the form of a software product that contributes to the relevant technology. The software product is stored in a storage medium, including several instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the methods described in each embodiment of the present application. The aforementioned storage medium includes: various media that can store program codes, such as a U disk, a mobile hard disk, a read-only memory (ROM), a disk or an optical disk. In this way, the embodiment of the present application is not limited to any specific hardware, software or firmware, or any combination of hardware, software, and firmware.

[0171] An embodiment of the present application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program that can be executed on the processor, and when the processor executes the program, some or all of the steps in the above method are implemented.

[0172] The embodiment of the present application also provides a computer-readable storage medium on which a computer program is stored, and when the computer program is executed by a processor, some or all of the steps in the above method are implemented. The computer-readable storage medium can be transient or non-transient.

[0173] An embodiment of the present application also provides a computer program, including a computer-readable code. When the computer-readable code is run in a computing device, a processor in the computing device executes some or all of the steps for implementing the above method.

[0174] The present application also provides a computer program product, which includes a non-transitory computer-readable storage medium storing a computer program, and when the computer program is read and executed by a computer, some or all of the steps in the above method are implemented. The computer program product can be implemented specifically by hardware, software, or a combination thereof. In some embodiments, the computer program product is specifically embodied as a computer storage medium, and in other embodiments, the computer program product is specifically embodied as a software product, such as a software development kit (SDK), etc.

[0175] It should be noted here that the description of the various embodiments above tends to emphasize the differences between the various embodiments, and the same or similar aspects can be referenced to each other. The description of the above device, storage medium, computer program and computer program product embodiments is similar to the description of the above method embodiment, and has similar beneficial effects as the method embodiment. For technical details not disclosed in the embodiments of the device, storage medium, computer program and computer program product of this application, please refer to the description of the method embodiment of this application for understanding.

[0176] It should be noted that Figure 8 A schematic diagram of a hardware entity of a computer device in an embodiment of the present application is shown in FIG. Figure 8 As shown, the hardware entity of the computer device 800 includes: a processor 801, a communication interface 802 and a memory 803, wherein:

[0177] Processor 801 generally controls the overall operation of computer device 800 .

[0178] The communication interface 802 enables the computer device to communicate with other terminals or servers through a network.

[0179] The memory 803 is configured to store instructions and applications executable by the processor 801, and can also cache data to be processed or processed by the processor 801 and each module in the computer device 800 (for example, image data, audio data, voice communication data, and video communication data), which can be implemented by flash memory (FLASH) or random access memory (Random Access Memory, RAM). Data transmission between the processor 801, the communication interface 802 and the memory 803 can be carried out through the bus 804.

[0180] It should be understood that "one embodiment" or "an embodiment" mentioned throughout the specification means that specific features, structures or characteristics related to the embodiment are included in at least one embodiment of the present application. Therefore, "in one embodiment" or "in an embodiment" appearing throughout the specification does not necessarily refer to the same embodiment. In addition, these specific features, structures or characteristics can be combined in one or more embodiments in any suitable manner. It should be understood that in various embodiments of the present application, the size of the serial number of each step / process mentioned above does not mean the order of execution, and the execution order of each step / process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiment of the present application. The serial numbers of the embodiments of the present application mentioned above are for description only and do not represent the advantages and disadvantages of the embodiments.

[0181] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the sentence "comprises a ..." does not exclude the existence of other identical elements in the process, method, article or device including the element.

[0182] In the several embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as: multiple units or components can be combined, or can be integrated into another system, or some features can be ignored or not executed. In addition, the coupling, direct coupling, or communication connection between the components shown or discussed can be through some interfaces, and the indirect coupling or communication connection of the devices or units can be electrical, mechanical or other forms.

[0183] The units described above as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units; they may be located in one place or distributed on multiple network units; some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.

[0184] In addition, all functional units in the embodiments of the present application may be integrated into one processing unit, or each unit may be a separate unit, or two or more units may be integrated into one unit; the above-mentioned integrated units may be implemented in the form of hardware or in the form of hardware plus software functional units.

[0185] A person skilled in the art can understand that all or part of the steps of implementing the above method embodiment can be completed by hardware related to program instructions, and the aforementioned program can be stored in a computer-readable storage medium. When the program is executed, it executes the steps of the above method embodiment; and the aforementioned storage medium includes: mobile storage devices, read-only memories (ROM), magnetic disks or optical disks, etc., various media that can store program codes.

[0186] Alternatively, if the above-mentioned integrated unit of the present application is implemented in the form of a software function module and sold or used as an independent product, it can also be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application can essentially or in other words, the part that contributes to the relevant technology can be embodied in the form of a software product, which is stored in a storage medium and includes a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the methods described in each embodiment of the present application. The aforementioned storage medium includes: various media that can store program codes, such as mobile storage devices, ROMs, magnetic disks, or optical disks.

[0187] The above is only an implementation method of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application.

Claims

1. A method for determining an injury level, characterized in that: The method comprises: Obtaining vital sign data and injury information of the injured person; the injury information is used to describe the injury of the injured person; Generate an injury text based on the vital sign data and the injury information; The injury text is analyzed using a pre-trained injury classification model to obtain the injury grade and injury type of the injured person, and the injury grade and / or injury type are output.

2. The method for determining the injury level according to claim 1, characterized in that: The method further comprises: In the case where the injury level of the injured person is low, based on the injury type of the injured person, extracting self-rescue measures corresponding to the injury type from the rescue database; Based on the self-rescue measures, the injured person is guided by voice to perform self-rescue.

3. The method for determining the injury level according to claim 2, characterized in that: The method further comprises: The injury level, the injury type, and the acquired location of the injured are uploaded to a rescue center; the rescue center is used to provide rescue to the injured according to the number of the injured, the location of the injured, the injury level, and the injury type.

4. The method for determining the injury level according to claim 1, characterized in that: The obtaining of the vital sign data of the injured person includes: Collecting initial vital sign data of the injured person; the initial vital sign data includes at least one of heart rate data, blood oxygen data and respiratory rate data; When the change rate of two adjacent data in the same type of initial vital sign data is greater than a preset value, a segment of the same type of initial vital sign data is selected for averaging to obtain an average value of the same type of initial vital sign data; An average value of each type of the initial vital sign data is determined as the vital sign data.

5. The method for determining the injury level according to claim 1, characterized in that: The obtaining of the injury information of the injured person includes: Collecting and converting the voice describing the injury of the injured person into text to obtain voice text; The speech text is sequentially processed for error correction and terminology standardization to obtain the injury information of the injured person.

6. The method for determining the injury level according to claim 5, characterized in that: The step of sequentially performing error correction and terminology standardization processing on the voice text to obtain the injury information of the injured person includes: Correcting the speech text using candidate words corresponding to words in the speech text to obtain a corrected speech text; By using the correspondence between preset medical terms and vernacular terms, the corrected speech text is processed with term standardization to obtain the injury information of the injured person.

7. The method for determining the injury level according to claim 6, characterized in that: The step of correcting the speech text by using candidate words corresponding to words in the speech text to obtain the speech text after error correction includes: Based on a preset recall strategy, determining candidate words corresponding to the words in the speech text; the preset recall strategy includes a homophone recall strategy and / or a front and back nasal recall strategy; Using each of the candidate words to replace the words in the wrong position in the speech text, and obtaining the replaced speech text; Based on each of the replaced speech texts, determining the perplexity of each of the replaced speech texts; The speech text with the lowest confusion is determined as the speech text after error correction.

8. The method for determining the injury level according to claim 7, characterized in that: The method further comprises: Segmenting the speech text to obtain a word list; the word list includes the words after segmentation of the speech text; Based on the preset word scores, determining the scores of at least two adjacent words in the word list after being concatenated; The positions corresponding to the words with low scores after splicing are determined as wrong positions.

9. The method for determining the injury level according to any one of claims 1 to 8, characterized in that: The method further comprises: Acquire a training set; the training set includes training samples of each category; the training samples include the first vital sign data, first injury information, first injury level and first injury type of the injured person; The pre-trained medical model is used to train part or all of the training samples in the training set to obtain the pre-trained triage classification model.

10. A device for determining the level of injury, characterized in that: The device comprises: The first acquisition module is used to acquire the vital sign data and injury information of the injured person; the injury information is used to describe the injury of the injured person; A generating module, used for generating an injury text based on the vital sign data and the injury information; The analysis module is used to analyze the injury text using a pre-trained injury classification model to obtain the injury level and injury type of the injured person, and output the injury level and / or the injury type.

11. A computer device comprising a memory and a processor, wherein the memory stores a computer program that can be run on the processor, characterized in that: When the processor executes the program, the steps in the method according to any one of claims 1 to 9 are implemented.

12. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps in the method according to any one of claims 1 to 9 are implemented.

Images