JP2020013234A - Emergency notification listening support system and emergency notification listening support method - Google Patents

Abstract

To provide a technique which reduces the burden on a user of determining a situation and making a decision and acquires appropriate information.SOLUTION: A system which supports determining a condition of a sick or wounded person is provided. This system includes: a storage unit in which collected information about the sick or wounded person is stored; a determination unit which determines a condition of the sick or wounded person in accordance with the collected information about the sick or wounded person; and an estimation unit which estimates question items corresponding to information to be collected, from the collected information about the sick or wounded person and determines priorities of the question items obtained by estimation.SELECTED DRAWING: Figure 1

Description

  The present invention relates to a service for listening to an emergency report in emergency medical care, and to a technique for supporting the service.

  In recent years, the number of emergency reports from the general public has tended to increase due to social backgrounds such as an increase in the elderly population. Emergency resources such as ambulances and paramedics are limited, and there is a need for more efficient emergency medical services.

  As one of the measures against the above-mentioned problem, the degree of urgency of the victim (the extent to which the passage of time affects the danger of life) is determined from the contents of the hearing at the time of the emergency report, and the provision of emergency services according to the case is determined. Call triage has been proposed. Call triage requires a protocol to determine the degree of urgency, and the Fire and Disaster Management Agency and a group promoting call triage are developing and improving the protocol. Usually, protocols do not only determine the degree of urgency from one or more symptoms of the victim obtained by hearing, but also, when a certain symptom is obtained, ask one or more related questions related to it. It has been created so that it can be enumerated and facilitate the smooth listening work.

  While such a protocol is being created, Non-Patent Document 1 proposes an algorithm for creating a model for determining the degree of urgency from past cases of listening to an emergency report, and for accurately determining the degree of urgency.

OHSHIGE, Kenji, et al. Evaluation of an algorithm for estimating a patient's life threat risk from an ambulance call. BMC emergency medicine, 2009, 9.1: 21.

  In a situation where there is no time constraint, it is possible to easily determine the emergency level of the victim according to the emergency level determination protocol. However, in actual emergency medical practice, the strength of time constraints varies depending on the situation of the victim. In particular, when the condition of the victim is severe, the listening time is limited. The protocol may list multiple questions related to a symptom, but does not usually state which items should be given priority. This is because it is necessary to make a judgment in accordance with the situation, but in the occurrence phase of the injured and ill, it takes experience to judge and listen to the most important items.

  Therefore, there is a demand for a support technology for reducing the burden on the user for situation determination and decision making and obtaining appropriate information.

  One preferred aspect of the present invention is a system that assists in determining the condition of a patient. The system includes a storage unit that stores information about the collected victims, a determination unit that determines the state of the victim from the collected information about the victims, and information to be collected from the collected information about the victims. An estimating unit for estimating the corresponding question item and determining the priority of the question item obtained by the estimation.

  Another preferred aspect of the present invention is a method for assisting in listening to an emergency report, which uses an arithmetic device, a storage device, an input device, and an output device to assist in determining the state of a victim. The method includes a first step of inputting information about the victim from an input device, a second step of storing the input information about the victim in a storage device, and determining a state of the victim from the stored information about the victim. A third step of estimating a question item corresponding to the information to be collected from the stored information on the victim, and a fifth step of determining the priority of the question item obtained by the estimation. Executing the sixth step of displaying the question items according to the priority, and repeating the first step using the information collected in response to the display of the question items in the sixth step as information on a new patient. .

  ADVANTAGE OF THE INVENTION According to this invention, the burden on a user's situation determination and decision-making is reduced, and the support for acquiring appropriate information becomes possible.

FIG. 1 is a block diagram showing a configuration example of a first embodiment of an emergency call listening support system and a program. 9 is a flowchart illustrating an example of a processing procedure according to the first embodiment. FIG. 4 is an image diagram illustrating an example of an emergency report listening screen according to the first embodiment. FIG. 4 is a partially enlarged image view showing an example of an emergency report listening screen according to the first embodiment. FIG. 4 is a partially enlarged image view showing an example of an emergency report listening screen according to the first embodiment. FIG. 4 is a partially enlarged image view showing an example of an emergency report listening screen according to the first embodiment. FIG. 4 is a partially enlarged image view showing an example of an emergency report listening screen according to the first embodiment. 6 is a data flow diagram illustrating an example of a data flow according to the first embodiment. FIG. 4 is a block diagram showing a configuration example of a second embodiment of the emergency call listening support system and program. 9 is a flowchart illustrating an example of a processing procedure according to the second embodiment. FIG. 10 is an image diagram illustrating an example of an emergency report listening screen according to the second embodiment. 9 is a data flow diagram illustrating an example of a data flow according to the second embodiment.

  Hereinafter, embodiments will be described with reference to the drawings. Note that the present invention is not construed as being limited to the description of the embodiments below. It is easily understood by those skilled in the art that the specific configuration can be changed without departing from the spirit or spirit of the present invention.

  In the structures of the invention described below, the same portions or portions having similar functions are denoted by the same reference numerals in different drawings, and description thereof is not repeated in some cases. When there are a plurality of elements having the same or similar functions, the same reference numerals may be given different subscripts for explanation. However, when there is no need to distinguish a plurality of elements, the description may be omitted with suffixes omitted.

  Notations such as “first”, “second”, and “third” in this specification and the like are used to identify constituent elements, and necessarily limit the number, order, or content thereof. is not. Also, numbers for identifying components are used for each context, and numbers used in one context do not necessarily indicate the same configuration in another context. Also, this does not prevent a component identified by a certain number from also having a function of a component identified by another number.

  The position, size, shape, range, or the like of each component illustrated in the drawings and the like is not accurately represented in some cases in order to facilitate understanding of the present invention. For this reason, the present invention is not necessarily limited to the position, size, shape, range, and the like disclosed in the drawings and the like.

  An outline of an embodiment described in detail below will be described. The proposed emergency report listening support system and program are support systems for assisting the determination of the degree of urgency, based on the hearing of the reporter, input means for inputting information on the victim and output for displaying information to the user. Means. A listening information input storage unit for storing the input listening content; and a question estimation feature amount for extracting a question estimation feature amount for estimating a question item to be heard from the input listening content or date. Extraction means. In addition, a question candidate estimating means having a candidate estimating model for estimating a question candidate using the question estimating feature amount as an input, and a question to be heard next among a plurality of question items obtained from the question candidate estimating means is optimized. And a question highlighting means for highlighting and displaying the optimized question candidate on the output means.

  Furthermore, to construct a candidate estimation model, from the listening order history table storing the features of the past listening content and the features of the listening order, and from other variables such as the content and date of the listening information input storage means, An urgency determination feature amount extraction means for extracting a feature amount for performing an urgency degree determination; and an urgency degree determination processing means having an urgency degree determination model for determining an urgency degree and a suspicious disease name by using the emergency degree determination feature amount as input. Prepare. Also, an urgency determination output means for displaying a judgment result obtained from the urgency judgment processing means on an output means, and an emergency listening which stores a feature amount of past listening contents used for constructing an emergency judgment model. It has a history table and a suspicious disease name table which is also used to construct an urgency determination model. In addition, a question table for storing text data of a question item is provided.

  Further, in a preferred embodiment, in order to determine whether the caller is in an emergency, a voice input unit for obtaining a call voice as an input, and a call voice feature amount extraction unit for extracting a feature amount from the call voice. An emergency judgment processing means having an emergency judgment model for judging an emergency by using the speech voice feature quantity as an input; and an emergency judgment output means for displaying a judgment result obtained from the emergency judgment processing means on the output means. And a call voice history table that stores a feature amount of past call voice and is used to construct an emergency judgment model.

  FIG. 1 is a diagram illustrating an example of a configuration of an emergency report listening support system according to the present embodiment. In the present embodiment, a method of indicating a question item and its order to the user will be described. The emergency report listening support system 100 of the present embodiment includes a processing terminal 101 such as a server or a PC (personal computer), a user input device 102 such as a mouse and a keyboard, and a user output device 103 such as a display. It consists of. The emergency report listening support system 100 is operated by a user (operator). The processing terminal 101 includes an arithmetic device 104 and a storage device 105. The storage device 105 stores a question candidate estimating unit 106 and an urgency determining unit 113 as programs executed by the arithmetic device 104.

  The question candidate estimating unit 106 for estimating the next and subsequent question items based on the content of the listening includes a question candidate estimating unit 107 for performing an estimating process, a candidate estimating model 108 which is a machine learning model used for estimating, and And a question estimation feature amount extraction unit 109 for extracting a feature amount as follows. Further, an order optimizing unit 111 for optimizing the plurality of question candidates obtained by the estimation in the order to be heard and a question candidate highlighting display unit 112 for highlighting and displaying the question candidates to be heard on the output device 103 are provided. Prepare. In addition, a listening order history table 110 storing past listening information (question items and corresponding listening contents) and a listening order is provided for use in constructing a machine learning model. Further, a question table 121 for storing text data of a question item corresponding to the question item ID is provided. The contents of the question table 121 include, for example, text displayed on a display screen described later with reference to FIG. 3A.

  Further, an urgency determination unit 113 that determines the state of the patient, for example, urgency, based on the input of the listening content, includes an urgency determination processing unit 114 that performs determination processing, and an urgency determination model that is a machine learning model used for the determination. 115, and an urgency determination feature value extraction unit 116 that extracts a feature value that is input to the determination process. It also includes a suspicious disease name table 118 that stores a list of suspicious disease names, and an urgency determination output unit 119 that displays an urgency and a suspicious disease name on the output device 103. Further, an emergency listening history table 117 storing past listening information is provided for use in constructing a machine learning model. The emergency listening history table 117 may be replaced with the listening order history table 110.

  In the present embodiment, the candidate estimation model 108 and the urgency determination model 115, which are machine learning models, are configured by a neural network. As described above, the machine learning model uses the data of the listening order history table 110 and the emergency listening history table 117 that store the past listening information, uses these data to which a human has given a correct answer as teacher data, I decided to learn by learning. The learning method of the neural network itself may follow a known technique.

  For example, the candidate estimation model 108 uses information such as obtained symptoms as problem data and corrects what kind of question to ask next for a skilled doctor or the like to identify or further narrow down the suspected disease name and urgency. The model is learned as data. In addition, for example, the urgency determination model 115 learns the model by using information such as obtained symptoms as problem data, and as a correct answer data what kind of suspected disease and urgency are determined by a skilled doctor or the like. . In the candidate estimation model 108, since a new question is a question other than a question asked in the past, it is preferable that the data of the listening order history table 110 used for learning and training of the model include information on the order of the question.

  The storage device 105 includes a listening information input storage unit 120 for storing listening information to be input to the question candidate estimation unit 106 and the urgency determination unit 113. In the present embodiment, functions such as calculation and control are realized by executing a program stored in the storage device 105 of the PC by the arithmetic device 104 in cooperation with other hardware. . A program executed by a PC or the like, its function, or means for realizing the function may be referred to as “function”, “means”, “unit”, or the like. The above configuration may be configured by a single computer, or any part of the input device 102, the output device 103, the arithmetic device 104, and the storage device 105 may be configured by another computer connected via a network. You may.

  It should be noted that the candidate estimation model 108 and the urgency determination model 115 may be configured by a model that leads to a conclusion by conditional branching instead of artificial intelligence. For example, in the urgency determination model 115, when "pain in the chest and upper abdomen" and "severe pain" are obtained as information, "angina pectoris" and "myocardial infarction" are extracted as names of suspected diseases. In the above case, the candidate estimation model 108 preferentially extracts “Is pain persistent?” As a question candidate, and identifies “angina” and “myocardial infarction”. Such a model is separately created by the user and stored in the storage device 105.

  FIG. 2 is a flowchart describing the procedure of the processing of this embodiment from receiving the emergency report to ending the listening of the emergency report. Hereinafter, description will be made along each step of the flowchart.

  In step S201, the user of the emergency report listening support system 100 receives an emergency report. The report is made via a telephone line or the Internet. It is assumed that the reporter makes a report by voice from his / her telephone or smartphone.

  In step S202, the user starts listening from the reporter. First, for example, the name, age, and gender of the victim, the relationship between the reporter and the victim, and the like are heard, and information is input to the emergency report listening support system 100 using the input device 102. The input information is stored in the listening information input storage unit 120.

  In step S203, following step S202, the user listens to the condition and symptoms of the patient. When the caller voluntarily starts explaining, the user listens to the content. If not, the user starts communication of the condition / symptom so that the explanation is easy to start, such as whether the patient is injured or sick.

  In step S <b> 204, the user inputs the contents of the reporter's response to the screen that is the input device 102 of the emergency report listening support system 100. As will be described later with reference to FIG. 3A, the input screen is a method of selecting an item by clicking on the corresponding item, and basically assumes input by a mouse click or a tap on a touch panel. When an item is selected, the listening information input storage unit 120 stores the ID of the item and information indicating the order of the input item. The item is usually a question itself or a content corresponding to the question, and for example, a pull-down input as shown in FIG. 3A or the like is possible. Steps S201 to S204 are the input processing to the emergency report listening support system 100 in which the user who is the operator of the system intervenes. It is also possible for the user to perform input processing using a voice synthesis or voice recognition interface instead of having a direct conversation with the reporter.

  In step S205, the urgency determination unit 113 determines the degree of urgency and the name of the suspicious disease using the listening information input to the listening information input storage unit 120. The urgency determination feature value extraction unit 116 performs a process for obtaining a feature value to be input to the urgency determination processing unit 114 that performs the determination. The feature amount can be obtained by directly using the information recorded in the listening information input storage unit 120 and information such as date and time, weather, and country, or by processing such information. By inputting the obtained feature amount to the urgency determination processing unit 114 having the learned urgency determination model 115, the resulting urgency and suspicious disease ID are output. The algorithm for determining the degree of urgency is also reported in Non-Patent Document 1.

  In step S206, the urgency determination output unit 119 obtains the urgency and the suspicious disease ID obtained as a result of step S205, and displays them on the output device 103. At this time, the suspicious disease name is preferably displayed on the output device, and the suspicious disease name ID is converted into the suspicious disease name using the suspicious disease name table 118.

  In step S207, the question candidate estimating unit 106 estimates the next question candidate to be heard using the listening information input to the listening information input storage unit 120. The question estimation feature amount extraction unit 109 performs a process for obtaining a feature amount to be input to the question candidate estimation processing unit 107 for performing estimation. The feature amount can be obtained by directly using the information recorded in the listening information input storage unit 120 or information such as date and time, weather, and land, or by processing the information. By inputting the feature amount obtained in this way to the question candidate estimation processing unit 107 having the learned candidate estimation model 108, a score indicating how suitable each question item ID is for the next question is obtained. Is output.

  In step S208, the order optimizing unit 111 obtains each question item ID and its score as a result of step S208, and sorts them in descending order of the score. Then, by excluding scores below a certain value or by obtaining only a predetermined number of question items from the top of the score, optimize the order according to the score and extract only questions with a relatively high score I do.

  In step S209, the question candidate highlighting display unit 112 acquires the question item ID and the score thereof that have been optimized in order, and displays them on the output device 103. At this time, it is preferable to display the question items on the output device, and the question item ID is converted into the question items using the question table 121. As shown in FIG. 2, the processing of the question candidate estimating unit 106 and the processing of the urgency determining unit 113 are executed substantially in parallel.

  FIG. 3A is a diagram illustrating an example of a screen displayed on the output device 103 by the question candidate highlighting unit 112. FIGS. 3B, 3C, 3D, and 3E show enlarged views of ranges 3001, 3002, 3003, and 3004 indicated by dotted lines, respectively. Basically, this screen is displayed on the output device 103 during the user's work, and the content changes as the user's input work progresses. In the present embodiment, the item corresponding to the question is displayed on the screen, and when the input is made, the display is changed so that the input can be determined. For example, in FIG. 3B, “16 to 39 years old (youth)” has been input in the items of “basic information” and “age”, and the display is inverted. “Pale” has already been entered in the “vital sign” and “face” fields. As can be seen from the figure, since there are a wide variety of question items, it is not practical to fill in all items, especially in an emergency. In this embodiment, the system assists the user by determining the importance of the item to be asked and displaying the listening priority.

  Here, as shown in FIG. 3C, the question candidate highlighting unit 112 highlights a screen area on which information corresponding to the corresponding question item ID is displayed, with a highlight frame 301. When there are a plurality of question items to be highlighted, the screen area displaying the question item with the highest score is highlighted with the darkest highlighting frame 301, and the question item with the next highest score is displayed. By highlighting the screen area with a lighter color with a semi-emphasized frame line 302, the user can know which question item estimation unit 106 has determined which question item should be heard next most. . In addition to displaying the content of the question item and the question candidate display area 303 displaying the score in the screen area in conjunction with the display method, the user can know the estimation result of the question candidate estimation unit 106. FIG. 3E shows that the score of “presence or absence of numbness” is 0.86 and the score of “presence or absence of paralysis” is 0.65. Note that the display method shown in each figure in FIG. 3 is an example, and as other examples of the highlighting, various modes such as changing a color, changing a size, and hiding a low priority one are considered. Can be

  In step S210, as shown in FIG. 3E, the user is displayed based on the judgment result 305 of the degree of urgency and suspicious disease displayed on the output device 103, the question candidate and the contents of the previous hearing and the strength of the time constraint, and the like. It is determined whether the listening should be terminated or the listening should be continued based on the urgency level. If the urgency determination unit 113 cannot determine the degree of urgency and the name of the suspected disease depending on the content of the hearing so far and allows a case where there is no determination result, the urgency determination cannot be performed, and the listening is continued.

  In step S211, when listening is continued, step S212 is performed, and when listening is ended, step S213 is performed.

  In step S212, the next question is performed to continue listening. The next question item is determined with reference to the candidates displayed on the output device 103 as a result of step S209. The answer obtained in the question is input in step S204 and stored in the listening information input storage unit 120. If the listening is not finished, urgency determination and question candidate estimation are further performed based on the contents of the listening information input storage unit 120.

  In step S213, the listening ends. The user who has completed the hearing performs the work after the hearing, such as a request for dispatch of an ambulance, and starts the hearing if there is a next emergency report. When an ambulance is requested to be dispatched, the content of the hearing and the urgency and the name of the suspected illness as a result of the hearing are transmitted to emergency personnel to be dispatched.

  FIG. 4 is an example of a data flow diagram when processing is performed according to the flowchart of FIG. The data flow 401 describes the input of the hearing information of the victim by the user in steps S202 and S204. The data flow 402 describes that the input listening information is stored in the listening information input storage unit 120.

  In step S205, the data flow 403 is a process in which the urgency determination feature value extraction unit 116 reads out information necessary for calculating the urgency determination feature value, and in step S207, the question estimation feature value extraction unit 109 determines whether the question estimation feature value Describes the data flow of the process of reading the information necessary for the calculation of. After reading the data, the arithmetic unit 104 calculates the urgency determination feature amount and the question estimation feature amount.

  The data flow 404 describes a process of reading the urgency determination model 115 in step S205. The arithmetic unit 104 determines the degree of urgency and the name of the suspicious disease from the urgency determination feature amount and the degree of urgency determination model 115.

  The data flow 405 describes that the urgency determination output unit 119 outputs the urgency and the suspicious disease name to the output device 103 in step S206.

  The data flow 406 describes the process of reading the candidate estimation model 108 in step S207. The arithmetic unit 104 estimates a question candidate from the question estimation feature amount and the candidate estimation model 108.

  The data flow 407 describes that the question candidate highlighted display unit 112 outputs the question candidate whose order has been optimized by the order optimization unit 111 to the output device 103 in step S209.

  The data flow 408 describes a process of inputting the hearing information of the patient again in step S204 after determining to continue listening in step S211. Thereafter, exchange of data of the data flows 402 to 407 is repeatedly executed, but the description in FIG. 4 is omitted.

  In the above, the flow of the process of optimizing the item to be the next question candidate and presenting it to the user when the input of the listening is started has been described with reference to the flowchart of FIG. 2 and the data flow diagram of FIG. . By these processes, the question candidate estimation processing unit 107 estimates the question items to be heard, and the order optimization unit 111 determines the items to be asked next, thereby reducing the burden on the user for situation determination and decision making. Can be reduced.

  In the above-described embodiment, an example in which the processing is performed on one computer has been described. However, if the data flow diagram of FIG. 4 can be realized, the input device 102, the output device 103, the arithmetic device 104, and the storage device 105 Need not be the same computer. The storage device 105 does not need to be a single device as long as it can transmit and receive data to and from each other.

  According to the embodiment described above, the question candidate estimation processing unit 107 estimates a question item to be heard, the order optimization unit 111 determines an item candidate to be asked next, and the question candidate highlight display unit 112 emphasizes the item. indicate. As a result, it is possible to reduce the burden on the user for situation determination and decision making, and to quickly obtain appropriate information.

  In the second embodiment, after the reporter emergency determination processing unit determines the strength of the time constraint of the listening, the question candidate estimation processing unit estimates the question items to be heard, and the order optimizing unit determines the next An example of determining an item to be asked will be described. The same components as those in the first embodiment are denoted by the same reference numerals, and description thereof is omitted. Here, the emergency situation refers to a situation where the time constraint is determined to be stronger than usual. Therefore, processing such as limiting the number of question candidates presented to the user is performed.

  FIG. 5 is a diagram illustrating a configuration example of the support system 500 according to the present embodiment. The processing terminal 501 has a configuration in which a reporter emergency situation determination unit 521 is added to the processing terminal 101 shown in FIG. A reporter emergency state determination unit 521 that determines whether an emergency such as the safety of the victim or the reporter's whereabouts is not ensured includes an emergency state determination processing unit 522 that performs determination processing, and a machine used for the determination. An emergency judgment model 523 which is a learning model, a call sound input unit 524 for acquiring a call sound used for the judgment, a call sound feature extraction unit 526 for extracting a feature amount to be input for the judgment processing, and a past call sound It comprises a stored call voice history table 525 and an emergency judgment output unit 527 for displaying a judgment result on the output device 103.

  FIG. 6 is a flowchart describing the procedure of the processing according to the present embodiment from receiving the emergency report to finishing listening to the emergency report. Hereinafter, description will be made along each step of the flowchart.

  In step S605, the reporter emergency situation determination unit 521 performs the determination. First, the call voice input unit 524 acquires call voice data input from the input device 102. The call voice feature extraction unit 526 performs a process for analyzing the call voice and obtaining a feature amount to be input to the emergency judgment processing unit 522 that performs the determination. The call voice data includes the environmental sound around the caller in addition to the voice of the caller, and the feature amount is extracted for one or both of them. By inputting the obtained feature amount to the emergency determination processing unit 522 having the learned emergency determination model 523, an emergency determination result is output.

  For example, at an accident site on a highway, it is desirable to immediately evacuate off the road to ensure the safety of the caller. The call voice feature extraction unit 526 detects, for example, a feature amount of noise caused by the traveling of the car on the highway included in the call voice data, and determines that the notification is from the highway. Alternatively, the stress state is detected from the inflection of the caller's voice, and the urgency of the situation is determined.

  In step S606, the emergency judgment output unit 527 outputs the information on the emergency judgment result obtained in step S605 to the storage device 105 so that the question candidate estimation unit 106 can use it. The emergency judgment result output to the storage device 105 is used in step S207b described later. Further, the emergency judgment output unit 527 may output information on the emergency judgment result to the output device 103. In that case, by displaying an icon in a partial area of the screen used by the user for listening, the user knows that the caller emergency determination unit 521 has determined that the call is being performed in an emergency situation. be able to. Thereby, the user can confirm with the informant whether the emergency can be avoided.

  In steps S207b to S209b, similarly to the first embodiment, the question candidate estimating unit 106 performs estimation, order optimization, and display of question candidates using the listening information input to the listening information input storage unit 120. However, in the second embodiment, when an output indicating an emergency is obtained from the emergency determination output unit 527, for example, at least one of the following processes is performed.

  (1) In step S207b, an emergency-specific question is added in addition to a normal question. For example, "Is your place on a highway?" If the answer to the above-mentioned question is “Yes”, an emergency-specific instruction such as “Please evacuate immediately to a safe place” may be selected instead of the question.

  (2) In step S208b, the order is optimized so that questions and instructions specific to an emergency are given priority over ordinary questions. The question or instruction specific to an emergency may be stored in the question table 121 as an item dedicated to the emergency with the maximum score value being added in advance. In step S209b, the question candidate highlighting display unit 112 acquires the order-optimized question item IDs and their scores, and displays them on the output device 103.

  (3) In step S207b, the question candidate estimation processing unit 107 estimates a question candidate using a candidate estimation model dedicated to an emergency. In emergency situations, questions and answers are often time-limited, so questions should be limited to those that are more important than usual and the question-and-answer time reduced.

  (4) In step S208b, the order optimizing unit 111 performs a process of limiting the number of question items. For this purpose, for example, the order optimization unit 111 increases the threshold value of the score for excluding the question item.

As described above, when it is determined that an emergency has occurred, the number of question items to be presented to the user is reduced by changing the algorithm of each step. FIG. 7 is a diagram showing an example of a screen to be displayed. As in FIG. 3A, the candidate to be asked next is highlighted. Further, in the screen example of FIG. 7, in the question candidate display area 701, an icon “!” Indicating an emergency is displayed next to the score. Thus, the user can know that the questionrer emergency determination unit 521 has determined that an emergency has occurred and is presenting a question candidate.

  FIG. 8 is an example of a data flow diagram when processing is performed according to the flowchart of FIG. The description of the same flow as in FIG. 4 will be omitted, and the parts unique to the second embodiment will be described. The data flow 801 acquires, in addition to the user's input of the hearing information of the victim, the background sound around the reporting location, position information, and the like, which are used in the reporter emergency judgment S605.

  The data flow 802 describes that the input listening information and background sound are stored in the listening information input storage unit 120. The data flow 803 describes a process of reading the emergency judgment model 523 in step S605. The arithmetic unit 104 determines whether or not an emergency has occurred from the emergency determination feature amount and the emergency determination model 523, and uses the result for question candidate estimation in step S207b. Subsequent data flows 403 to 809 are the same as those in the first embodiment except that questions, instructions, and information specific to an emergency are added with priority.

  As described above, according to the flowchart of FIG. 6 and the data flow diagram of FIG. 8, it is determined whether the injured or injured person is in an emergency, and the next question candidate item is optimized in order and presented to the user. Has been described. With these processes, when the question candidate estimation processing unit 107 estimates a question item to be heard, the question optimization unit 111 estimates a question item to be heard after determining whether it is an emergency, and the order optimization unit 111 By determining the item to be asked next, it is possible to further reduce the burden on the user in determining the situation and making a decision.

  In the first embodiment, an example in which the processing is performed on one computer is described. However, if the data flow diagram of FIG. 4 can be realized, the input device 102, the arithmetic device 104, and the storage device 105 are the same computer. No need. The storage device does not need to be a single device as long as it can transmit and receive data to and from each other. FIG. 5 illustrates an example in which the input device 102 and the output device 103 are connected to the processing terminal 501 via the network NW. For example, the input device 102 and the output device 103 are, for example, a portable information processing terminal or a personal computer, and the processing terminal 501 is a server capable of performing more complicated processing. The server includes an output device 103 for displaying a question item obtained by the estimation, and an interface IF for communicating via the network NW with at least one of the input devices 102 for inputting information on the collected victims. ing.

  As described above, in an actual emergency medical setting, the strength of the time constraint varies depending on the situation where the victim or the informant is placed. In particular, in the case of an emergency such as a situation in which the location of the victim or the informant is not secured, the listening time is very limited. The protocol may list multiple questions related to a symptom, but does not usually state which items should be given priority. This is because it is necessary to judge according to the situation, but in an emergency situation, it is necessary to have experience to judge and listen to the most important items. Also, judging whether an emergency is in the first place requires experience. In the above embodiment, after the emergency determination processing unit determines the strength of the time constraint of the listening, the question candidate estimation processing unit estimates the question item to be heard, and the order optimization unit should ask the next question. By determining the items, it is possible to further reduce the burden on the user in determining the situation and making a decision.

  101: processing terminal, 102: input device, 103: output device, 104: arithmetic device, 105: storage device, 106: question candidate estimation unit, 107: question candidate estimation processing unit, 108: candidate estimation model, 109: question estimation Feature amount extraction unit, 110: listening order history table, 111: order optimization unit, 112: question candidate highlighting unit, 113: urgency determination unit, 114: urgency determination processing unit, 115: urgency determination model, 116 : Emergency level judgment feature quantity extraction unit, 117: Emergency listening history table, 118: Suspicious disease name table, 119: Urgency level determination output unit, 120: Listening information input storage unit

Claims (15)

A system that assists in determining the condition of a patient,
A storage unit for storing information on the collected victims;
From the collected information about the victim, a determination unit that determines the state of the victim,
From the information about the collected victim, an estimation unit that estimates a question item corresponding to the information to be collected, and determines the priority of the question item obtained by the estimation,
An emergency call listening support system characterized by having: The determination unit and the estimation unit operate in parallel,
The emergency report hearing support system according to claim 1. The storage unit stores information corresponding to the question item obtained by the estimation,
The emergency report hearing assistance system according to claim 1, wherein the determination unit further determines the state of the patient from information corresponding to the question item obtained by the estimation. The estimating unit includes a question estimation feature amount extraction unit, a question candidate estimation processing unit, a question table that stores question items, and an order optimization unit.
The question estimation feature amount extraction unit extracts a feature amount from information stored in the storage unit,
The question candidate estimation processing unit receives the feature amount as input, and attaches a score to the question item stored in the question table using a candidate estimation model that is a machine learning model,
The order optimization unit sorts the question items according to the score,
The emergency report hearing support system according to claim 1. The order optimization unit excludes scores below a certain value,
The emergency report hearing support system according to claim 4. The order optimizing unit obtains only a predetermined number of question items from a higher score,
The emergency report hearing support system according to claim 4. An output device that displays a question item obtained by the estimation,
The display of the priority in the output device is at least one of highlighting and score display,
The emergency report hearing support system according to claim 1. An output device for displaying the question item obtained by the estimation, and an interface for communicating via a network with at least one of the input devices for inputting information on the collected patient,
The emergency report hearing support system according to claim 1. It further includes an emergency judgment section,
The emergency determination unit includes a call voice input unit, a call voice feature extraction unit, an emergency determination processing unit, and an emergency determination output unit,
The call voice input unit inputs a call voice obtained when collecting information on the patient,
The call voice feature extraction unit extracts a call voice feature amount from the call voice,
The emergency determination processing unit receives the call voice feature amount as input, and determines the presence or absence of an emergency using an emergency determination model that is a machine learning model.
The emergency determination output unit outputs the presence or absence of the emergency,
The emergency report hearing support system according to claim 4. The order optimization unit performs a process of limiting the number of the question items based on an output of the emergency determination output unit,
The emergency report listening support system according to claim 9. The order optimizing unit performs a process of sorting by giving priority to a question item dedicated to an emergency, based on an output of the emergency determination output unit.
The emergency report listening support system according to claim 9. The question candidate estimation processing unit, based on the output of the emergency judgment output unit, estimates a question candidate using a candidate estimation model dedicated to emergency,
The emergency report listening support system according to claim 9. Using an arithmetic device, a storage device, an input device, and an output device, an emergency call listening support method for supporting determination of the state of the patient,
A first step of inputting information about the victim from the input device;
A second step of storing the input information on the patient in a storage device;
A third step of determining the state of the patient from the stored information on the patient;
A fourth step of estimating a question item corresponding to information to be collected from the stored information on the victim;
A fifth step of determining the priority of the question item obtained by the estimation;
Executing a sixth step of displaying the question item according to the priority,
Repeating the first step as information collected in response to the display of the question item in the sixth step as information on a new patient;
Emergency call listening support method. A seventh step of inputting information on a place where the victim exists from the input device;
Performing an eighth step of determining the presence or absence of an emergency from the information on the place where the victim exists,
Changing at least one of the algorithm of the fourth step and the algorithm of the fifth step based on the presence or absence of the emergency;
14. The support method for listening to an emergency report according to claim 13. Reducing the number of displayed question items in the sixth step by changing the algorithm;
The support method for listening to an emergency report according to claim 14.

Images