Emergency treatment green channel information management method and system and storage medium

Abstract

The invention discloses an emergency treatment green channel information management method and system and a storage medium, and belongs to the technical field of medical data processing.The method comprises the steps that a plurality of first sensors are arranged in a hospital area, a second sensor is worn on the body of a patient, and the first sensors generate stay information of the second sensor; generating a processing node based on the stay information, and generating a first processing timeline of the patient based on the processing node; recording a first video based on the portable recorder, recording a second video based on the fixed camera, and segmenting the first video and the second video to obtain a first sub-video and a second sub-video corresponding to the processing node; analyzing the first sub-video and the second sub-video to obtain operation process information and cooperation process information, and respectively mapping the operation process information and the cooperation process information to a first processing timeline to obtain a second processing timeline; and finally evaluating the treatment process of each second processing timeline based on the electronic medical record. According to the invention, the management and evaluation of the emergency treatment process are more accurate and comprehensive.

Description

Technical Field

[0001] The present invention belongs to the technical field of medical data processing, and in particular relates to an emergency green channel information management method, system and storage medium. Background Art

[0002] The emergency green channel refers to a fast and orderly diagnosis and treatment channel for critically ill patients, aiming to shorten the treatment time and improve the success rate of treatment. In this context, how to optimize the emergency treatment process through technical means and improve the work efficiency of medical staff has become an important research direction in the field of medical information management.

[0003] For example, the Chinese patent document with publication number CN109003664A discloses an emergency green channel information management method and device, which manages the emergency green channel emergency process through smart terminals A and B, positioning equipment, information management platform and server. Smart terminal A collects patient identity information and user account information, smart terminal B obtains patient identity information and binds the patient, and reads location tag information and current server time at the same time. The server stores and processes data to realize intelligent collection of emergency green channel patient full process treatment time, medical staff arrival time information, etc., thereby improving the management efficiency and accuracy of the emergency treatment process.

[0004] However, the solution in the above document mainly focuses on the collection and storage of information, and has limited capabilities for real-time monitoring and integrated analysis of video data during the treatment process. Secondly, when processing the timeline information of the patient's treatment process, the solution lacks in-depth mining and analysis of video data, and cannot comprehensively evaluate the efficiency and effectiveness of each link in the treatment process. Summary of the invention

[0005] To solve the above problems, the present invention provides an emergency green channel information management method, system and storage medium to solve the problems existing in the prior art.

[0006] In order to achieve the above-mentioned invention object, the present invention proposes an emergency green channel information management method, comprising: A plurality of first sensors are arranged in the hospital area, a second sensor is worn on the patient's body, and when the second sensor reaches the detection range of the first sensor, the first sensor generates stay information; generating a processing node based on the stay information, and generating a first processing timeline of the patient based on the processing node; Recording a first video based on a portable recorder and recording a second video based on a fixed camera, segmenting the first video and the second video according to the first processing timeline to obtain a first sub-video and a second sub-video corresponding to the processing node; Analyze the first sub-video and the second sub-video to obtain operation process information and coordination process information respectively, and map the first sub-video, the second sub-video, the operation process information and the coordination process information to the first processing timeline respectively to obtain a second processing timeline; The electronic medical record of the patient is obtained, a disease manifestation is assigned to each second treatment timeline based on the electronic medical record, and the second treatment timelines of the same disease manifestation are analyzed to evaluate the treatment process.

[0007] Further, analyzing the first sub-video and the second sub-video includes the following steps: Define the first sub-video and the second sub-video as target videos, split the target video into multiple static frames, identify each of the static frames based on a deep neural network, obtain an operation event contained therein, merge the static frames of the same operation event to obtain an operation video corresponding to the operation event, determine the occurrence time period of the operation event according to the position of the operation video in the target video, divide the video length of the target video into multiple time windows based on the occurrence time period, and construct a first feature matrix with the time windows as row features and the operation events as column features; Preset arrangement rules, rearrange the first feature matrix based on the arrangement rules to obtain a second feature matrix, perform feature recognition on the second feature matrix based on a convolutional neural network to extract feature events from the second feature matrix, and the feature events include the operation process information and the coordination process information.

[0008] Furthermore, dividing the time window comprises the following steps: An initial window duration is set, and the video duration is divided into a plurality of equally spaced first windows based on the initial window duration. The proportion of the occurrence time period of the operation event in each of the first windows is obtained. If the proportion is greater than a first threshold, the operation event is defined to have occurred in the corresponding first window. The event overlap rate of adjacent first windows including the operation events is counted. If the event overlap rate is greater than a second threshold, two first windows are merged into a second window. The density of the operation events included in the second window is counted. The second window with the density greater than a third threshold is located, and the second window is split to obtain a third window and a fourth window. The unmerged first window, the undivided second window after merging, the third window, and the fourth window are defined as the time window.

[0009] Furthermore, performing window segmentation on the second window includes the following steps: The first window is selected as the split window in the second window, and the second window is split into a first sub-window and a second sub-window based on the split window, and a first number and a second number of the operation events included in the first sub-window and the second sub-window are counted, and a difference value between the first number and the second number is calculated, and another first window is selected as the split window from the second window again and split again, and the difference value obtained after the split is calculated, and this step is repeated until the first windows in the second window are traversed, and the first sub-window and the second sub-window corresponding to the minimum difference value are located as the segmentation result of the second window.

[0010] Further, projecting the first sub-video and the second sub-video into the first processing timeline includes the following steps: Define the first sub-video and the second sub-video as target sub-videos, set a tag in the target sub-video, the tag corresponds to the processing node, set a first storage area and a second storage area, redundantly store the target sub-video in the first storage area and the second storage area, losslessly compress the target sub-video to obtain a first compressed file, and when checking the processing node, decompress the first compressed file in the first storage area to obtain the corresponding target sub-video; If the existence time of the first compressed file reaches the first retention time, the first compressed file in the first storage area is deleted, and the second storage area performs lossy compression on the first compressed file to obtain a second compressed file. When viewing the processing node, the second compressed file in the second storage area is decompressed to obtain the corresponding target sub-video. If the existence time of the second compressed file reaches the second retention time, the target sub-video is deleted. When viewing the processing node, the corresponding operation process information or the coordination process information is displayed.

[0011] Furthermore, a viewing time is preset, and the first compressed file or the second compressed file is restored before the viewing time is reached.

[0012] Further, evaluating the treatment process includes the following steps: The disease manifestations of the patient are extracted based on the electronic medical record, and the disease manifestations are used as the treatment attributes of the second processing timeline. The second processing timeline is clustered based on the treatment attributes to obtain symptom clusters, and corresponding historical data are acquired based on the symptom clusters. The historical data are analyzed to obtain a standard processing procedure, and the standard processing procedure includes time standards, mandatory operations, and standard collaboration modes. Time efficiency, operation specification, and team collaboration are evaluated for each second processing timeline based on the standard processing procedure. The treatment level is determined based on the evaluation results, and the treatment process is evaluated based on the treatment level.

[0013] Furthermore, the stay information includes arrival time, departure time and location information of the second sensor.

[0014] The present invention also provides an emergency green channel information management system, which is used to implement the above-mentioned emergency green channel information management method, and the system includes: The sensing module includes a plurality of first sensors arranged in a hospital area, a second sensor worn on a patient's body, and when the second sensor reaches a detection range of the first sensor, the first sensor generates stay information; a first analysis module, generating a processing node based on the stay information, and generating a first processing timeline of the patient based on the processing node; A second analysis module, based on recording the first video with a portable recorder and recording the second video with a fixed camera, the second analysis module divides the first video and the second video according to the first processing timeline to obtain the first sub-video and the second sub-video corresponding to the processing node, analyzes the first sub-video and the second sub-video to obtain operation process information and coordination process information respectively, and maps the first sub-video, the second sub-video, the operation process information and the coordination process information to the first processing timeline respectively to obtain a second processing timeline; An evaluation module is used to obtain the patient's electronic medical record, assign a disease manifestation to each second processing timeline based on the electronic medical record, analyze the second processing timelines with the same disease manifestation, and evaluate the treatment process.

[0015] The present invention also discloses a computer storage medium, wherein the computer storage medium stores program instructions, wherein when the program instructions are executed, the device where the computer storage medium is located is controlled to execute the above-mentioned method.

[0016] Beneficial effects: The present invention realizes real-time monitoring of the patient's movement path by deploying multiple first sensors and second sensors worn by the patient in the hospital area, and then generates a first processing timeline in combination with the patient's stay information at various locations, records video data through a portable recorder and a fixed camera, and identifies it to obtain operation process and coordination process information, maps it to the processing timeline, and forms a second processing timeline, which makes it easy to view the process nodes of the emergency treatment process. In addition, by obtaining the patient's electronic medical record and assigning disease manifestations, it is possible to perform cluster analysis on the processing timeline of the same disease manifestation, evaluate the efficiency and effectiveness of the treatment process, and thus optimize the emergency treatment process. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 A flowchart of the steps of an emergency green channel information management method of the present invention; Figure 2 It is a schematic diagram of the principle of dividing the second window of the present invention; Figure 3 The present invention is a structural schematic diagram of an emergency green channel information management system. DETAILED DESCRIPTION

[0018] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0019] It is understood that the terms "first", "second", etc. used in this application may be used herein to describe various elements, but unless otherwise specified, these elements are not limited by these terms. These terms are only used to distinguish a first element from another element. For example, without departing from the scope of this application, a first xx script may be referred to as a second xx script, and similarly, a second xx script may be referred to as a first xx script.

[0020] like Figure 1 As shown, the present invention provides an emergency green channel information management method, which specifically includes the following steps: S1: multiple first sensors are deployed in the hospital area, and the second sensor is worn on the patient's body. When the second sensor reaches the detection range of the first sensor, the first sensor generates stay information.

[0021] Specifically, the stay information includes the arrival time, departure time and location information of the second sensor.

[0022] The first sensor includes an RFID sensor or a Bluetooth gateway, and the second sensor includes a beacon or a Bluetooth sensor. When an emergency patient arrives at the hospital, the second sensor can be worn on the patient, or the second sensor can be directly installed on the transfer bed. The first sensor is deployed at the entrance of each department, ward, operating room, etc. The first sensor itself has a location tag, such as the emergency entrance. When the second sensor enters the monitoring range of the first sensor, the second sensor establishes a connection with the first sensor. When the second sensor leaves the monitoring range of the first sensor, the second sensor disconnects from the first sensor. The first sensor can generate stay information based on the entry and exit time of the second sensor, and send the stay information to the server for storage.

[0023] S2: Generate a processing node based on the stay information, and generate a first processing timeline of the patient based on the processing node.

[0024] For example, processing node 1 is at the emergency entrance from 8:00 to 8:02, processing node 2 is at triage registration from 8:05 to 8:07, and processing node 3 is at preliminary diagnosis in emergency room A from 8:08 to 8:12, then the above processing nodes are arranged in order as the first processing timeline.

[0025] S3: Record the first video based on the portable recorder, record the second video based on the fixed camera, split the first video and the second video according to the first processing timeline, and obtain the first sub-video and the second sub-video of the corresponding processing nodes.

[0026] S4: Analyze the first sub-video and the second sub-video to obtain operation process information and coordination process information respectively, and map the first sub-video, the second sub-video, the operation process information and the coordination process information to the first processing timeline respectively to obtain the second processing timeline.

[0027] During the emergency treatment process, medical staff will wear video recorders to record the first video of the emergency treatment process. The second video of the entire emergency area can be obtained through the surveillance cameras installed in the hospital. Then the first video and the second video are segmented according to the above processing nodes. If processing node 1 is 8:00-8:02, the segment between 8:00-8:02 in the first video is intercepted as the first sub-video, and the segment between 8:00-8:02 in the second video is intercepted as the second sub-video.

[0028] Then, image recognition is performed on the first sub-video and the second sub-video to obtain operation process information and coordination process information. The operation process information includes the medical staff's treatment operation process, such as wound treatment, injection, equipment use, etc., and the coordination process information includes communication, the arrival time of the consulting doctor, etc. Finally, the first sub-video, the second sub-video, the operation process information and the coordination process information are mapped to the processing node corresponding to the first processing timeline. When viewing the second processing timeline later, not only can the process nodes of the entire emergency treatment process be seen, but also the specific scene video of the corresponding node can be seen, which is convenient for review and quality inspection.

[0029] S5: Obtain the patient's electronic medical record, assign a disease manifestation to each second treatment timeline based on the electronic medical record, analyze the second treatment timelines for the same disease manifestation, and evaluate the treatment process.

[0030] The electronic medical record records the patient's personal information, diagnosis information, complications, etc. The above information is extracted from the patient's electronic medical record and used as the disease manifestation corresponding to the second processing timeline. After the disease manifestation is vectorized, the K-means or DBSCAN algorithm is used for clustering, so that the second processing timelines with the same disease manifestation are clustered into one category, that is, patients with the same physical condition and disease manifestation are clustered into one category. Finally, by comparing and analyzing each emergency treatment process under this category, the rescue process of each time is evaluated, the problem points are found, and the data basis is provided for the subsequent optimization of the emergency process.

[0031] The present invention realizes real-time monitoring of the patient's movement path by deploying multiple first sensors and second sensors worn by the patient in the hospital area, and then generates a processing timeline based on the patient's stay information at various locations, records video data through a portable recorder and a fixed camera, and identifies it to obtain operation process and coordination process information, which is mapped to the processing timeline to form a second processing timeline, which makes it easy to view the process nodes of the emergency treatment process. In addition, by obtaining the patient's electronic medical record and assigning disease manifestations, it is possible to perform cluster analysis on the processing timeline of the same disease manifestation, evaluate the efficiency and effectiveness of the treatment process, and thus optimize the emergency treatment process.

[0032] The present invention can not only intelligently collect the full-process treatment time of emergency green channel patients, but also comprehensively evaluate the efficiency and effect of each link in the treatment process through video analysis technology. This method makes up for the problem of insufficient real-time monitoring of the treatment process and video data analysis capabilities in the existing technology. Through the present invention, the management and evaluation of the emergency treatment process are more accurate and comprehensive.

[0033] In this embodiment, analyzing the first sub-video and the second sub-video includes the following steps: Define the first sub-video and the second sub-video as target videos, split the target video into multiple static frames, identify each static frame based on a deep neural network, obtain the operation event contained therein, merge the static frames of the same operation event to obtain the operation video of the corresponding operation event, determine the occurrence time period of the operation event according to the position of the operation video in the target video, divide the video length of the target video into multiple time windows based on the occurrence time period, and construct a first feature matrix with the time window as the row feature and the operation event as the column feature.

[0034] For ease of description, the first sub-video and the second sub-video are uniformly defined as target videos. When analyzing the target video, the target video is first split into multiple static frames, and then each static frame is identified based on a deep neural network. For the first sub-video that is strongly related to medical tools, this embodiment uses YOLOv8 to detect the tools in the static frames and obtain the operation events included therein, such as hand-held syringes, hand-held cotton swabs, etc. For the second sub-video that requires target tracking, this embodiment uses the DeepSORT target tracking algorithm to identify and track the target, and then the following operation events can be obtained, such as doctor A arriving at the target department for consultation at 18:00.

[0035] For the injection operation, it may last for 30 seconds, and there may be hundreds of frames of images in which the operation event is a hand-held syringe. Therefore, the static frames of the hand-held syringe are merged to obtain a complete operation video, and then the position of the operation video in the target video is further determined according to the time tags of each static frame. For example, if the entire video length of the target video is 30 minutes, and the operation event of the hand-held syringe appears in 10 minutes to 11 minutes, then the appearance time period of the operation video is 10 minutes to 11 minutes. After that, multiple time windows are generated according to the appearance time period of each operation video. For example, after division, there are time window 1, time window 2 and time window 3. Time window 1 includes multiple operation events such as a hand-held syringe and doctor A coming for consultation. The purpose of constructing the first feature matrix is ​​to facilitate subsequent computer processing.

[0036] A preset arrangement rule is set, and the first feature matrix is ​​rearranged based on the arrangement rule to obtain a second feature matrix. Feature recognition is performed on the second feature matrix based on a convolutional neural network to extract feature events from the second feature matrix, and the feature events include operation process information and coordination process information.

[0037] The arrangement rules are set manually. For example, the disinfection procedure includes handheld syringes, handheld cotton swabs, and handheld disinfectant. However, in the generated first feature matrix, the above elements may be far apart in the matrix. In order to facilitate the recognition of the convolutional neural network, their positions need to be adjusted. For example, arrangement rule 1 is to concentrate handheld cotton swabs, handheld disinfectant, and handheld syringes. If the current first feature matrix is , where elements 1, 2, and 3 represent handheld cotton swabs, handheld disinfectant, and handheld syringes, respectively; by translating the handheld syringes and handheld disinfectants upward, the handheld syringes, handheld cotton swabs, and handheld disinfectants are more concentrated, and the second characteristic matrix is ​​obtained as follows: , so that the convolution kernel of the convolutional neural network can be easily identified and extracted, that is, it is convenient for the convolutional neural network to extract co-occurrence features. The present invention uses a 3*3 convolution kernel for feature extraction and determines feature events. For example, by extracting the three features of handheld syringe, handheld cotton swab and handheld disinfectant in time windows 1, 2, and 3, it is determined that there is an injection event in time windows 1-3, that is, operation process information. The number of convolutional neural network convolution kernels, the size of each convolution kernel, the setting of the pooling layer, and the selection of the activation function can be determined by those skilled in the art based on actual experience and will not be described here.

[0038] In this embodiment, dividing the time window includes the following steps: An initial window duration is set, and the video duration is divided into multiple equally spaced first windows based on the initial window duration. The proportion of the time period of the operation event in each first window is obtained. If the proportion is greater than a first threshold, the operation event is defined to have occurred in the corresponding first window. The event overlap rate of adjacent first windows including the operation events is counted. If the event overlap rate is greater than a second threshold, the two first windows are merged into a second window. The density of the operation events included in the second window is counted. The second window with a density greater than a third threshold is located, and the window is split to obtain a third window and a fourth window. The unmerged first window, the undivided second window after merging, the third window, and the fourth window are defined as time windows.

[0039] For example, if the initial window duration is set to 10s and the video duration is 10 minutes, the video duration will be divided into 60 first time windows, and then the proportion of the time period of the operation event in each first window will be obtained. For example, the time span of operation event A is within the first windows 1-4, and it appears in the first window 1 for 5s, then the proportion in the first window 1 is 6S. If the proportion in the first windows 1-4 exceeds the first threshold, for example, the first threshold is 5s, then it is defined that operation event A appears in the first windows 1-4, and the judgment process for other operation events is the same.

[0040] Then, the event overlap rate of the operation events that appear in the adjacent first windows is counted. If the operation events A, B, C, and E appear in the first window 1 and the operation events A, B, C, D, and E appear in the first window 2, the event overlap rate is calculated using the first formula: ,in, is to compare the event overlap rate of the two first windows, X is the number of the same operation events that appear in the two first windows, To obtain the maximum number of operation events in the two first windows, are the number of operation events included in the two compared first windows respectively; the event overlap rate of the above example is 4 / 5*100%=80%, and the second threshold is set to 78%, then the first window 1 and the first window 2 are merged into the second window 1, and then the event overlap rate of the second window 1 and the first window 3 are compared. If the event overlap rate is greater than the second threshold, the first window 3 is merged into the second window 1, otherwise the second window 1 is retained, and the first window 3 and the first window 4 are compared, and the process is repeated until the last first window is processed.

[0041] After the merging is completed, the density of the operation events included in each second window is counted. If the density is greater than the third threshold, which is set to 5, it means that the second window includes too many operation events, which is not conducive to feature recognition by the deep learning model. Therefore, the second window needs to be segmented again. The specific segmentation method will be introduced later. After defining the segmentation of the second window, the third window and the fourth window are obtained. Finally, the currently existing first window, second window, third window and fourth window are defined as time windows.

[0042] In this embodiment, window segmentation of the second window includes the following steps: A first window is selected as a split window in the second window, and the second window is split into a first sub-window and a second sub-window based on the split window. A first number and a second number of operation events included in the first sub-window and the second sub-window are counted, and a difference value between the first number and the second number is calculated. Another first window is selected as a split window from the second window again and split again, and the difference value obtained after the split is calculated. This step is repeated until the first windows in the second window are traversed, and the first sub-window and the second sub-window corresponding to the minimum difference value are located as the segmentation result of the second window.

[0043] like Figure 2 As shown in FIG. 1 , since the second window is formed by merging multiple first windows, each second window will include multiple first windows. In particular, if the number of first windows included in the second window is less than 3, it will not be segmented. In actual use, if the number of first windows included is less than 3, the density will not be greater than the third threshold. When segmenting, select in reverse chronological order. Select Figure 2The first window 9 in the image is split with the first window 9 as the split window to obtain the first sub-window (first window 1-8) and the second sub-window (first window 9-10). The difference value is the absolute value of the difference between the operation events included in the two sub-windows. The first sub-window includes 12 operation events (AL) and the second sub-window includes 3 operation events (H, J, L). The difference value is 9.

[0044] When the first window 6 is used as the split window for segmentation, the first sub-window includes 6 operation events (AF) and the second sub-window includes 6 operation events (GL), the difference value is 0, which is the smallest, and the first sub-window (first window 1-5) and the second sub-window (first window 6-10) are used as the final segmentation result of the second window. The second window can be segmented reasonably by this method, so that the segmented windows can include multiple complete operation events.

[0045] In this embodiment, projecting the first sub-video and the second sub-video into the first processing timeline includes the following steps: Define the first sub-video and the second sub-video as target sub-videos, set a label in the target sub-video, the label corresponds to the processing node, set a first storage area and a second storage area, redundantly store the target sub-video in the first storage area and the second storage area, losslessly compress the target sub-video to obtain a first compressed file, and when checking the processing node, decompress the first compressed file in the first storage area to obtain the corresponding target sub-video.

[0046] For the sake of simplicity, the first sub-video and the second sub-video are collectively referred to as the target sub-video. When storing the target sub-video, a label corresponding to the processing node is marked in the target sub-video so that when accessing the processing node, the corresponding target sub-video can be located. The present embodiment also sets a first storage area and a second storage area, and stores the target sub-video in both storage areas, thereby achieving redundant storage and preventing accidental loss of the video. In order to save storage space, the target sub-video is losslessly compressed into a first compressed file for storage. When viewing, the first compressed file in the first storage area is decompressed to obtain the target sub-video. If there is no first compressed file in the first storage area, the target sub-video is searched from the second storage area.

[0047] If the existence time of the first compressed file reaches the first retention time, the first compressed file in the first storage area is deleted, and the second storage area performs lossy compression on the first compressed file to obtain a second compressed file. When checking the processing node, the second compressed file in the second storage area is decompressed to obtain the corresponding target sub-video. If the existence time of the second compressed file reaches the second retention time, the target sub-video is deleted. When checking the processing node, the corresponding operation process information or coordination process information is displayed.

[0048] The first retention time is set to 6 years. To further save storage space, after the existence time of the first compressed file exceeds 6 years, if it is in the first storage space, it will be deleted. If it is in the second storage space, it will be decompressed first to obtain the target sub-video, and then the target sub-video will be lossily compressed to obtain the second compressed file. In this way, the storage space of the first storage area and the second storage area can be further released. Finally, if the existence time of the second compressed file reaches the second retention time, the second retention time is 2 years, that is, the target sub-video has been stored for 8 years, then the second storage space will be deleted. However, the operation process information or coordination process information corresponding to the target sub-video is retained, so that when viewing the processing node, the processing status at that time can still be obtained according to the text information.

[0049] In this embodiment, a viewing time is also preset, and the first compressed file or the second compressed file is restored before the viewing time is reached.

[0050] The viewing time is manually set. For example, if the emergency treatment process of the previous week is set to be viewed at 15:00 every Wednesday, the system will decompress the first compressed file of the previous week in the first storage area at 14:40 every Wednesday, thereby further speeding up the video display speed. The specific viewing time setting rules are determined according to actual needs.

[0051] The evaluation and treatment process of this embodiment includes the following steps: The patient's disease manifestations are extracted based on the electronic medical record, and the disease manifestations are used as the treatment attributes of the second processing timeline. The second processing timeline is clustered based on the treatment attributes to obtain symptom clusters, and the corresponding historical data are obtained based on the symptom clusters. The historical data is analyzed to obtain the standard processing procedure. The standard processing procedure includes time standards, mandatory operations and standard collaboration modes. Based on the standard processing procedure, each second processing timeline is evaluated in terms of time efficiency, operation specifications and team collaboration. The treatment level is determined based on the evaluation results, and the treatment process is evaluated based on the treatment level.

[0052] Specifically, the disease manifestations also include the patient's main complaint and physical signs. The main complaint is such as squeezing pain behind the sternum with cold sweats, and the physical signs are systolic blood pressure>140 and heart rate>100. After that, the second processing timeline is clustered based on the treatment attributes. Assume that the second processing timeline of 3,000 chest pain patients in the database is clustered, and it is determined that the symptom cluster of the patient is myocardial infarction. For myocardial infarction, obtain 200 similar cases that have been successful in the past in the same cluster, and determine the time standard, required operations and standard collaboration mode for treatment by analyzing similar cases. For time standards, the median of historical data can be extracted, such as the standard time for electrocardiogram is 8 minutes. For required operations, the treatment operations that will be performed in 90% of cases are selected, such as electrocardiogram and myocardial infarction tests. For standard collaboration mode, the proportion of various cross-department handovers is obtained, such as 85% of direct handovers between the cardiology department and the catheterization room, so as to obtain the standard handover mode. For example, handovers between the emergency department, laboratory department, cardiology department, and catheterization room are required. In addition, the average response time of each department is obtained as the standard response time, such as the average response time of cardiology consultation is 15 minutes.

[0053] When conducting time efficiency evaluation, operation standard evaluation and team collaboration evaluation, the time efficiency score, operation standard score and team collaboration score are calculated based on the following second formula, third formula and fourth formula respectively. The second formula is ,in, Score for time efficiency, is the number of processing nodes included in the second processing timeline, is the deviation between the actual processing time of the nth processing node in the second processing timeline and the corresponding standard processing time, is the standard processing time of the nth processing node. If the total deviation value of each processing node is 30 minutes (numerator) and the total standard processing time is 120 minutes (denominator), the time efficiency score is 1-30 / 120=0.75.

[0054] The third formula is ,in, Rate the operating specifications. is the number of treatment operations actually performed in the second treatment timeline, is the number of standard treatment operations. For example, if 3 operations were actually performed and there were 6 standard treatment operations, the operation standard score would be 3 / 6=0.5.

[0055] The fourth formula is ,in, Score teamwork. is the number of actual department handovers in the second processing timeline, The number of handovers in standard handover mode, The number of departments that conducted the consultation. is the deviation between the actual response time of the i-th department and the corresponding standard response time, is the standard response time of the i-th department. If the number of actual department handovers is 3, the number of handovers in the standard handover mode is 2, the total deviation value of the response of each department is 7 minutes (numerator), and the total standard response time is 30 minutes (denominator), then the teamwork score is 0.635.

[0056] Finally, the time efficiency score, operation standard score and teamwork score are weighted and summed to obtain a comprehensive score. Then, multiple numerical ranges are set, and each numerical range corresponds to a level of treatment, such as 0.6-0.7 points corresponding to level D, thus completing the evaluation of the treatment process.

[0057] like Figure 3 As shown, the present invention also provides an emergency green channel information management system, which is used to implement the above-mentioned emergency green channel information management method, and the system includes: The sensing module includes a plurality of first sensors arranged in the hospital area, a second sensor worn on the patient's body, and the first sensor generates stay information when the second sensor reaches the detection range of the first sensor; A first analysis module generates a processing node based on the stay information, and generates a first processing timeline of the patient based on the processing node; A second analysis module, based on recording the first video with a portable recorder and recording the second video with a fixed camera, the second analysis module divides the first video and the second video according to the first processing timeline to obtain a first sub-video and a second sub-video of corresponding processing nodes, analyzes the first sub-video and the second sub-video to obtain operation process information and coordination process information respectively, and maps the first sub-video, the second sub-video, the operation process information and the coordination process information to the first processing timeline respectively to obtain a second processing timeline; The evaluation module is used to obtain the patient's electronic medical record, assign disease manifestations to each second treatment timeline based on the electronic medical record, analyze the second treatment timelines of the same disease manifestations, and evaluate the treatment process.

[0058] The present invention also discloses a computer storage medium, wherein the computer storage medium stores program instructions, wherein when the program instructions are executed, the device where the computer storage medium is located is controlled to execute the above-mentioned method.

[0059] The technical features of the above-mentioned embodiments can be combined arbitrarily. In order to make the description concise, not all possible combinations of the technical features in the above-mentioned embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0060] The above embodiments only express several implementation methods of the present invention, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the patent of the present invention. It should be pointed out that, for those of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present invention, and these all belong to the protection scope of the present invention. Therefore, the protection scope of the patent of the present invention shall be subject to the attached claims.

[0061] The above are only preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. An emergency green channel information management method, characterized in that: include: A plurality of first sensors are arranged in the hospital area, a second sensor is worn on the patient's body, and when the second sensor reaches the detection range of the first sensor, the first sensor generates stay information; generating a processing node based on the stay information, and generating a first processing timeline of the patient based on the processing node; Recording a first video based on a portable recorder and recording a second video based on a fixed camera, segmenting the first video and the second video according to the first processing timeline to obtain a first sub-video and a second sub-video corresponding to the processing node; Analyze the first sub-video and the second sub-video to obtain operation process information and coordination process information respectively, and map the first sub-video, the second sub-video, the operation process information and the coordination process information to the first processing timeline respectively to obtain a second processing timeline; The electronic medical record of the patient is obtained, a disease manifestation is assigned to each second treatment timeline based on the electronic medical record, and the second treatment timelines of the same disease manifestation are analyzed to evaluate the treatment process.

2. The method according to claim 1, characterized in that Analyzing the first sub-video and the second sub-video includes the following steps: Define the first sub-video and the second sub-video as target videos, split the target video into multiple static frames, identify each of the static frames based on a deep neural network, obtain an operation event contained therein, merge the static frames of the same operation event to obtain an operation video corresponding to the operation event, determine the occurrence time period of the operation event according to the position of the operation video in the target video, divide the video length of the target video into multiple time windows based on the occurrence time period, and construct a first feature matrix with the time windows as row features and the operation events as column features; Preset arrangement rules, rearrange the first feature matrix based on the arrangement rules to obtain a second feature matrix, perform feature recognition on the second feature matrix based on a convolutional neural network to extract feature events from the second feature matrix, and the feature events include the operation process information and the coordination process information.

3. The method according to claim 2, characterized in that Dividing the time window comprises the following steps: An initial window duration is set, and the video duration is divided into a plurality of equally spaced first windows based on the initial window duration. The proportion of the occurrence time period of the operation event in each of the first windows is obtained. If the proportion is greater than a first threshold, the operation event is defined to have occurred in the corresponding first window. The event overlap rate of adjacent first windows including the operation events is counted. If the event overlap rate is greater than a second threshold, two first windows are merged into a second window. The density of the operation events included in the second window is counted. The second window with the density greater than a third threshold is located, and the second window is split to obtain a third window and a fourth window. The unmerged first window, the undivided second window after merging, the third window, and the fourth window are defined as the time window.

4. The method according to claim 3, characterized in that The step of performing window segmentation on the second window comprises the following steps: The first window is selected as the split window in the second window, and the second window is split into a first sub-window and a second sub-window based on the split window, and a first number and a second number of the operation events included in the first sub-window and the second sub-window are counted, and a difference value between the first number and the second number is calculated, and another first window is selected as the split window from the second window again and split again, and the difference value obtained after the split is calculated, and this step is repeated until the first windows in the second window are traversed, and the first sub-window and the second sub-window corresponding to the minimum difference value are located as the segmentation result of the second window.

5. The method according to claim 1, characterized in that: Projecting the first sub-video and the second sub-video into the first processing timeline comprises the following steps: Define the first sub-video and the second sub-video as target sub-videos, set a tag in the target sub-video, the tag corresponds to the processing node, set a first storage area and a second storage area, redundantly store the target sub-video in the first storage area and the second storage area, losslessly compress the target sub-video to obtain a first compressed file, and when checking the processing node, decompress the first compressed file in the first storage area to obtain the corresponding target sub-video; If the existence time of the first compressed file reaches the first retention time, the first compressed file in the first storage area is deleted, and the second storage area performs lossy compression on the first compressed file to obtain a second compressed file. When viewing the processing node, the second compressed file in the second storage area is decompressed to obtain the corresponding target sub-video. If the existence time of the second compressed file reaches the second retention time, the target sub-video is deleted. When viewing the processing node, the corresponding operation process information or the coordination process information is displayed.

6. The method according to claim 5, characterized in that A viewing time is preset, and the first compressed file or the second compressed file is restored before the viewing time is reached.

7. The method according to claim 4, characterized in that The evaluation process includes the following steps: The disease manifestations of the patient are extracted based on the electronic medical record, and the disease manifestations are used as the treatment attributes of the second processing timeline. The second processing timeline is clustered based on the treatment attributes to obtain symptom clusters, and corresponding historical data are acquired based on the symptom clusters. The historical data are analyzed to obtain a standard processing procedure, and the standard processing procedure includes time standards, mandatory operations, and standard collaboration modes. Time efficiency, operation specification, and team collaboration are evaluated for each second processing timeline based on the standard processing procedure. The treatment level is determined based on the evaluation results, and the treatment process is evaluated based on the treatment level.

8. The method according to claim 5, characterized in that The stay information includes arrival time, departure time and location information of the second sensor.

9. An emergency green channel information management system, used to implement the method according to any one of claims 1 to 8, characterized in that: include, The sensing module includes a plurality of first sensors arranged in a hospital area, a second sensor worn on a patient's body, and when the second sensor reaches a detection range of the first sensor, the first sensor generates stay information; a first analysis module, generating a processing node based on the stay information, and generating a first processing timeline of the patient based on the processing node; A second analysis module, based on recording the first video with a portable recorder and recording the second video with a fixed camera, the second analysis module divides the first video and the second video according to the first processing timeline to obtain the first sub-video and the second sub-video corresponding to the processing node, analyzes the first sub-video and the second sub-video to obtain operation process information and coordination process information respectively, and maps the first sub-video, the second sub-video, the operation process information and the coordination process information to the first processing timeline respectively to obtain a second processing timeline; An evaluation module is used to obtain the patient's electronic medical record, assign a disease manifestation to each second processing timeline based on the electronic medical record, analyze the second processing timelines with the same disease manifestation, and evaluate the treatment process.

10. A computer storage medium, characterized in that: The computer storage medium stores program instructions, wherein when the program instructions are executed, the device where the computer storage medium is located is controlled to execute the method according to any one of claims 1 to 8.

Images