Unplanned extubation pre-identification and extubation alarm method and system based on marker

Abstract

The invention provides an unplanned extubation identification method based on a marker, which comprises the following steps: inputting an image containing the marker into a marker identification module for attitude judgment to obtain attitude information, and intercepting the marker to obtain an intercepting frame containing the marker; inputting the interception box into a semantic segmentation module for supervised object segmentation to obtain a supervised object and a moving object; and judging the moving state of the moving body in the supervised object according to the moving body, the interception frame and / or the spatial position of the supervised object, and sending out non-planned extubation supervision alarm information. Alarm information and early warning information are given to different behavior models through classification of postures and action behaviors of patients, and therefore the purpose of accurately preventing unplanned extubation is achieved.

Description

Technical Field

[0001] The invention belongs to the technical field of catheter removal prevention, and in particular relates to an unplanned catheter removal identification method based on a marker. Background Art

[0002] In order to deliver necessary water and food to patients who cannot swallow under special circumstances, a gastric tube is placed into the esophagus through the mouth or nostrils. If the trachea is pulled out unplanned, the patient's mucosa may be damaged and the esophagus may be stimulated to cause strong physiological discomfort. Some patients with respiratory diseases cannot maintain adequate ventilation or oxygenation and cannot have enough gas exchange, so they usually use a ventilator to assist breathing. If the trachea is pulled out unplanned, the patient may suffer from respiratory arrest and die. Improper operation when the patient changes position or changes clothes may cause excessive pulling and easy dislocation of the tube. When the patient is uncomfortable or unconscious, he or she may pull out the catheter by hand.

[0003] Such actions often take a short time and are difficult to stop through real-time manual monitoring. The Chinese patent document, application number: 202410510955.5, discloses a method for monitoring and dynamically capturing patient extubation actions based on deep learning. It does not solve the technical problem of how to stop the patient's extubation action in advance. Summary of the invention

[0004] In view of the above technical problems, the present invention achieves the purpose of early warning and alarm of unplanned extubation by establishing different models corresponding to different patient postures and movements.

[0005] The first aspect of the present invention provides a marker-based unplanned tube removal pre-identification and removal alarm method, comprising: inputting an image containing a passive marker into a passive marker recognition module to perform posture judgment to obtain posture information, and performing passive marker interception to obtain an interception frame containing the passive marker;

[0006] Inputting the image containing the passive marker into a semantic segmentation module to perform supervised object segmentation to obtain supervised objects, active objects and monitoring areas;

[0007] The activity state of the moving body on the monitoring object is judged according to the moving body, the monitoring area and / or the spatial position of the monitoring object, and an unplanned extubation monitoring alarm message is issued.

[0008] The second aspect of the present invention provides: a marker-based unplanned tube removal pre-identification and removal alarm system, the system comprising at least one image acquisition end, at least one processor; and a memory storing instructions, which, when executed by at least one processor, implements the steps of the above method.

[0009] The third aspect of the present invention provides: a computer-readable storage medium for a marker-based unplanned tube extubation pre-identification and extubation alarm method, on which a computer program / instruction is stored, characterized in that the computer program / instruction implements the steps of the above method when executed by a processor.

[0010] The fourth aspect of the present invention provides: a computer program product for a marker-based unplanned tube extubation pre-identification and extubation alarm method, comprising a computer program / instruction, characterized in that the computer program / instruction implements the steps of the above method when executed by a processor.

[0011] The fifth aspect of the present invention provides: a network of unplanned tube removal pre-identification and removal alarm methods based on markers, characterized by a first server and a second server, wherein the first server or the second server can execute the steps according to the above method.

[0012] The beneficial effect of the present invention is that by classifying the patient's posture and action behavior, alarm information and early warning information are given to different behavior models, thereby achieving the purpose of accurately preventing unplanned extubation. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] Figure 1 Flow chart of alarm identification according to an embodiment of the present invention;

[0014] Figure 2 Flowchart of restraining patient alarms according to an embodiment of the present invention;

[0015] Figure 3 Non-constrained patient warning flow chart of an embodiment of the present invention;

[0016] Figure 4 Flow chart of the identification and alarm of unrestrained patients and restrained patients according to an embodiment of the present invention;

[0017] Figure 5 Flow chart of obtaining constraint status information according to an embodiment of the present invention;

[0018] Figure 6 Flow chart of posture information acquisition according to an embodiment of the present invention;

[0019] Figure 7 The interception frame acquisition flow chart of the embodiment of the present invention;

[0020] Figure 8 Flow chart of posture information classification according to an embodiment of the present invention;

[0021] Fig. 9 Flowchart of the supervision task distribution according to the embodiment of the present invention;

[0022] Fig.10 Schematic diagram of the system structure of the embodiment of the present invention Figure 1 ;

[0023] Fig.11 Flow chart of posture information classification and recognition according to an embodiment of the present invention;

[0024] Fig.12 Schematic diagram of the system structure of the embodiment of the present invention Figure 2 ;

[0025] Fig.13 Schematic diagram of the system structure of the embodiment of the present invention Figure 3 ;

[0026] Fig.14 Schematic diagram of the system structure of the embodiment of the present invention Figure 4 ;

[0027] Fig.15 Schematic diagram of the system structure of the embodiment of the present invention Figure 5 ;

[0028] Fig.16 Schematic diagram of the algorithm structure of an embodiment of the present invention;

[0029] Fig.17 Diagram of the early warning interface of the nurse station monitoring terminal intention in the embodiment of the present invention;

[0030] Fig.18 Diagram of the alarm interface of the extubation completion stage of the monitoring terminal at the nurse station in the embodiment of the present invention;

[0031] Fig.19 Patient facial recognition in an embodiment of the present invention Figure 1 ;

[0032] Fig. 20 Patient facial recognition in an embodiment of the present invention Figure 2 ;

[0033] Fig.21 Diagram of a bed interface for turning off the unplanned extubation recognition function with one click at a nurse station monitoring terminal according to an embodiment of the present invention. DETAILED DESCRIPTION

[0034] The following examples further illustrate the content of the present invention, but should not be construed as limiting the present invention. Without departing from the spirit and substance of the present invention, modifications or substitutions made to the methods, steps or conditions of the present invention all fall within the scope of the present invention.

[0035] It should also be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the presence of other identical elements in the process, method, article or device including the elements.

[0036] In some embodiments of the present invention, Figure 1-21 As shown: A marker-based unplanned tube removal pre-identification and removal alarm method, characterized by comprising:

[0037] The image containing the passive marker is input into the passive marker recognition module for posture judgment to obtain posture information, and the passive marker is intercepted to obtain a capture frame containing the passive marker. The capture frame is as follows: Fig.19 , 20 The blue box in

[0038] The image containing the passive marker is input into the semantic segmentation module for supervised object segmentation to obtain supervised objects, active objects and monitoring areas; wherein the supervised objects are, for example, identified pipelines such as endotracheal tubes, etc.; active objects are, for example, hands, gripping tools, etc.; monitoring areas are, for example: Fig.19 , 20 The yellow box containing the face;

[0039] The activity state of the moving body on the monitoring object is judged according to the moving body, the monitoring area and / or the spatial position of the monitoring object, and an unplanned extubation monitoring alarm message is issued.

[0040] In these embodiments:

[0041] This solution is mainly aimed at the unplanned extubation of patients in the ICU ward of the hospital. Under the requirements of pre-identification and alarm of unplanned extubation, the core recognition objects are human faces, tubes, and human hands. Human faces have already had good recognition accuracy in the existing technical framework. Therefore, the recognition accuracy of trachea and human hands can be further improved to achieve accurate recognition of unplanned extubation actions. Among them, the monitoring object can be an endotracheal tube, a gastric tube, etc.; the passive marker can be a tape of a special material, an easily recognizable color, etc.; the interception frame includes the recognition frame of the face, such as Fig.19 In the blue box, the warning information includes early warning information and alarm information.

[0042] In these embodiments:

[0043] When the patient is uncomfortable or unconscious, he or she pulls out the catheter by hand. Monitoring and warning of unplanned extubation are key issues in clinical risk management. In this embodiment, unplanned extubation is mainly divided into the following scenarios:

[0044] 1. Provide timely warnings for intubated patients whose hands enter their faces;

[0045] Non-binding patient conditions:

[0046] By using the face and hands within the facial recognition area, when it is detected that a hand enters the patient's face and remains in the area for a period of time, it is considered that there is a risk of extubation and an early warning is issued immediately.

[0047] Patient conditions with constraints:

[0048] Hand intrusion warning: timely alarm will be issued for the behavior of intubated patients' hands entering their faces, and the alarm information will be pushed to the nurse station monitoring terminal and the nurse mobile terminal in a timely manner.

[0049] Lying position identification: supine position, left lateral position, right lateral position, prone position.

[0050] Identify the lines: endotracheal tube, high-flow oxygen catheter, gastric tube.

[0051] Both day and night scenes can be recognized. The camera (image acquisition end) is installed just above the end of the bed, with the camera position within 2.5 meters above the bed surface. By adjusting the focal length, clear faces can be captured.

[0052] In the recognition scenario, the optimal recognition conditions are: the face is not severely obstructed, the pipes are not obstructed, and the person is lying flat.

[0053] Medical staff issue a request for a testing bed and start bed detection. If a hand is detected entering the facial area, an alert will be issued.

[0054] 2. Alarm when an intubated patient removes the tube from his nose or mouth;

[0055] Unplanned extubation: Alarm for extubation from the nose or mouth. If the hand is detected on or near the tube in the face area and the tube is out of the functional area, it is considered extubation and an alarm message is sent to the system.

[0056] Both daytime and nighttime scenes can be recognized. The camera (image acquisition end) is installed just above the end of the bed, and the camera position is within 2.5 meters above the bed surface. Through focal length adjustment, clear faces can be captured. The collected images can clearly identify faces, pipes, and tape colors.

[0057] In the recognition scenario, the optimal recognition conditions are: the face is not severely blocked, the pipeline is not blocked, and the person is lying flat. When the unplanned extubation recognition alarm function is used, the medical staff sends a bed detection, turns on the bed detection, and recognizes that the hand is close to the pipeline and the pipeline is out of the functional area to alarm.

[0058] 3. Receive early warning information and close monitoring of non-extubation beds;

[0059] 4. Alarm information is pushed to the nurse terminal in time;

[0060] The system mainly uses video analysis to identify the patient's extubation status. Different analysis rules are preset in the camera scene. Once the target violates the predefined analysis rules in the scene, the system will trigger the pre-set linkage rules. This effectively realizes the monitoring of unplanned extubation of patients.

[0061] In some embodiments of the present invention, Figure 2 , 4 As shown, the difference from the above embodiment is that: the step of judging the activity state of the moving body on the monitoring object according to the moving body, the interception frame and / or the spatial position of the monitoring object and issuing the unplanned extubation monitoring alarm information includes:

[0062] The step of judging the activity state of the moving object on the monitoring object according to the moving object, the interception frame and / or the spatial position of the monitoring object and issuing the unplanned extubation monitoring alarm information comprises:

[0063] Obtain constraint status information of the supervised object;

[0064] Determine whether the supervised object is restrained. If the supervised object is restrained, determine whether the active object exists in the monitoring area, and if so, issue an unplanned extubation alarm message;

[0065] If the supervised object is not restrained, it is determined whether the moving body exists in the monitoring area and whether the staying time of the moving body in the monitoring area exceeds the warning threshold. If so, an unplanned extubation warning message is issued.

[0066] In some embodiments of the present invention, Figure 3 , 4 As shown, the difference from the above embodiment is that:

[0067] Obtain constraint status information of the supervised object;

[0068] Determine whether the supervised object is restrained. If the supervised object is not restrained, determine whether the active object exists in the monitoring area and whether the stay time of the active object in the monitoring area exceeds the warning threshold. If it exceeds, issue an unplanned extubation warning message.

[0069] In these embodiments: Passive marker recognition basis: In order to improve the recognition accuracy of the target object, the distinction between the target and the non-target object is increased by increasing the color contrast, the screen ratio and other factors to reduce the recognition difficulty. Conventional medical equipment and human body sizes are relatively fixed, while in the ward, the colors of the bed sheets, patient gowns and other environments are mainly white and blue, and the colors are relatively single. Objects with obviously different colors can be selected as markers for auxiliary recognition and judgment.

[0070] Change the surface color of the endotracheal tube and other pipelines, the auxiliary fixation belt, the color of the fixing tape, etc. In this embodiment, a colored tape with a clear color contrast with bed-related items is selected, such as purple and orange, so that image recognition is easier to identify and analyze, thereby improving the recognition accuracy of the pipeline.

[0071] The basis for distinguishing between unrestrained and restrained patients: The installation of endotracheal intubation, gastric intubation, and other tubes that play an important role in maintaining the patient's vital signs needs to be strictly monitored, and different early warning conditions and alarm conditions must be set. In order to avoid extubation in an unconscious state, some patients will be specially monitored, and the patient's behavioral state will also be subject to a certain degree of restraint, such as restraining their hands with sleeves or straps to limit their ability to move.

[0072] 1. For non-restrained patients, pre-identification - judged by whether the patient's arm marker stays in the facial recognition frame for a period of time exceeding the threshold. In this embodiment, the threshold is 0.5s to 5s, specifically 0.5s, 1s, 1.5s, 2s, 2.5s, 3s, 3.5s, 4.0s, 4.5s, 5s; preferably 1s. The arm marker is a passive marker. Passive marker: a colored tape with a significant color contrast to bed-related items, such as purple and orange, which is easier to identify and analyze with image recognition.

[0073] 2. For restrained patients, pre-identification - enter the facial recognition situation through the patient's arm marker and immediately alarm.

[0074] 3. Post-event alarm - For ordinary dangerous level intubated patients (generally unrestrained patients), the passive marker of the tube on the patient's face is identified by whether it has left the face to determine whether the tube has fallen off; for severely dangerous level intubated patients (generally restrained patients), in addition to the alarm of whether the passive marker on the patient's face exists, active parent-child markers can also be installed to improve the accuracy of the alarm.

[0075] Among them, the active marker is driven by a button battery and has a magnetic unit. One part of the module can be placed on the pipeline, and the other part can be pasted and fixed on the patient's face. After the pipeline is displaced, the two can issue a separation alarm.

[0076] In some embodiments of the present invention, Figure 4, 19 , as shown in 20, which is different from the above embodiment in that:

[0077] The intercepted frame is input into the semantic segmentation module to segment the supervised object, and the supervised object, the active object and the monitoring area are obtained, including the extubation warning step:

[0078] Inputting the captured frame into a semantic segmentation module for supervised object segmentation;

[0079] Obtain the object to be monitored, the moving object and the monitoring area, and if the object to be monitored is not in the monitoring area, issue an extubation alarm message.

[0080] In some embodiments of the present invention, Figure 5 , 17 , as shown in FIG. 18, the difference from the above embodiment is that: the step of obtaining the constraint status information of the supervised object includes:

[0081] Associating the supervised object information according to the image acquisition terminal information;

[0082] The constraint state information is associated with the supervision object information to obtain the constraint state information.

[0083] In these embodiments:

[0084] In this embodiment, non-restrained patients have corresponding case records and identifications at the monitoring end, and it can be determined whether the patient is a restrained patient through their case records and identifications. The bed number corresponding to the patient and the image acquisition device corresponding to the bed number, such as a high-definition camera, have corresponding device numbers. Based on this, a mapping of the patient, patient restraint status, bed number, camera, and constraint status information of the camera captured image, or an information management table, is established to form a judgment on the patient's restraint status.

[0085] At the same time, the identification can be generated on the monitoring end, such as selecting the restraint status on the operation interface, or designing a restraint reminder button on the patient-related equipment, such as ventilators, gastric tube equipment, etc., and a medical staff presses the reminder button, which is synchronized to the monitoring end to form a restrained patient identification. The patient's restraint status can be quickly obtained, and it is also convenient for further supervision of the patient's extubation.

[0086] In some embodiments of the present invention, Figure 1-4 , 6, and 16, which are different from the above embodiments in that: the step of inputting the image containing the marker into the marker recognition module for posture judgment to obtain the posture information includes:

[0087] Inputting the image containing the marker into the marker recognition module, performing a plurality of marker point matching, and obtaining a plurality of marker points;

[0088] The posture of the supervised object is identified according to the distribution information of the identification points to obtain the posture information of the supervised object.

[0089] In some embodiments of the present invention, Figure 7 , 16 As shown, the difference from the above embodiment is that: the step of performing marker interception and obtaining a interception frame containing the passive marker includes:

[0090] Inputting the image containing the passive marker into a marker recognition module;

[0091] Perform semantic segmentation to obtain the clipped box containing the passive marker.

[0092] In some embodiments of the present invention, Figure 8 , 10 As shown, the difference from the above embodiment is that:

[0093] The step of identifying the posture of the supervised object according to the distribution information of the identification points and obtaining the posture information of the supervised object also includes:

[0094] Obtaining posture information of the supervised object, classifying the posture information of the supervised object, obtaining posture information classification information and sending the posture information classification information to the monitoring end;

[0095] If the posture information classification information meets the recognition standard (lying upright), passive marker interception is performed.

[0096] In these embodiments:

[0097] Take yolov8 as an example: use yolov8's human key point model pose and yolov8's semantic segmentation model segment; the pose model can find 17 key points of the human body; the segmentation model is responsible for detecting the tube.

[0098] Among them: the key point model pose is a functional module in the YOLOv8 series specifically used for human posture estimation and key point detection. It can be used for real-time human posture estimation, action recognition, and behavior analysis. Correspondingly, you can also choose the OpenPose multi-person key point detection system, MediaPipe Pose detection module, HRNet human posture estimation and key point detection network, and SimpleBaseline human key point detection model combination and replacement.

[0099] The semantic segmentation model segment is a functional module that focuses on image semantic segmentation. It can segment an image into multiple regions and assign semantic labels to each region, thereby achieving accurate recognition and distinction of different objects in the image. Correspondingly, you can also choose U-Net image segmentation model, DeepLab series image segmentation model, Mask R-CNN image segmentation model, PSPNet (Pyramid Scene Parsing Network) image segmentation model as a combination and replacement of segment model.

[0100] The recognition results are shown in the figure Fig.19 and 20 As shown: Use the pose model for detection, the red frame is the pose model result; judge the lying posture based on the facial key point results of the pose; generate a yellow frame based on the facial key points to determine the person's face.

[0101] In some embodiments of the present invention, Fig. 9 , 11 , 12, 13, 14, 15, and 21, which are different from the above embodiments in that:

[0102] The identification method further comprises the step of issuing a supervision task:

[0103] The monitoring terminal selects the supervised object and the non-supervised object and issues the supervision task;

[0104] The corresponding image acquisition end corresponding to the supervised object transmits the image containing the passive marker to the processor.

[0105] The alarm statistics results are shown in Table 1 Alarm Statistics Table

[0106] Table 1

[0107]

[0108]

[0109] In some embodiments of the present invention, Fig.10 , 11 As shown in 12, 13, 14 and 15, a marker-based unplanned extubation pre-identification and extubation alarm system comprises at least one image acquisition terminal, a monitoring terminal, at least one processor and a memory; the image acquisition terminal communicates with the processor, and the monitoring terminal communicates with the processor; the memory stores instructions, and when the instructions are executed by at least one processor, the steps of the above method are implemented.

[0110] In some embodiments of the present invention, Figure 1As shown, the difference from the above embodiment is: a computer-readable storage medium for a marker-based unplanned extubation identification method, on which a computer program / instruction is stored, characterized in that the computer program / instruction implements the steps of the above method when executed by a processor.

[0111] In some embodiments of the present invention, Figure 1 As shown, the difference from the above embodiment is that: a computer program product of a marker-based unplanned extubation identification method includes a computer program / instruction, characterized in that the computer program / instruction implements the steps of the above method when executed by a processor.

[0112] In some embodiments of the present invention, Figure 1 As shown, the difference from the above embodiment is: a marker-based unplanned extubation identification network, characterized by a first server and a second server, wherein the first server or the second server can execute the steps according to the above method.

[0113] The image acquisition end collects images of the patient's face and adjacent areas, and transmits the images to the server through the network. The server uses an AI algorithm to intelligently identify the image content, stores the identification results in the database, and transmits the identification results to the monitoring end through the database, the integration platform, and the network. The image acquisition end is preferably a network camera, each camera is uniformly facing above the bed, and the image is adjusted through the camera zoom function to bring the patient's upper body into the picture.

[0114] The processor may be a server, and the server may be an ECS server, which acquires and configures computing resources such as CPU, memory, disk, and network according to the computing requirements generated by the scene where the identification device is located.

[0115] The processor may include an intelligent recognition unit, which is pre-installed with an image recognition program. It can determine the patient's posture based on the patient's face and adjacent areas, such as side-lying, supine, etc., and can also define a facial recognition frame in the image; further, it can identify the tubes worn by the patient and unplanned extubation actions through semantic segmentation.

[0116] The memory may be a database, which stores the image generated after the intelligent recognition unit recognizes it, and inserts, updates, deletes and retrieves the generated image data according to the recognition information.

[0117] The network structure may include an integration platform, which is a software system or infrastructure for realizing the interconnection, integration and collaboration between two or more different applications, business processes and data sources of the intelligent identification unit and the monitoring terminal.

[0118] The network can be a wireless network or a combination of a wired network and a wireless network. The wireless network can be a Bluetooth module, a WIFI module, a GSM module, a GPRS module, a CDMA module, a CDMA2000 module, a WCDMA module, a TD-SCDMA module, a Zigbee module or a LoRa module. By adopting a variety of wireless communication methods, it can not only increase the flexibility of the wireless communication method, but also meet the needs of different users and different occasions. In particular, when the LoRa module is adopted, its communication distance is longer and the communication performance is more stable, which is suitable for occasions with higher communication quality requirements.

[0119] The monitoring end preferably displays and conveys the early warning and alarm information of unplanned extubation to medical staff through an electronic screen.

[0120] The first monitoring terminal is the nurse's mobile terminal, which sends out warning and alarm signals through watches, mobile phones and other devices with display screens, sound prompt speakers, vibration prompters and other warning and alarm information. It can be carried by medical staff to monitor or receive warning and alarm information at any time.

[0121] The second monitoring end is the nurse station monitoring end, which monitors and receives early warning and alarm information from medical staff through desktop computers with display screens, sound prompt speakers, and other external display devices, as well as touch screens, keyboards, mice and other interactive devices that can perform human-computer interaction. Or start and temporarily monitor the video of a patient in a bed.

[0122] The technical effects of the present invention are multifaceted. By classifying the patient's posture and action behaviors, alarm information and early warning information are given to different behavior models, thereby achieving the purpose of accurately preventing unplanned extubation.

[0123] Finally, it should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the presence of other identical elements in the process, method, article or device including the elements.

[0124] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referenced to each other.

[0125] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present invention. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments described herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.

[0126] Although the present invention has been described in detail above by general description, specific implementation methods and experiments, it is obvious to those skilled in the art that some modifications or improvements can be made to the present invention. Therefore, these modifications or improvements made without departing from the spirit of the present invention all belong to the scope of protection claimed by the present invention.

Claims

1. A marker-based unplanned tube removal pre-identification and removal alarm method, characterized in that: include: Inputting the image containing the passive marker into the passive marker recognition module to perform posture judgment to obtain posture information, and performing passive marker interception to obtain an interception frame containing the passive marker; Inputting the image containing the passive marker into a semantic segmentation module to perform supervised object segmentation to obtain supervised objects, active objects and monitoring areas; The activity state of the moving body on the monitoring object is judged according to the moving body, the monitoring area and / or the spatial position of the monitoring object, and an unplanned extubation monitoring alarm message is issued.

2. The marker-based unplanned tube removal pre-identification and removal alarm method according to claim 1, characterized in that: The step of judging the activity state of the moving object on the monitoring object according to the moving object, the interception frame and / or the spatial position of the monitoring object and issuing the unplanned extubation monitoring alarm information comprises: Obtain constraint status information of the supervised object; Determine whether the supervised object is restrained. If the supervised object is restrained, determine whether the active object exists in the monitoring area, and if so, issue an unplanned extubation alarm message; If the supervised object is not restrained, it is determined whether the active body exists in the monitoring area and whether the stay time of the active body in the monitoring area exceeds the warning threshold. If so, an unplanned extubation warning message is issued.

3. The marker-based unplanned tube removal pre-identification and removal alarm method according to claim 1, characterized in that: The step of judging the activity state of the moving object on the monitoring object according to the moving object, the interception frame and / or the spatial position of the monitoring object and issuing the unplanned extubation monitoring alarm information comprises: Obtain constraint status information of the supervised object; Determine whether the supervised object is restrained. If the supervised object is not restrained, determine whether the active object exists in the monitoring area and whether the stay time of the active object in the monitoring area exceeds the warning threshold. If it exceeds, issue an unplanned extubation warning message.

4. The marker-based unplanned tube removal pre-identification and removal alarm method according to claim 1, characterized in that: The intercepted frame is input into the semantic segmentation module to segment the supervised object, and the supervised object, the active object and the monitoring area are obtained, including the extubation warning step: Inputting the captured frame into a semantic segmentation module for supervised object segmentation; Obtain the object to be monitored, the moving object and the monitoring area, and if the object to be monitored is not in the monitoring area, issue an extubation alarm message.

5. The marker-based unplanned tube removal pre-identification and removal alarm method according to claim 2, characterized in that: The step of obtaining constraint status information of the supervised object comprises: Associating the supervised object information according to the image acquisition terminal information; The constraint state information is associated with the supervision object information to obtain the constraint state information.

6. The marker-based unplanned tube removal pre-identification and removal alarm method according to claim 1, characterized in that: The steps of inputting the image containing the marker into the marker recognition module for posture judgment to obtain posture information include: Inputting the image containing the marker into the marker recognition module, performing a plurality of marker point matching, and obtaining a plurality of marker points; The posture of the supervised object is identified according to the distribution information of the identification points to obtain the posture information of the supervised object.

7. The marker-based unplanned tube removal pre-identification and removal alarm method according to claim 1, characterized in that: The steps of performing marker interception and obtaining an interception frame including the passive marker include: Inputting the image containing the passive marker into a marker recognition module; Perform semantic segmentation to obtain the clipped box containing the passive marker.

8. The marker-based unplanned tube removal pre-identification and removal alarm method according to claim 6, characterized in that: The step of identifying the posture of the supervised object according to the distribution information of the identification points and obtaining the posture information of the supervised object also includes: Obtaining posture information of the supervised object, classifying the posture information of the supervised object, obtaining posture information classification information and sending the posture information classification information to the monitoring end; If the posture information classification information meets the recognition standard, passive marker interception is performed.

9. The marker-based unplanned tube removal pre-identification and removal alarm method according to claim 8, characterized in that: The identification method further comprises the step of issuing a supervision task: The monitoring terminal selects the supervised object and the non-supervised object and issues the supervision task; The corresponding image acquisition end corresponding to the supervised object transmits the image containing the passive marker to the processor.

10. A marker-based unplanned extubation pre-identification and extubation alarm system, the system comprising at least one image acquisition terminal, a monitoring terminal, at least one processor and a memory; the image acquisition terminal communicates with the processor, and the monitoring terminal communicates with the processor; the memory stores instructions, and when the instructions are executed by at least one processor, the steps of the method described in any one of claims 1 to 9 are implemented.

Images