Anti-falling monitoring method and system for emergency treatment monitoring based on sensor

Through sensor-based methods, the layout of obstacles and patient movement behaviors in the emergency ward are monitored in real time, and the risk of falling is predicted, which solves the problem of unpredictable falls in the existing technology, and improves the safety of emergency wards and the medical safety of patients. sex.

CN119091514AInactive Publication Date: 2024-12-06NANTONG TUMOR HOSPITAL

Patent Information

Application Number
CN202411572288.X
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-11-06
Publication Date
2024-12-06
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

The existing anti-fall monitoring methods cannot predict whether falls will occur based on the behavior of emergency patients in historical time periods, resulting in delayed rescue and increased risk of secondary injuries, and lack of effective preventive measures.

Method used

Using a sensor-based method, the emergency ward image data is collected in real time through visual sensors, the obstacle layout and patient behavior trajectory are extracted, the safety distance is calculated, the behavior prediction model is constructed, the risk of falling is predicted, and the reminder signal and early warning measures are issued.

Benefits of technology

Real-time monitoring and prediction of the movement behavior of emergency patients is achieved, the safety of emergency wards is improved, the occurrence of fall incidents is reduced, and early warning is provided in a timely manner to reduce patient injuries.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119091514A_ABST
    Figure CN119091514A_ABST
Patent Text Reader

Abstract

The invention discloses an anti-falling monitoring method and system for emergency treatment monitoring based on a sensor, and relates to the technical field of ward monitoring, the anti-falling monitoring method for emergency treatment monitoring based on the sensor comprises the following steps: extracting an obstacle layout in an emergency treatment ward from image data; extracting an action behavior track of the emergency patient from the image data, and dynamically tracking the action behavior track; calculating a safe distance between the action behavior track of the emergency patient and each obstacle; and collecting action behavior tracks of the emergency patient in a historical time period to construct a behavior prediction model, and predicting whether the emergency patient will fall down or not. According to the method, the behavior prediction model is established through collection of historical action behavior tracks, and the risk of falling of the patient can be evaluated, so that falling behaviors can be recognized in advance, preventive measures can be taken, and falling events are reduced.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of ward monitoring, and in particular to a fall prevention monitoring method and system for emergency monitoring based on sensors. Background Art

[0002] The emergency monitoring room is a facility in a medical institution specifically used to handle emergency patients. The emergency monitoring room is usually used to quickly evaluate, diagnose and treat patients, especially those who may be in emergency or life-threatening conditions. The medical environment is usually high-pressure and high-risk, so preventing emergency patients from falling can reduce the occurrence of secondary injuries, especially when the patient's original condition is serious, a fall may lead to more serious consequences. Therefore, it is extremely necessary to prevent patients from falling and provide timely intervention in the emergency room or other medical environments.

[0003] In addition, preventing patients from falling can reduce the need for more medical resources due to injuries, including emergency treatment, rehabilitation, and longer hospitalization services, which helps medical institutions use resources more effectively and improve service efficiency. Anti-fall monitoring is part of improving the quality and safety of medical services. By promptly detecting the risk of patients falling and taking preventive measures, the overall medical safety of patients can be improved.

[0004] However, existing fall prevention monitoring methods are usually unable to predict whether a patient will fall based on the behavior of emergency patients in a historical time period, and can only respond with rescue after the fall occurs, which not only leads to delayed rescue, but also increases the risk of secondary injury to emergency patients, thereby threatening the safety of emergency patients.

[0005] Currently, no effective solution has been proposed for the problems in the related technologies. Summary of the invention

[0006] In view of the deficiencies in the prior art, the present invention provides a fall prevention monitoring method and system for emergency monitoring based on sensors, which has the advantage of predicting fall action behaviors, thereby solving the problem of being unable to predict fall action behaviors.

[0007] In order to achieve the above advantages of predicting falling action behavior, the specific technical solutions adopted by the present invention are as follows: According to one aspect of the present invention, a fall prevention monitoring method for emergency monitoring based on a sensor is provided, and the fall prevention monitoring method for emergency monitoring based on a sensor comprises the following steps: S1. Using visual sensors to collect image data inside the emergency ward in real time, and extracting the layout of obstacles inside the emergency ward from the image data; S2, extracting the motion trajectory of the emergency patient from the image data, and dynamically tracking the motion trajectory; S3, calculating the safe distance between the movement trajectory of the emergency patient and each obstacle, and if the safe distance exceeds a preset threshold, issuing a warning signal command; S4. Collect the action and behavior trajectories of emergency patients in the historical period to build a behavior prediction model, and predict whether the emergency patients will fall; S5. Record and analyze fall events to determine the influencing factors of the fall events, and formulate alarm prevention measures based on the influencing factors.

[0008] Preferably, using a visual sensor to collect image data inside the emergency ward in real time, and extracting the obstacle layout inside the emergency ward from the image data includes the following steps: S11, using a visual sensor to collect image data inside the emergency ward, and dividing the image data into an obstacle image sequence and a patient action image sequence; S12, performing a two-dimensional discrete cosine transform on the obstacle image sequence to obtain a low-frequency coefficient, and using the low-frequency coefficient as an internal area feature of the emergency ward; S13, extracting and sampling the obstacle contours in the internal area features to form an ordered point list describing the obstacle contours; S14, performing clockwise and counterclockwise distance tracking on each sampling point in the ordered point list, and generating an obstacle shape descriptor according to the distance tracking result; S15. Generate local features of the obstacle shape description based on the obstacle shape descriptor, and generate the obstacle layout inside the emergency ward based on the local features of the obstacle shape description.

[0009] Preferably, performing clockwise and counterclockwise distance tracking on each sampling point in the ordered point list, and generating an obstacle shape descriptor according to the distance tracking result comprises the following steps: S141, tracking the sampling points clockwise and counterclockwise along the image boundaries in the obstacle image sequence to obtain two tracking points; S142, calculating the vertical distance between one of the tracking points and the line connecting the two tracking points, obtaining the arch height feature and the centroid distance of the obstacle contour, and using the arch height feature as the concave-convex feature of the obstacle contour; S143, fusing the arch height feature and the centroid distance to determine the shape description of the obstacle contour, and taking the arch height feature as the imaginary part of the obstacle contour and the centroid distance as the real part of the obstacle contour, thereby forming an arch height radius composite function; S144, performing discrete cosine transform on the composite function of arch height and radius to obtain a low-frequency cosine variation coefficient, and using the low-frequency cosine variation coefficient as an obstacle shape descriptor.

[0010] Preferably, extracting the motion behavior trajectory of the emergency patient from the image data and dynamically tracking the motion behavior trajectory includes the following steps: S21, extracting the action trajectory of the emergency patient from the patient action image sequence; S22. Construct a human body dynamics model based on emergency patients, and use the long short-term memory network as the backbone network of the human body dynamics model; S23, evaluating the motion rate of the emergency patient in combination with the human injury data of the emergency patient, and determining the motion load rate of the emergency patient according to the motion rate; S24, inputting the motion behavior trajectory and the motion load rate into the human body dynamics model as feature vectors, and using the human body dynamics model to calculate the velocity differential value of the time step, and at the same time cascading the velocity differential value and the feature vector to obtain a new time step; S25, calculating the final velocity differential value using the backbone network in the human body dynamics model, and calculating the change in the motion state of the emergency patient according to the final velocity differential value; S26. Use Euler integral to discretize the dynamic tracking algorithm, embed the human body dynamics model into the dynamic tracking algorithm, and then use the dynamic tracking algorithm to track the changes in the motion state of emergency patients in real time.

[0011] Preferably, evaluating the motion rate of the emergency patient in combination with the human injury data of the emergency patient, and determining the motion load rate of the emergency patient according to the motion rate comprises the following steps: S231. Combine the medical records and examination reports of emergency patients to obtain human injury data of emergency patients; S232, establishing an injury correlation between the human injury data and the motion rate of the emergency patient, so as to determine the influence of the human injury on the motion rate; S233. Analyze the motion load rate of the emergency patient in the motion behavior trajectory according to the impact degree judgment result.

[0012] Preferably, calculating the safe distance between the movement trajectory of the emergency patient and each obstacle, and if the safe distance exceeds a preset threshold, issuing a reminder signal instruction includes the following steps: S31, using a three-frame difference algorithm to perform difference calculation on three consecutive frames of images in the patient's motion image sequence to obtain a frame difference image, and performing binarization processing on the frame difference image; S32, performing an addition operation on the frame difference image after the binarization process to obtain a contour moment of the obstacle; S33, establishing a conversion relationship between the visual sensor and the world coordinates of the obstacle, and constructing an obstacle ranging model according to the conversion relationship; S34, calculating the safe distance between the obstacle and the emergency patient using an obstacle ranging model; S35: Compare the safety distance with a preset threshold, and send a reminder signal instruction based on the comparison result.

[0013] Preferably, comparing the safety distance with a preset threshold and sending a reminder signal instruction based on the comparison result comprises the following steps: S351. If the safety distance is less than the minimum preset threshold, it is determined that there is a risk of collision between the emergency patient and the obstacle, and a first-level reminder signal instruction is sent to the emergency patient and the medical staff at the same time; S352: If the safety distance is greater than the maximum preset threshold, it is determined that the distance between the emergency patient and the obstacle is far, and there is no need to send a signal instruction, and the safety distance between the emergency patient and the obstacle continues to be monitored; S353. If the safety distance is between the minimum preset threshold and the maximum preset threshold, it is determined that the distance between the emergency patient and the obstacle is relatively far, and a secondary reminder signal instruction is sent to the emergency patient.

[0014] Preferably, collecting the action behavior trajectories of emergency patients in a historical time period to build a behavior prediction model, and predicting whether the emergency patients will fall includes the following steps: S41, capturing the position information of the skeleton key points from the motion behavior trajectory of the emergency patient, and mapping each skeleton key point as a mapping feature point into a two-dimensional space; S42, reconstructing the weight matrix in combination with Lagrange's multiplication optimization, and building a behavior rule library; S43, collecting the action and behavior trajectories of emergency patients in a historical period to build a behavior prediction model; S44, using the behavior prediction model to predict the movement behavior of the emergency patient in the future time period, and comparing the predicted movement behavior with all the movement behaviors in the behavior rule library to determine whether the predicted movement behavior is a fall behavior; S45. If the predicted action behavior is determined to be a fall behavior, an alarm signal instruction is sent to medical staff and guardians at the same time, and rescue or guardianship measures are taken in time.

[0015] Preferably, the calculation formula for the safe distance between the obstacle and the emergency patient is: ; In the formula, d Indicates the safe distance between obstacles and emergency patients; u and p Both represent the center point of the image plane; o Represents the image plane; H Indicates the placement height of the visual sensor; h Indicates the sight height of the visual sensor; and All represent the intrinsic parameters of the visual sensor; x and y All represent coordinate values.

[0016] According to another aspect of the present invention, there is also provided an anti-fall monitoring system for emergency monitoring based on sensors, the anti-fall monitoring system for emergency monitoring based on sensors comprising an obstacle extraction module, a behavior trajectory tracking module, a collision reminder module, a fall behavior prediction module and a fall behavior warning module; Among them, the obstacle extraction module, the behavior trajectory tracking module, the collision reminder module, the fall behavior prediction module and the fall behavior warning module are connected in sequence; An obstacle extraction module is used to collect image data inside the emergency ward in real time using a visual sensor, and extract the obstacle layout inside the emergency ward from the image data; The behavior trajectory tracking module is used to extract the action trajectory of emergency patients from the image data and dynamically track the action trajectory; The collision warning module is used to calculate the safe distance between the movement trajectory of the emergency patient and each obstacle. If the safe distance exceeds the preset threshold, a warning signal command is issued; The fall behavior prediction module is used to collect the action behavior trajectories of emergency patients in the historical time period to build a behavior prediction model and predict whether emergency patients will fall; The fall behavior warning module is used to record and analyze fall events to determine the influencing factors of fall events and formulate alarm prevention measures based on the influencing factors.

[0017] Compared with the prior art, the present invention provides a fall prevention monitoring method and system for emergency monitoring based on sensors, which has the following beneficial effects: (1) The present invention can detect the abnormal movements and behaviors of emergency patients in a timely manner by dynamically tracking their movement and behavior trajectories, and can enable medical staff to understand the activity patterns, posture characteristics and other information of emergency patients. At the same time, dynamic tracking of movement and behavior can provide real-time position information of patients on the bed, so as to respond to the needs or emergency situations of emergency patients more quickly.

[0018] (2) The present invention can monitor the environment in the emergency ward in real time by extracting the obstacle layout from the image data. At the same time, by calculating the safe distance between the emergency patient's movement trajectory and the obstacle, the collision risk between the emergency patient and the obstacle can be monitored, thereby improving the safety of the emergency ward. At the same time, if there is a collision risk, a reminder signal instruction can be issued in time, which helps medical staff to respond quickly to possible emergency situations and enable medical staff to take countermeasures in time.

[0019] (3) The present invention establishes a behavior prediction model by collecting historical action behavior trajectories, which can assess the risk of patients falling, thereby helping to identify falling behaviors in advance and take preventive measures to reduce the occurrence of falling events. Based on the results of the behavior prediction model, medical staff can formulate personalized care plans and provide early warning signals before the patient is about to fall, allowing medical staff to intervene in time, thereby greatly reducing the damage caused by falls and further improving the safety of emergency patients. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0021] Figure 1 4 is a flow chart of a fall prevention monitoring method for emergency monitoring based on sensors according to an embodiment of the present invention.

[0022] Figure 2 4 is a principle block diagram of a fall prevention monitoring system for emergency monitoring based on sensors according to an embodiment of the present invention.

[0023] In the figure: 1. Obstacle extraction module; 2. Behavior trajectory tracking module; 3. Collision reminder module; 4. Fall behavior prediction module; 5. Fall behavior warning module. DETAILED DESCRIPTION

[0024] To further illustrate each embodiment, the present invention provides drawings, which are part of the disclosure of the present invention and are mainly used to illustrate the embodiments. They can be used in conjunction with the relevant descriptions in the specification to explain the operating principles of the embodiments. With reference to these contents, ordinary technicians in this field should be able to understand other possible implementation methods and advantages of the present invention.

[0025] According to an embodiment of the present invention, a fall prevention monitoring method and system for emergency monitoring based on sensors are provided.

[0026] The present invention is further described with reference to the accompanying drawings and specific embodiments. Figure 1 As shown, according to an anti-fall monitoring method for emergency monitoring based on sensors according to an embodiment of the present invention, the anti-fall monitoring method for emergency monitoring based on sensors includes the following steps: S1. Use visual sensors to collect image data inside the emergency ward in real time, and extract the obstacle layout inside the emergency ward from the image data.

[0027] The process of collecting image data of the interior of the emergency ward in real time using a visual sensor and extracting the obstacle layout of the interior of the emergency ward from the image data includes the following steps: S11. Collect image data inside the emergency ward using a visual sensor, and divide the image data into an obstacle image sequence and a patient action image sequence.

[0028] It should be noted that common obstacles in emergency wards include medical equipment, mobile equipment, and ward-deployed equipment.

[0029] S12. Perform a two-dimensional discrete cosine transform on the obstacle image sequence to obtain low-frequency coefficients, and use the low-frequency coefficients as internal area features of the emergency ward.

[0030] It should be noted that performing a two-dimensional discrete cosine transform on the obstacle image sequence to obtain low-frequency coefficients, and using the low-frequency coefficients as internal area features of the emergency ward includes the following steps: performing a two-dimensional discrete cosine transform on each image frame or image in the image sequence, converting the image into a frequency domain representation by applying DCT (discrete cosine transform), and obtaining low-frequency coefficients in the frequency domain; extracting the low-frequency coefficients from the result of the DCT transform, and considering the extracted low-frequency coefficients as internal area features of the emergency ward; integrating the extracted low-frequency coefficients into the feature representation of the emergency ward, and performing environmental recognition, anomaly detection or other related tasks; optimizing and verifying the extracted internal area features to ensure that these features can effectively reflect the important attributes of the internal environment of the emergency ward.

[0031] S13. Extract obstacle contours from the internal area features and perform sampling to form an ordered point list describing the obstacle contours.

[0032] S14. Perform clockwise and counterclockwise distance tracking on each sampling point in the ordered point list, and generate an obstacle shape descriptor according to the distance tracking result.

[0033] The following steps are used to track the distance of each sampling point in the ordered point list in a clockwise and counterclockwise direction and generate an obstacle shape descriptor according to the distance tracking result: S141 , tracking the sampling points clockwise and counterclockwise along the image boundaries in the obstacle image sequence to obtain two tracking points.

[0034] S142, calculating the vertical distance between one of the tracking points and the line connecting the two tracking points, obtaining the arch height feature and the centroid distance of the obstacle contour, and using the arch height feature as the concave-convex feature of the obstacle contour.

[0035] S143, the shape description of the obstacle contour is determined by fusing the arch height feature and the centroid distance, and the arch height feature is used as the imaginary part of the obstacle contour, and the centroid distance is used as the real part of the obstacle contour, thereby forming an arch height radius composite function.

[0036] It should be noted that the arch height refers to the vertical distance from the obstacle outline to the farthest point above it. This feature is usually used to indicate the height of the obstacle or the degree of protrusion in the image. A larger arch height may indicate that the obstacle is taller or has a more protruding shape.

[0037] The centroid is the average position of all points in the obstacle contour, and the centroid distance is the average distance from each point in the contour to the centroid. This feature can provide information about the overall distribution of the obstacle contour. A larger centroid distance may indicate that the contour is more dispersed, while a smaller centroid distance may indicate that the contour is more concentrated.

[0038] S144, performing discrete cosine transform on the composite function of arch height and radius to obtain a low-frequency cosine variation coefficient, and using the low-frequency cosine variation coefficient as an obstacle shape descriptor.

[0039] It should be noted that performing discrete cosine transform on the arch height radius composite function to obtain low-frequency cosine variation coefficients, and using the low-frequency cosine variation coefficients as obstacle shape descriptors includes the following steps: regarding the arch height radius composite function as a time domain signal, and applying discrete cosine transform to obtain the corresponding frequency domain representation; extracting low-frequency cosine variation coefficients from the result of DCT transform, these low-frequency coefficients usually contain the main features of the signal; using the extracted low-frequency cosine variation coefficients as obstacle shape descriptors, these descriptors reflect the main features of the arch height radius composite function in the frequency domain, which are mainly used to represent and compare the shape information of the obstacle.

[0040] S15. Generate local features of the obstacle shape description based on the obstacle shape descriptor, and generate the obstacle layout inside the emergency ward based on the local features of the obstacle shape description.

[0041] S2. Extract the motion trajectory of the emergency patient from the image data and dynamically track the motion trajectory.

[0042] The steps of extracting the motion behavior trajectory of the emergency patient from the image data and dynamically tracking the motion behavior trajectory include the following steps: S21. Extracting the motion behavior trajectory of the emergency patient from the patient motion image sequence.

[0043] It should be noted that the motion behavior trajectory of emergency patients is extracted from the patient motion image sequence: the patient motion image sequence is analyzed and processed to identify the patient's motion behavior; the identified patient motion behavior may be discrete motion fragments, and these motion behaviors are fused using time series data processing methods to generate a more continuous and smooth motion behavior trajectory; the fused motion behavior is converted into a motion trajectory and analyzed using the generated trajectory data.

[0044] S22. Construct a human body dynamics model based on emergency patients, and use the long short-term memory network as the backbone network of the human body dynamics model.

[0045] S23. Evaluate the movement rate of the emergency patient in combination with the human injury data of the emergency patient, and determine the movement load rate of the emergency patient based on the movement rate.

[0046] The steps of evaluating the motion rate of the emergency patient in combination with the human injury data of the emergency patient and determining the motion load rate of the emergency patient according to the motion rate include: S231. Acquire the human injury data of emergency patients by combining their medical records and examination reports.

[0047] S232. Establish an injury correlation between the human injury data and the movement rate of emergency patients to determine the degree of influence of the human injury on the movement rate.

[0048] It should be noted that human injuries (such as fractures, joint injuries, and postoperative effects) may limit the patient's ability to move, resulting in a significant decrease in movement rate. Minor injuries or pain may affect the patient's movement rate, but the degree of impact may be relatively small. Comprehensive evaluation and analysis of these factors will help to more accurately understand the patient's movement status and needs.

[0049] Establishing the injury correlation between the human injury data and the movement rate of emergency patients to determine the degree of influence of human injury on movement rate includes the following steps: using the Pearson correlation coefficient to analyze the correlation between injury data and movement rate, the correlation analysis will provide quantitative information about the relationship between the two variables; dividing the data set into a training set and a test set, the training set is used to establish a multivariate regression model, and the test set is used to evaluate the performance of the model; using evaluation indicators to measure the accuracy and generalization ability of the model, and analyzing the coefficients or feature importance of the model to explain the degree of influence of human injury on movement rate.

[0050] S233. Analyze the motion load rate of the emergency patient in the motion behavior trajectory according to the impact degree judgment result.

[0051] S24, inputting the motion behavior trajectory and the motion load rate into the human body dynamics model as feature vectors, and using the human body dynamics model to calculate the velocity differential value of the time step, and at the same time cascading the velocity differential value and the feature vector to obtain a new time step.

[0052] S25. Calculate the final velocity differential value using the backbone network in the human body dynamics model, and calculate the change in the motion state of the emergency patient based on the final velocity differential value.

[0053] It should be noted that the calculation of the final velocity differential value using the backbone network in the human body dynamics model, and the calculation of the change in the motion state of the emergency patient based on the final velocity differential value include the following steps: using a numerical simulation method to calculate the final velocity of the backbone network; differentiating the numerical result of the final velocity to obtain the derivative of the velocity with respect to time, that is, the differential value of the final velocity; collecting relevant information of the emergency patient and using this information as input parameters of the model; inputting the information of the emergency patient into the established human body dynamics model for model calculation and obtaining the final velocity differential value; and using the final velocity differential value and other relevant data to calculate the change in the motion state of the emergency patient.

[0054] S26. Use Euler integral to discretize the dynamic tracking algorithm, embed the human body dynamics model into the dynamic tracking algorithm, and then use the dynamic tracking algorithm to track the changes in the motion state of emergency patients in real time.

[0055] It should be noted that the Euler integration method is a basic numerical integration method that approximates the solution of a differential equation by dividing the time continuity into discrete time steps. The general form of the Euler integration is to approximate the derivative in the differential equation by using differences.

[0056] It should be noted that the discretization design of the dynamic tracking algorithm using Euler integral includes the following steps: converting the continuous time in the dynamic tracking algorithm into discrete time, that is, dividing the time continuity into discrete time steps; using the discretized Euler integral model and iteratively applying the update rule, the state changes in discrete time steps can be obtained.

[0057] S3. Calculate the safe distance between the movement trajectory of the emergency patient and each obstacle. If the safe distance exceeds a preset threshold, issue a warning signal command.

[0058] Among them, calculating the safe distance between the movement trajectory of the emergency patient and each obstacle, and if the safe distance exceeds the preset threshold, issuing a reminder signal instruction includes the following steps: S31, using a three-frame difference algorithm to perform difference calculation on three consecutive frames of images in the patient's motion image sequence to obtain a frame difference image, and performing binarization processing on the frame difference image.

[0059] It should be noted that the three-frame difference algorithm is used to perform differential calculation on three consecutive frames in the patient's action image sequence to obtain a frame difference image, and the frame difference image is binarized, including the following steps: differential calculation is performed on three consecutive frames, and the difference image between the current frame and the previous frame is obtained by subtracting the grayscale value of the previous frame, and then the difference image is subtracted from the difference image of the previous two frames to obtain a frame difference image of three consecutive frames; the frame difference image is enhanced, and the action area is highlighted by using a filter or adjusting the image contrast and brightness; the frame difference image is binarized to convert the pixel values ​​in the image into binary values ​​(0 or 1). This can be achieved by setting a threshold, setting pixel values ​​greater than the threshold to 1, and setting pixel values ​​less than or equal to the threshold to 0. The binarized image is easier to analyze the shape and contour.

[0060] S32, performing an addition operation on the frame difference images after the binarization process to obtain a contour moment of the obstacle.

[0061] S33. Establish a conversion relationship between the visual sensor and the world coordinates of the obstacle, and construct an obstacle ranging model based on the conversion relationship.

[0062] It should be noted that establishing the conversion relationship between the visual sensor and the world coordinates of the obstacle, and constructing the obstacle ranging model based on the conversion relationship includes the following steps: for each pixel point in the image, the image coordinates are converted into the external parameters of the visual sensor coordinates using the visual sensor internal parameters; for the detected obstacle, the position of the obstacle in the world coordinate system is determined according to the conversion relationship from the visual sensor coordinate system to the world coordinate system, and the obstacle ranging model is constructed.

[0063] S34. Calculate the safe distance between the obstacle and the emergency patient using an obstacle ranging model.

[0064] Among them, the calculation formula for the safe distance between obstacles and emergency patients is: ; In the formula, d Indicates the safe distance between obstacles and emergency patients; u and p Both represent the center point of the image plane; It should be noted that the center point of the image plane represents the optical center or principal point of the visual sensor.

[0065] o Represents the image plane; H Indicates the placement height of the visual sensor; h Indicates the sight height of the visual sensor; and All represent the intrinsic parameters of the visual sensor; x and y All represent coordinate values.

[0066] It should be noted that the intrinsic parameters of a visual sensor are a set of parameters that describe its internal characteristics. These parameters are very important for tasks such as converting image coordinates into world coordinates and performing geometric correction.

[0067] S35: Compare the safety distance with a preset threshold, and send a reminder signal instruction based on the comparison result.

[0068] The step of comparing the safety distance with a preset threshold and sending a reminder signal instruction based on the comparison result includes the following steps: S351. If the safety distance is less than the minimum preset threshold, it is determined that there is a risk of collision between the emergency patient and the obstacle, and a first-level reminder signal instruction is sent to the emergency patient and medical staff at the same time.

[0069] S352: If the safety distance is greater than the maximum preset threshold, it is determined that the distance between the emergency patient and the obstacle is far, and there is no need to send a signal instruction, and the safety distance between the emergency patient and the obstacle continues to be monitored.

[0070] S353. If the safety distance is between the minimum preset threshold and the maximum preset threshold, it is determined that the distance between the emergency patient and the obstacle is relatively far, and a secondary reminder signal instruction is sent to the emergency patient.

[0071] It should be noted that the distinction between the first-level warning signal instructions and the second-level warning signal instructions generally indicates different levels of urgency or importance. The purpose of this classification and distinction is to take appropriate responses and actions in different situations.

[0072] A Level 1 warning signal command usually indicates a more urgent or critical situation that requires rapid response and processing. When a Level 1 warning signal command is sent to emergency patients and medical staff, there may be a risk of collision and quick measures need to be taken to avoid potential dangers. This level of warning usually involves urgent safety issues and immediate measures may need to be taken to ensure safety.

[0073] The second-level warning signal instruction is a relatively low level, indicating that the situation is relatively far away or relatively urgent. In this case, although the safe distance from the obstacle is within a certain range, it is relatively far away, and emergency measures may not be required immediately, but it still needs attention. The second-level reminder is usually used to remind emergency patients to maintain a certain safe distance, but it is not an immediately critical situation.

[0074] S4. Collect the action and behavior trajectories of emergency patients in the historical period to build a behavior prediction model, and predict whether the emergency patients will fall.

[0075] Among them, collecting the action and behavior trajectories of emergency patients in the historical time period to build a behavior prediction model and predicting whether the emergency patients will fall includes the following steps: S41. Capture the position information of the skeleton key points from the motion behavior trajectory of the emergency patient, and map each skeleton key point into the two-dimensional space as a mapping feature point.

[0076] S42. Reconstruct the weight matrix by combining Lagrange's multiplication optimization and build a behavior rule library.

[0077] It should be noted that the reconstruction of the weight matrix in combination with Lagrange multiplication optimization and the construction of the behavior rule library include the following steps: by introducing Lagrange multipliers, the constrained optimization problem is transformed into an unconstrained problem, and the optimal solution is obtained by solving the corresponding Lagrange equation; the weight matrix is ​​usually used to represent the degree of influence of different factors on a process, and Lagrange multiplication helps to find the optimal weight matrix to best meet your optimization goals and constraints; define the behavior rule library based on the optimal reconstructed weight matrix.

[0078] S43. Collect the action and behavior trajectories of emergency patients in a historical period to build a behavior prediction model.

[0079] S44. Use the behavior prediction model to predict the movement behavior of the emergency patient in the future time period, and compare the predicted movement behavior with all the movement behaviors in the behavior rule library to determine whether the predicted movement behavior is a fall behavior.

[0080] It should be noted that using a behavior prediction model to predict the movement behavior of emergency patients in future time periods is an approach of applying machine learning or other predictive modeling methods. Once the prediction results of the movement behavior are obtained, the following steps can be used to compare it with the behavior rule library to determine whether the predicted movement behavior is a fall behavior: In the behavior rule library, clearly define the rules for fall behavior, which may include characteristics of movement patterns, postures, speeds, etc., so as to accurately judge the fall; extract corresponding features from the prediction results to ensure that the extracted features match the features defined in the behavior rule library; compare the extracted features with the rules in the behavior rule library; if the features of the predicted movement behavior match the fall behavior rules, it can be determined that the patient is about to fall; if it is determined that the patient is about to fall, trigger corresponding response measures, such as sending an emergency notification, calling medical staff, etc.

[0081] S45. If the predicted action behavior is determined to be a fall behavior, an alarm signal instruction is sent to medical staff and guardians at the same time, and rescue or guardianship measures are taken in time.

[0082] S5. Record and analyze fall events to determine the influencing factors of the fall events, and formulate alarm prevention measures based on the influencing factors.

[0083] It should be noted that for each fall incident, a detailed event record should be established, including but not limited to the patient's time, surrounding environment (such as light, ground conditions, etc.), patient status and other information; data related to the fall event, as well as the patient's current physiological parameters and environmental conditions, etc. are collected; the collected data are used for analysis to identify factors that may affect the occurrence of falls, which may include specific ground conditions, insufficient light, the patient's physical condition, etc.; based on the analysis results, corresponding alarm prevention measures are formulated. For example, if it is found that the risk of falling is higher under certain environmental conditions, you can consider increasing vigilance under similar conditions, or reducing the risk of falling by providing additional support facilities.

[0084] like Figure 2 As shown, according to another embodiment of the present invention, there is also provided an anti-fall monitoring system for emergency monitoring based on sensors. The anti-fall monitoring system for emergency monitoring based on sensors includes an obstacle extraction module 1, a behavior trajectory tracking module 2, a collision reminder module 3, a fall behavior prediction module 4 and a fall behavior warning module 5.

[0085] Among them, the obstacle extraction module 1, the behavior trajectory tracking module 2, the collision reminder module 3, the fall behavior prediction module 4 and the fall behavior warning module 5 are connected in sequence.

[0086] The obstacle extraction module 1 is used to collect image data inside the emergency ward in real time using a visual sensor, and extract the obstacle layout inside the emergency ward from the image data.

[0087] The behavior trajectory tracking module 2 is used to extract the action behavior trajectory of the emergency patient from the image data and dynamically track the action behavior trajectory.

[0088] The collision warning module 3 is used to calculate the safe distance between the movement trajectory of the emergency patient and each obstacle. If the safe distance exceeds a preset threshold, a warning signal instruction is issued.

[0089] The fall behavior prediction module 4 is used to collect the action behavior trajectories of emergency patients in a historical period to build a behavior prediction model and predict whether the emergency patients will have a fall event.

[0090] The fall behavior warning module 5 is used to record and analyze the fall events to determine the influencing factors of the fall events and formulate alarm prevention measures based on the influencing factors.

[0091] In summary, with the help of the above-mentioned technical scheme of the present invention, the present invention can detect the abnormal movement behavior of the emergency patient in time by dynamically tracking the movement behavior trajectory of the emergency patient, and can enable medical staff to understand the activity pattern, posture characteristics and other information of the emergency patient. At the same time, dynamic tracking of movement behavior can provide real-time position information of the patient on the bed, and respond to the needs or emergency situations of the emergency patient more quickly; the present invention can monitor the environment in the emergency ward in real time by extracting the obstacle layout from the image data, and at the same time, by calculating the safe distance between the movement behavior trajectory of the emergency patient and the obstacle, the collision risk between the emergency patient and the obstacle can be monitored, thereby improving the emergency ward. The invention can improve the safety of the room, and at the same time, if there is a risk of collision, it can issue a warning signal instruction in time, which helps medical staff to respond quickly to possible emergencies, so that medical staff can take countermeasures in time; the present invention establishes a behavior prediction model by collecting historical action behavior trajectories, which can evaluate the risk of patients falling, thereby helping to identify falling behaviors in advance and take preventive measures to reduce the occurrence of falling events. Based on the results of the behavior prediction model, medical staff can formulate personalized care plans and provide early warning signals before the patient is about to fall, so that medical staff can intervene in time, thereby greatly reducing the injuries caused by falls and further improving the safety of emergency patients.

[0092] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A fall prevention monitoring method for emergency monitoring based on sensors, characterized in that: The anti-fall monitoring method for emergency monitoring based on sensors comprises the following steps: S1. Using visual sensors to collect image data inside the emergency ward in real time, and extracting the layout of obstacles inside the emergency ward from the image data; S2, extracting the motion trajectory of the emergency patient from the image data, and dynamically tracking the motion trajectory; S3, calculating the safe distance between the movement trajectory of the emergency patient and each obstacle, and if the safe distance exceeds a preset threshold, issuing a warning signal command; S4. Collect the action and behavior trajectories of emergency patients in the historical period to build a behavior prediction model, and predict whether the emergency patients will fall; S5. Record and analyze fall events to determine the influencing factors of the fall events, and formulate alarm prevention measures based on the influencing factors.

2. The anti-fall monitoring method for emergency monitoring based on sensors according to claim 1, characterized in that: The method of collecting image data inside the emergency ward in real time by using a visual sensor and extracting the obstacle layout inside the emergency ward from the image data comprises the following steps: S11, using a visual sensor to collect image data inside the emergency ward, and dividing the image data into an obstacle image sequence and a patient action image sequence; S12, performing a two-dimensional discrete cosine transform on the obstacle image sequence to obtain a low-frequency coefficient, and using the low-frequency coefficient as an internal area feature of the emergency ward; S13, extracting and sampling the obstacle contours in the internal area features to form an ordered point list describing the obstacle contours; S14, performing clockwise and counterclockwise distance tracking on each sampling point in the ordered point list, and generating an obstacle shape descriptor according to the distance tracking result; S15. Generate local features of the obstacle shape description based on the obstacle shape descriptor, and generate the obstacle layout inside the emergency ward based on the local features of the obstacle shape description.

3. The anti-fall monitoring method for emergency monitoring based on sensors according to claim 2, characterized in that: The step of tracking the distance of each sampling point in the ordered point list in a clockwise and counterclockwise direction and generating an obstacle shape descriptor according to the distance tracking result comprises the following steps: S141, tracking the sampling points clockwise and counterclockwise along the image boundaries in the obstacle image sequence to obtain two tracking points; S142, calculating the vertical distance between one of the tracking points and the line connecting the two tracking points, obtaining the arch height feature and the centroid distance of the obstacle contour, and using the arch height feature as the concave-convex feature of the obstacle contour; S143, fusing the arch height feature and the centroid distance to determine the shape description of the obstacle contour, and taking the arch height feature as the imaginary part of the obstacle contour and the centroid distance as the real part of the obstacle contour, thereby forming an arch height radius composite function; S144, performing discrete cosine transform on the composite function of arch height and radius to obtain a low-frequency cosine variation coefficient, and using the low-frequency cosine variation coefficient as an obstacle shape descriptor.

4. The anti-fall monitoring method for emergency monitoring based on sensors according to claim 3, characterized in that: The method of extracting the motion behavior trajectory of the emergency patient from the image data and dynamically tracking the motion behavior trajectory comprises the following steps: S21, extracting the action trajectory of the emergency patient from the patient action image sequence; S22. Construct a human body dynamics model based on emergency patients, and use the long short-term memory network as the backbone network of the human body dynamics model; S23, evaluating the motion rate of the emergency patient in combination with the human injury data of the emergency patient, and determining the motion load rate of the emergency patient according to the motion rate; S24, inputting the motion behavior trajectory and the motion load rate into the human body dynamics model as feature vectors, and using the human body dynamics model to calculate the velocity differential value of the time step, and at the same time cascading the velocity differential value and the feature vector to obtain a new time step; S25, calculating the final velocity differential value using the backbone network in the human body dynamics model, and calculating the change in the motion state of the emergency patient according to the final velocity differential value; S26. Use Euler integral to discretize the dynamic tracking algorithm, embed the human body dynamics model into the dynamic tracking algorithm, and then use the dynamic tracking algorithm to track the changes in the motion state of emergency patients in real time.

5. The anti-fall monitoring method for emergency monitoring based on sensors according to claim 4, characterized in that: The step of evaluating the motion rate of the emergency patient in combination with the human injury data of the emergency patient and determining the motion load rate of the emergency patient according to the motion rate comprises the following steps: S231. Combine the medical records and examination reports of emergency patients to obtain human injury data of emergency patients; S232, establishing an injury correlation between the human injury data and the motion rate of the emergency patient, so as to determine the influence of the human injury on the motion rate; S233. Analyze the motion load rate of the emergency patient in the motion behavior trajectory according to the impact degree judgment result.

6. The anti-fall monitoring method for emergency monitoring based on sensors according to claim 5, characterized in that: The step of calculating the safe distance between the movement trajectory of the emergency patient and each obstacle and issuing a warning signal instruction if the safe distance exceeds a preset threshold comprises the following steps: S31, using a three-frame difference algorithm to perform difference calculation on three consecutive frames of images in the patient's motion image sequence to obtain a frame difference image, and performing binarization processing on the frame difference image; S32, performing an addition operation on the frame difference image after the binarization process to obtain a contour moment of the obstacle; S33, establishing a conversion relationship between the visual sensor and the world coordinates of the obstacle, and constructing an obstacle ranging model according to the conversion relationship; S34, calculating the safe distance between the obstacle and the emergency patient using an obstacle ranging model; S35: Compare the safety distance with a preset threshold, and send a reminder signal instruction based on the comparison result.

7. The anti-fall monitoring method for emergency monitoring based on sensors according to claim 6, characterized in that: The step of comparing the safety distance with a preset threshold and sending a reminder signal instruction based on the comparison result includes the following steps: S351. If the safety distance is less than the minimum preset threshold, it is determined that there is a risk of collision between the emergency patient and the obstacle, and a first-level reminder signal instruction is sent to the emergency patient and the medical staff at the same time; S352: If the safety distance is greater than the maximum preset threshold, it is determined that the distance between the emergency patient and the obstacle is far, and there is no need to send a signal instruction, and the safety distance between the emergency patient and the obstacle continues to be monitored; S353. If the safety distance is between the minimum preset threshold and the maximum preset threshold, it is determined that the distance between the emergency patient and the obstacle is relatively far, and a secondary reminder signal instruction is sent to the emergency patient.

8. The anti-fall monitoring method for emergency monitoring based on sensors according to claim 7, characterized in that: The step of collecting the action and behavior trajectories of emergency patients in a historical time period to build a behavior prediction model and predicting whether the emergency patients will fall includes the following steps: S41, capturing the position information of the skeleton key points from the motion behavior trajectory of the emergency patient, and mapping each skeleton key point as a mapping feature point into a two-dimensional space; S42, reconstructing the weight matrix in combination with Lagrange's multiplication optimization, and building a behavior rule library; S43, collecting the action and behavior trajectories of emergency patients in a historical period to build a behavior prediction model; S44, using the behavior prediction model to predict the movement behavior of the emergency patient in the future time period, and comparing the predicted movement behavior with all the movement behaviors in the behavior rule library to determine whether the predicted movement behavior is a fall behavior; S45. If the predicted action behavior is determined to be a fall behavior, an alarm signal instruction is sent to medical staff and guardians at the same time, and rescue or guardianship measures are taken in time.

9. The anti-fall monitoring method for emergency monitoring based on sensors according to claim 8, characterized in that: The calculation formula for the safe distance between the obstacle and the emergency patient is: ; In the formula, d Indicates the safe distance between obstacles and emergency patients; u and p Both represent the center point of the image plane; o Represents the image plane; H Indicates the placement height of the visual sensor; h Indicates the sight height of the visual sensor; and All represent the intrinsic parameters of the visual sensor; x and y All represent coordinate values.

10. A fall prevention monitoring system for emergency monitoring based on sensors, used to implement the fall prevention monitoring method for emergency monitoring based on sensors according to any one of claims 1 to 9, characterized in that: The anti-fall monitoring system for emergency monitoring based on sensors includes an obstacle extraction module, a behavior trajectory tracking module, a collision reminder module, a fall behavior prediction module and a fall behavior warning module; Wherein, the obstacle extraction module, the behavior trajectory tracking module, the collision reminder module, the fall behavior prediction module and the fall behavior warning module are connected in sequence; The obstacle extraction module is used to collect image data inside the emergency ward in real time using a visual sensor, and extract the obstacle layout inside the emergency ward from the image data; The behavior trajectory tracking module is used to extract the action behavior trajectory of the emergency patient from the image data and dynamically track the action behavior trajectory; The collision warning module is used to calculate the safe distance between the movement trajectory of the emergency patient and each obstacle, and if the safe distance exceeds a preset threshold, a warning signal instruction is issued; The fall behavior prediction module is used to collect the action behavior trajectories of emergency patients in a historical period to build a behavior prediction model and predict whether the emergency patients will have a fall event; The fall behavior warning module is used to record and analyze fall events to determine the influencing factors of the fall events and formulate alarm prevention measures based on the influencing factors.

Citation Information

Patent Citations

  • Real-time fall prediction and power-assisted recovery method based on decision tree model

    CN115054237A

  • Monitoring method and device for indoor toilet of nursing home, storage medium and electronic equipment

    CN117746576A

  • Target tracking and trajectory prediction method and device, equipment and storage medium

    CN117890922A

  • Fall protection intelligent alarm method based on millimeter wave radar

    CN118230502A

Cited By

  • Emergency treatment anti-falling monitoring method and system for emergency treatment monitoring

    CN119851420A