CN115910310A - Method for determining standard threshold of cardio-pulmonary resuscitation pressing posture and processor - Google Patents

Abstract

The present invention relates to a method and a processor for determining standard thresholds of cardiopulmonary resuscitation compression postures. The method at least includes: using a first optical component and a second optical component with different acquisition angles to simultaneously acquire CPR actions, based on the first The first action data collected by the optical component and the second action data collected by the second optical component extract the bone point data of the human body and at least calculate the posture angle data of the arms and the center of gravity matching angle data related to the CPR action, and select a number of CPR The unilateral skewed distribution law of the arm posture angle data and the center of gravity matching angle data of the action specification is used to determine the reasonable range of the arms posture angle data and the reasonable range of the center of gravity matching angle data. The present invention constructs the standard threshold of cardiopulmonary resuscitation compression posture based on the confidence level, so that the evaluation of CPR action has a data standard, constructs an objective and standardized evaluation standard, and avoids the deviation of the instructor's subjective judgment.

Description

A method and processor for determining the standard threshold of compression posture for cardiopulmonary resuscitation

technical field

The invention relates to the technical field of artificial intelligence first aid training, in particular to a method and a processor for determining a standard threshold value of a cardiopulmonary resuscitation compression posture.

Background technique

Once a cardiac arrest occurs, if it is not rescued in time, it will cause irreversible damage to the brain and other vital organs and tissues of the human body after 4 to 6 minutes, and even endanger life. Immediate high-quality cardiopulmonary resuscitation (CPR) is required on the spot. High-quality CPR is the foundation of both basic and advanced life support. The reason for emphasizing high quality is that if the standard is not met, the tissues and organs will not be able to obtain sufficient perfusion, resulting in a significant reduction in the success rate of resuscitation, especially the nervous system is very sensitive to ischemia and hypoxia. Although many patients restore circulation through resuscitation, they cannot Cause irreversible brain damage and seriously affect the quality of life after resuscitation. The first guidelines for CPR were published in 1966 by an ad hoc CPR committee of the National Research Council of the Division of Medicine of the National Academy of Sciences. More than half a century after the first guidelines were published, cardiac arrest remains the leading cause of death and health. The guidelines have always strived to use current relevant evidence to develop clear and actionable criteria to optimize CPR performance, thereby saving lives and being effective. However, it is not easy to apply these indicators to CPR rescue, and requires professional multiple or even repeated CPR training. Most of the current CPR training relies on the personal ability and subjective judgment of trainers, and lacks unified and standardized monitoring methods, resulting in uneven training levels.

Not only that, most of the currently known human bone extraction algorithms are based on natural standing positions or other special sports positions. However, the kneeling position is used during CPR operation. At present, there is no bone extraction algorithm that specifically recognizes CPR actions. The currently known bone extraction algorithms are not accurate enough for CPR action extraction. The bone extraction algorithm is the basis of subsequent CPR with artificial intelligence technology, which is very important. Therefore, in order to solve the deficiencies of the prior art, the present invention determines the standard threshold of cardiopulmonary resuscitation compression posture based on the bone extraction algorithm.

In order to realize high-quality training of cardiopulmonary resuscitation, many kinds of training systems and detection systems have also been proposed in the prior art. For example, the Chinese patent with publication number CN105224383A discloses a cardiopulmonary resuscitation simulation system, the user can select the operation mode and execution parameters of the cardiopulmonary resuscitation through the setting display part, and the processing part can The parameters simulate cardiopulmonary resuscitation and simulate the simulation results. The curve generation part can generate pressure waveforms, cardiac output waveforms, and cerebral blood flow/coronary blood flow waveforms and display them in the curve display area. The cycle average calculation part can calculate The cycle average value of each parameter is displayed in the parameter display area, and the evaluation department can compare the simulated result with the standard threshold to obtain a comparison result for evaluating whether the implemented cardiopulmonary resuscitation is effective.

For another example, the Chinese patent with publication number CN110990649A discloses an interactive training system for cardiopulmonary resuscitation based on gesture recognition technology, including a somatosensory recognition probe and interactive software. When connected with the computer equipment and collected by the interactive software, the following steps are completed when the interactive software is executed by the computer: obtain the real-time posture data of the human body through the body-sensory recognition probe; combine the real-time posture data with the cardiopulmonary resuscitation standard posture data in the database Perform comparison and output the comparison result; display the comparison result on the display device. The invention can realize the posture acquisition of personnel during cardiopulmonary resuscitation training through the somatosensory recognition probe. At the same time, it can be compared with the standard cardiopulmonary resuscitation action, and can output the comparison result, so that it can be checked in real time whether the action is standard. Action correction, complete training.

The monitoring standard of the above-mentioned prior art comes from the direct training of the collected CPR actions by the neural network, but because the collected actions are blocked by the clothing on the surface of the human body and the rigidity of the different characteristics of the human body, the collected CPR actions are not consistent with the operator's skeleton. There are large differences in actions. On this basis, the CPR action detection model based on neural network training is not accurate enough. Not only that, the CPR action detection model obtained by direct training of CPR action based on neural network requires that the acquisition angle of the CPR action image must be the same as the acquisition angle of the modeling to be able to accurately detect the two-dimensional CPR action, otherwise it cannot be accurately detected. detection. This restricts the location of the collection equipment, so that the collection equipment cannot be placed arbitrarily based on the actual site conditions.

Therefore, based on the defects that the detection standard progress of the CPR detection model in the current prior art is relatively low, and the collection angle of the collection device is limited, the present invention hopes to provide a new method for determining the standard threshold, and also provides a collection device Arbitrary setting of the acquisition angle can make the detection model realize the standard threshold of high-precision detection.

In addition, on the one hand, due to differences in the understanding of those skilled in the art; The present invention does not possess the characteristics of these prior art, on the contrary, the present invention already possesses all the characteristics of the prior art, and the applicant reserves the right to add relevant prior art to the background technology.

Contents of the invention

Aiming at the deficiencies of the prior art, the present invention provides a method for determining the standard threshold of cardiopulmonary resuscitation compression posture, the method at least includes: using the first optical assembly and the second optical assembly at different acquisition angles to simultaneously acquire CPR actions, based on The first motion data collected by the first optical component and the second motion data collected by the second optical component extract the skeletal point data of the human body and at least calculate the posture angle data of the arms and the center-of-gravity matching angle related to the CPR motion Data, select the unilateral skewed distribution law of the double-arm posture angle data and the center of gravity matching angle data of several CPR action specifications to determine the reasonable range of the double-arm posture angle data and the reasonable range of the center of gravity matching angle data.

Aiming at the defect that the CPR action detection model in the prior art has no clear standard threshold value, the present invention establishes a standard threshold value for cardiopulmonary resuscitation compression posture, so that the CPR action detection model can accurately provide a more accurate judgment comparison for the monitored action based on the standard threshold value , so that the action model can be applied to the training of the CPR action and the standard assessment of the CPR action, so as to improve the accuracy of the training and assessment of the CPR action.

Specifically, the present invention calculates the double-arm angle and center-of-gravity matching angle of each sample data by performing statistics and calculations on several sample data. The posture of the CPR action has a high correlation with the angle of the arms and the matching angle of the center of gravity. When the angle of both arms and the matching angle of the center of gravity are in a reasonable range, the force of the arms of the CPR operator will be relatively balanced. Unreasonable force on the back leads to excessive fatigue in the lower back. When both arm angles and center matching angles are within a reasonable range, the triangle formed by the straight line between the shoulders and wrists of both arms and the line between the shoulders is closer to an isosceles triangle, which makes the difference in the force of the arms the same. would be too large, while the lateral offset of the operator's center of gravity is small.

If the range of the operator's arms and the matching angle range of the center of gravity are unreasonable, it indicates that the operator's arms are not balanced, and the lateral shift of the center of gravity is large, and the operator is prone to limb fatigue due to the uneven force of the arms. An unreasonable matching angle of the center of gravity indicates that the operator's forward leaning range is insufficient or excessive, which makes the operator's back exert more force and the waist is prone to fatigue. When the CPR operator is prone to fatigue, it will make the operator's subsequent pressing force insufficient, thereby reducing the success rate of cardiopulmonary resuscitation.

Therefore, the angle range of the arms and the matching angle range of the center of gravity are both in a reasonable range, that is, the threshold range, which can obviously regulate the CPR operator's posture to a standard posture, delay the operator's fatigue time, and improve the success rate of cardiopulmonary resuscitation.

Preferably, the 5% percentile is selected based on the unilateral skew distribution law of the posture angle data of both arms to determine the reasonable range of the posture angle data of the arms, and the 95% percentile is selected based on the unilateral skew distribution law of the center of gravity matching angle data The number of bits determines a reasonable range for the center of gravity to match the angle data.

In the traditional CPR action detection, the system only pays attention to the range of motion or bending angle of the arms, and ignores the influence of the center of gravity on the strength of the body. As a result, even if the arms angle and compression force are qualified, the distribution of arm strength is actually balanced. Yes, so CPR trainees will feel tired and there will be insufficient pressure in the later stage. Compared with the traditional detection method that only restricts the angle of the arms and lacks the center of gravity constraint, the present invention uses the matching angle of the center of gravity as an objective parameter to detect the standard degree of posture, so that the CPR operator can balance the strength of both arms when performing the operation, and the back The forward lean range is reasonable, and it is easier to comprehend and understand the core standard posture requirements of balanced arm strength and reasonable forward lean range without the guidance of a human instructor, and it is easier to train and learn the standard posture of CPR movements.

Preferably, the collection angle deviation range of the first optical assembly and the second optical assembly with different collection angles is 30-90 degrees.

The latest published literature article "About the Application of Multimodal System in CPR" also mentioned the monitoring of CPR posture. An intelligent algorithm is designed to monitor the center of gravity change, but this research has obvious limitations. This research is a black-box algorithm obtained by machine learning, and the equipment must be kept as completely consistent as possible, otherwise the experimental results cannot be generalized and applied. For example, by moving the Kinect camera to a different location, or adding or removing a sensor from the current setup, the algorithms derived from this research will no longer apply. What is different from this study is that the present invention first uses an intelligent algorithm to extract the skeletal points of the CPR operator, and then compares it with the standard range obtained in this study. The parameters detected based on the present invention are the arm angle and the angle of the center of gravity, and the method of adding statistics by AI is adopted, so the angle and distance of the camera are not required to be exactly the same in each experiment and future application, as long as they change within a certain range, the results will not be affected. No noticeable effect. Secondly, in the multi-modality study, a single camera was used to collect the pressing posture of the subjects, and the specific distance and angle of the camera were not specified. During the implementation of the research, it was found that the single camera had a blind spot, and at least two angles were needed to collect simultaneously to get more information. The angle is more accurate to collect compression posture data. In addition, trainers in this study do not need to wear any equipment, and are not affected by other equipment. Its convenience, generalization and compatibility are better, and the feasibility of future promotion and application is higher.

Preferably, the collection angle deviation range of the first optical assembly and the second optical assembly at different collection angles is 45 degrees, wherein the first optical assembly collects the first action data of the CPR action in the first angular direction, The first angle direction faces the front of the body of the CPR operator, and the second optical assembly collects second movement data of the CPR action in the second angle direction.

When the collection angle deviation range is 45 degrees, the collected first motion data and second motion data can more accurately calculate the posture angle data of the arms and the center-of-gravity matching angle data, and the collection angle deviation range is 45 degrees. Ideally, calculate the most accurate deviation angle.

Preferably, the posture angle data of the arms and the matching angle data of the center of gravity are analyzed based on a line segment formed by connecting human skeleton points; wherein, the posture angle of the right arm refers to the angle of the right shoulder, the right elbow and the right wrist. The angle formed by the connection line of bone points, the left arm posture angle refers to the angle formed by the connection line of bone points of the left shoulder, left elbow joint and left wrist, the center of gravity matching angle refers to the distance between the moving direction of the center of gravity of the CPR operator and the normal vector of the plane angle.

Preferably, the method of selecting the double-arm posture angle data and the center-of-gravity matching angle data of several CPR action specifications at least includes: selecting the double-arm posture angle data and the center-of-gravity matching angle data with a confidence degree; For the labeling of the indicators, the angle data of the double-arm posture and the matching angle data of the center of gravity of the CPR actions whose labeled indicators are qualified for the CPR actions are selected for distribution statistics.

Preferably, the method further includes: the marking content of the indicators for the CPR action by professionals at least includes: arm straightening and its indicators, center of gravity matching angle and its indicators.

Preferably, the method further includes: before extracting the bone point data, performing data preprocessing on the first motion data collected by the first optical component and the second motion data collected by the second optical component , the data preprocessing method includes data missing value and outlier analysis, data cleaning, feature selection and/or data transformation.

Preferably, the posture angle data of both arms includes posture angle data of the left arm and posture angle data of the right arm, the reasonable range of the posture angle data of the left arm is 169.24°-180°, and the reasonable range of the posture angle data of the right arm is 168.49-180° °, the reasonable range of center of gravity matching angle data is 0-18.46°.

The present invention also provides a processor capable of running a method for determining a standard threshold of cardiopulmonary resuscitation compression postures, the processor is configured to: receive the first motion data collected by the first optical component and all the motion data from different collection angles The second action data collected by the second optical assembly, based on the first action data and the second action data, extracts the bone point data of the human body and at least calculates the posture angle data of the arms and the center-of-gravity matching angle data related to the CPR action , select the unilaterally skewed distribution law of the double-arm posture angle data and the center-of-gravity matching angle data of several CPR action specifications to determine the reasonable range of the double-arm posture angle data and the reasonable range of the center-of-gravity matching angle data.

For the first time, the present invention constructs the standard threshold of cardiopulmonary resuscitation compression posture specially used to identify CPR actions. Compared with other bone extraction algorithms, the present invention extracts CPR actions more accurately, laying an important foundation for future research, especially with artificial intelligence algorithms. .

Description of drawings

Fig. 1 is a schematic diagram of chest compression bone extraction in a preferred embodiment of the present invention;

Fig. 2 is an example diagram of human body key points identified by AlphaPose of a preferred embodiment of the present invention;

Fig. 3 is the confidence statistics table of the extracted human skeleton points of a preferred embodiment of the present invention;

Fig. 4 is a list of main pressing errors and occurrence rates of a preferred embodiment of the present invention;

Fig. 5 is a histogram of right arm and left arm posture angles of a preferred embodiment of the present invention;

Fig. 6 is a histogram of center of gravity matching angles of a preferred embodiment of the present invention;

Fig. 7 is a standard range of pressing posture norms in a preferred embodiment of the present invention;

Fig. 8 is a schematic diagram of the center of gravity matching angle of the CPR compression posture of the present invention;

Fig. 9 is a schematic diagram of the angle of both arms of the CPR compression posture of the present invention;

Figure 10 is a processor of the present invention for implementing standard threshold determination.

List of reference signs

0: nose; 1: cervical spine; 2: right shoulder; 3: right elbow; 4: right wrist; 5: left shoulder; 6: left elbow; 7: left wrist; 8: right hip; 9: right knee; 10: right Ankle; 11: left hip; 12: left knee; 13: left ankle; 14: right eye; 15: left eye; 16: right ear; 17: left ear; 100: acquisition end; 110: ZED camera device; 120: Simulator; 200: data extraction module; 300: preprocessing module; 400: posture detection module; 500: receiving terminal; 60: bone line segment; 61: bone endpoint; 70: actual distribution curve; 71: normal distribution curve; : center of gravity matching angle; α: right arm angle; β: left arm angle.

Detailed ways

A detailed description will be given below in conjunction with the accompanying drawings.

In the present invention, the arm angle refers to the bending angle between the upper arm and the forearm of the arm, as shown in FIG. 9 .

向量

与面法向量之间的夹角为重心匹配角度。The matching angle of the center of gravity refers to the angle between the moving direction of the center of gravity of the CPR operator and the normal vector of the plane. As shown in Figure 8, the midpoint A of the line connecting the right shoulder 2 and the left shoulder 5 moves to the midpoint B of the line connecting the right wrist 4 and the left wrist 7 to generate a vector

vector

The angle between it and the surface normal vector is the matching angle of the center of gravity.

Most of the currently known human bone extraction algorithms are based on natural standing positions or other special sports positions. However, the kneeling position is used during CPR operation, and there is currently no bone extraction algorithm that specifically recognizes CPR actions. The currently known bone extraction algorithms are not accurate enough for CPR action extraction. The bone extraction algorithm is the basis of subsequent CPR with artificial intelligence technology, which is very important. Therefore, the present invention provides a compression posture standard threshold capable of identifying CPR actions and a determination method thereof. Based on the determination method of the standard threshold of cardiopulmonary resuscitation compression posture, a CPR posture detection model can be constructed, and the CPR posture detection model can be applied in CPR training and clinical CPR rescue.

In application, it is only necessary to use optical components capable of taking pictures to collect CPR action images and send them to a processor equipped with a CPR posture detection model. The bone point data of CPR actions can be converted into values in real time and judged to determine whether it is standard or not. analysis results. The present invention can also use the CPR posture detection model to obtain suggestions on how to adjust the CPR action, so that unprofessional personnel can also perform emergency rescue of cardiopulmonary resuscitation in emergency situations.

The method for determining the standard threshold of cardiopulmonary resuscitation compression posture of the present invention, the method at least includes:

S1: The first optical component and the second optical component with different collection angles are used to simultaneously collect CPR actions,

S2: Extracting bone point data of the human body based on the first motion data collected by the first optical component and the second motion data collected by the second optical component;

S3: preprocessing the extracted bone point data;

S4: At least calculate the posture angle data of the arms and the matching angle data of the center of gravity related to the CPR action, and select the unilaterally skewed distribution law of the posture angle data of the arms and the matching angle data of the center of gravity of several CPR action specifications to determine the posture angle data of the arms A reasonable range for the center of gravity and a reasonable range for the matching angle data.

The determination method of the cardiopulmonary resuscitation compression posture standard threshold of the present invention, said method also includes:

At least two professionals marked the CPR action index, and selected the CPR action's double-arm posture angle data and center-of-gravity matching angle data for distribution statistics. As shown in FIG. 10 , the construction device for determining the press posture standard threshold of CPR actions includes at least a collection terminal 100 , a processor and a receiving terminal 500 . The processor establishes a wired or wireless connection with the collection terminal 100 and the receiving terminal to transmit information. Each collection terminal 100 , processor and receiving terminal 500 are respectively provided with independent power lines for power supply.

The collection end 100 includes a first optical component and a second optical component. The first optical component is used for collecting first motion information of a CPR posture with a first coordinate. The first coordinate serves as a reference coordinate system. The second optical component is used to collect the second motion information of the CPR posture with the second coordinates. The first coordinate system is different from the second coordinate system. The first optical component collects the first reference information of the CPR operator in the first coordinate system.

Preferably, the first optical assembly and the second optical assembly simultaneously collect dynamic images of the CPR posture from different angles. Preferably, the collection angle deviation between the first optical component and the second optical component is 45 degrees. Preferably, the collection angle deviation between the first optical component and the second optical component is not limited to 45 degrees, but can also be 30 degrees, 60 degrees and so on. Preferably, the collection angle deviation range between the first optical component and the second optical component is 30-90 degrees. If the collection angle deviation range between the first optical assembly and the second optical assembly is 90 degrees, then the collection angle for collecting the side of the CPR operator is not easy to collect the center of gravity offset vector of the CPR operator. The collection angle deviation between the first optical assembly and the second optical assembly is therefore preferably less than 90 degrees.

Preferably, the first optical component collects the first motion data at a collection angle of zero degrees directly in front of the CPR motion. The second optical component collects the second motion data at a collection angle of 45 degrees lateral to the CPR motion. The skeletal point data collected by the first optical component is used to determine the posture angle data of both arms. The center-of-gravity matching angle data is determined from the bone point data collected by the second optical component.

Alternatively, the first optical assembly collects at a side angle of 45 degrees, and the second optical assembly collects at a frontal angle. The center-of-gravity matching angle data is then determined from the bone point data collected by the first optical component. The skeletal point data collected by the second optical component is used to determine the posture angle data of both arms.

The first optical component includes at least a camera component and a computing component. The second optical component includes at least a camera component and a computing component. The camera component is, for example, the ZED camera device 110 . The calculation component includes a data extraction module 200 for extracting bone point data of the human body based on the AlphaPose algorithm through the CPR action images collected by the camera component. The skeletal point data includes features of a skeletal line segment 60 formed based on the skeletal point and its skeletal end point 61 . The data extraction module 200 sends the skeleton point data to the processor. The data extraction module 200 is a calculator capable of running the AlphaPose algorithm, and an ASIC chip is arranged inside it. The data extraction module 200 is connected to the camera component through a data line to receive and process image data. The data extraction module 200 is connected to the processor through a data bus, so as to extract bone point data from the image data and send it to the processor.

Preferably, the data extraction module 200 of the present invention may not be set in the optical assembly, but set in the processor to become a part of the processor.

As shown in FIG. 1 , when the operator of the CPR action is performing the CPR action on the simulated human 120 , the data extraction module 200 can extract the operator's time-related skeleton endpoint 61 and skeleton line segment 60 .

As shown in FIG. 2 , the bone points extracted by the data extraction module 200 include at least 18 main parts. Skeleton points mainly include nose 0, cervical spine 1, right shoulder 2, right elbow joint 3, right wrist 4, left shoulder 5, left elbow 6, left wrist 7, right hip 8, right knee 9, right ankle 10, left hip 11, left Knee 12, left ankle 13, right eye 14, left eye 15, right ear 16 and left ear 17.

The data extraction module 200 also performs confidence statistics on the skeleton point data. As shown in FIG. 9 , the average confidence statistical data of each human bone point collected by the first optical assembly and the second optical assembly.

The data extraction modules 200 of the two optical components respectively send the extracted bone data to the processor.

The processor may be one of a server, a remote server, and an ASIC. The processor is used for performing the analysis step and the statistical step of the preprocessed bone point data. Preferably, the processor may be a combined device of at least two dedicated integrated chips or CPU processors, and the processor may also be a separate dedicated integrated chip or CPU capable of running the data preprocessing module program and the gesture detection module program. ASIC or CPU can be applied in the form of server or cloud server.

The processor includes at least a preprocessing module 300 and a posture detection module 400 . Wherein, both the preprocessing module 300 and the posture detection module 400 may be dedicated integrated chips or separate hardware modules of a CPU processor. When the preprocessing module 300 and the posture detection module 400 are integrated on the same ASIC or CPU processor, the preprocessing module 300 and the posture detection module 400 are running programs with the processor as the hardware carrier.

The processor is also provided with a first data transmission port and a second data transmission port. In the case that the preprocessing module 300 and the posture detection module 400 are ASICs or CPU hardware modules respectively, the preprocessing module 300 is connected to the first data transmission port through a data transmission line. The preprocessing module 300 and the posture detection module 400 are connected through a data transmission line. The posture detection module 400 is connected to the second data transmission port through a data transmission line. The first data transmission port and the second data transmission port may be wired data transmission port components or wireless data transmission port components respectively, which depends on whether the set data transmission mode is wired transmission or wireless transmission. The wired data transmission port components are, for example, various types of USB transmission line ports. The wireless data transmission port component is, for example, a Bluetooth data transmission communication component, a WIFI data transmission communication component, a ZigBee data transmission communication component, and the like.

Preferably, the processor may only be provided with one data transmission port, which is connected in parallel with the preprocessing module 300 and the posture detection module 400 for sending and receiving data to the preprocessing module 300 and the posture detection module 400 respectively.

Preferably, the preprocessing module 300 also performs data preprocessing on the received skeleton point data.

Data preprocessing steps include at least:

S31: Data missing value and outlier analysis;

S32: data cleaning;

S33: feature selection;

S34: data conversion.

The preprocessing module 300 sends the preprocessed data to the posture detection module 400 through the transmission line. Preferably, the gesture detection module 400 is capable of running a gesture detection model. The posture detection model can calculate the posture angle data of the arms and the matching angle data of the center of gravity.

The posture detection module 400 at least calculates the double-arm posture angle data of the double-arm posture of the CPR operator based on the skeletal point data collected by the first optical component. Specifically, the bone point data collected by the first optical component includes at least the right shoulder 2 , the right elbow joint 3 , the right wrist 4 , the left shoulder 5 , the left elbow 6 and the left wrist 7 . The confidence levels for right shoulder 2, right elbow 3, right wrist 4, left shoulder 5, left elbow 6, and left wrist 7 are 0.94, 0.89, 0.93, 0.95, 0.90, and 0.87, respectively. As shown in Figure 9, the arm posture angle refers to the angle between the hand, elbow and shoulder, that is, the right arm posture angle is the angle α formed between the right shoulder 2, the right elbow joint 3 and the right wrist 4, and the left arm posture The angle is the angle β formed between the left shoulder 5 , the left elbow 6 and the left wrist 7 .

向量

与面法向量之间的夹角为重心匹配角度γ。The posture detection module 400 at least calculates the center of gravity matching angle data of the CPR operator based on the skeletal point data collected by the second optical component. Here the bone point data at least includes right shoulder 2 , left shoulder 5 , right wrist 4 and left wrist 7 . As shown in Fig. 9, the confidence levels of the bone point data of the right shoulder 2, the left shoulder 5, the right wrist 4 and the left wrist 7 are 0.91, 0.81, 0.89, and 0.88, respectively. As shown in Fig. 8, the matching angle of the center of gravity refers to the angle at which the moving direction of the center of gravity of the CPR implementer is perpendicular to the patient. As shown in Figure 10, the midpoint A of the line connecting the right shoulder 2 and the left shoulder 5 moves to the midpoint B of the line connecting the right wrist 4 and the left wrist 7 to generate a vector

vector

The angle between it and the surface normal vector is the center-of-gravity matching angle γ.

The posture detection module 400 calculates the included angle formed by the bone line segments based on the preset included angle calculation formula, or calculates the included angle between the vector and the normal vector.

The formula for calculating the included angle is:

m ₁ represents the slope of the first straight line and m ₂ represents the slope of the second straight line.

If the first straight line is defined by the points P ₁ =[x ₁ ,y ₁ ] and P ₂ =[x ₂ ,y ₂ ], then

ε为10^-9。The formula for calculating the slope m is:

ε is 10 ^-9 .

In the prior art, the latest research on the application of multimodal systems in CPR also mentioned the monitoring of CPR posture. An intelligent algorithm is designed to monitor the center of gravity change, but this research has obvious limitations. This research is a black-box algorithm obtained by machine learning, and the equipment must be kept as completely consistent as possible, otherwise the experimental results cannot be generalized and applied. For example, by moving the Kinect camera to a different location, or adding or removing a sensor from the current setup, the algorithms derived from this research will no longer apply. What is different from this study is that the present invention first uses an intelligent algorithm to extract the skeletal points of the CPR operator, and then compares it with the standard range obtained in this study. The parameters detected based on the present invention are the arm angle and the angle of the center of gravity, and the method of adding statistics by AI is adopted, so the angle and distance of the camera are not required to be exactly the same in each experiment and future application, as long as they change within a certain range, the results will not be affected. No noticeable effect. Secondly, in the multi-modality study, a single camera was used to collect the pressing posture of the subjects, and the specific distance and angle of the camera were not specified. During the implementation of the research, it was found that the single camera had a blind spot, and at least two angles were needed to collect simultaneously to get more information. The angle is more accurate to collect compression posture data. In addition, trainers in this study do not need to wear any equipment, and are not affected by other equipment. Its convenience, generalization and compatibility are better, and the feasibility of future promotion and application is higher. Not only that, the AI model in the prior art directly judges the angle of each posture of the arms through the action image of the CPR posture, which will be affected by the appearance characteristics of the CPR operator's body, such as fat and thin characteristics, and clothing occlusion characteristics. The judgment of CPR posture is biased and cannot be eliminated. The AI model in the prior art is not accurate enough to judge the CPR posture assessment, and the error is relatively large.

In the prior art, the center of gravity of the CPR posture cannot be determined due to the influence of human body characteristics and clothing occlusion, so the center of gravity deviation of the CPR posture cannot be assessed and judged, nor can it be judged whether the center of gravity deviation of the CPR operator is It is correct, and it is impossible to further judge whether the force of the CPR operator is correct.

The posture detection module of the present invention does not obtain angle data through a deep learning algorithm, but obtains angle data through specific calculation formulas. No matter how the collection end is set, as long as the angle of the collection end is reasonable, the accurate angle data of the CPR operator can be calculated and obtained. Therefore, the location of the collection terminal in the present invention is relatively random, the setting conditions are simple, and the location of the collection terminal will not affect the recognition progress of the posture detection module. That is, in the actual application scene, the change of the collection angle of the collection end of the present invention has relatively little influence on the angle result, and the setting range of the collection end is relatively wide.

Not only that, the posture detection model of the present invention extracts and calculates the posture angle data and center-of-gravity matching data of the arms through the skeleton points of the human body. The skeletal data composed of human skeletal points will not be affected by the fat and thin of the human body and the occlusion of clothing, so the calculation and judgment of the angle of the CPR posture will be more accurate.

Because the present invention uses bone point data to calculate, it gets rid of the rigidity of differences in human body characteristics and clothing. Although the center of gravity of the action cannot be directly determined, it can be formed by the midpoint of the line connecting the shoulders and the midpoint of the line connecting the two wrists. Vector to determine the center of gravity offset direction. According to the angle between the vector and the plane normal vector, it can be determined whether the CPR operator's effort is balanced and appropriate. If the matching angle of the center of gravity is reasonable, then the CPR operator's force is balanced and appropriate.

Preferably, the processor is connected to at least one terminal in a preferred or wireless manner. Specifically, the posture detection module 400 is connected to at least one terminal through a line or wireless signal, so as to send the video data of the CPR operator's action image, the calculated posture angle data of the arms, and the center of gravity matching angle data to the terminal. The terminal is used to display the CPR operator's motion image, the posture angle data of the arms and the center of gravity matching angle data to at least one expert. The terminal includes at least a display component, an interactive component and an information storage component. That is, a terminal is an electronic device that allows interaction. Terminals are, for example, electronic devices such as tablet computers, iPads, notebook computers, desktop computers, smart phones, smart watches, and smart glasses.

Preferably, on the display screen of the terminal, the posture angle data of the arms and the center-of-gravity matching angle data are displayed in a manner that does not block the actions of the CPR operator.

The terminal is used by professionals familiar with the standard of CPR practice. Preferably, one terminal is assigned to one professional. Preferably, based on the principles of scientific statistics, the professional staff is best composed of three persons. Motion images of CPR operators were individually annotated by three professionals. Professionals are annotated based on specified indicator content. The content of the index includes at least two items: arm straightening and its index, center of gravity matching angle and its index.

The index of arm straightness refers to the judgment of whether the arm posture is correct during CPR. The index of the matching angle of the center of gravity refers to whether the moving direction of the center of gravity of the CPR implementer is perpendicular to the patient. During the operation of the CPR operator, the compression errors of the CPR posture mainly include: wrist force, fingers not lifted, center of gravity deviation (including base center of gravity skew, center of gravity moving back and forth, center of gravity moving left and right), elbow bending, etc. Among them, wrist force, fingers not raised, center of gravity deviation (including base center of gravity skew, center of gravity moving back and forth, center of gravity moving left and right), and elbow bending are the errors with the highest incidence.

Therefore, the present invention also conducts professional labeling of CPR actions based on labeling indicators by viewing CPR action images by professionals, and excludes non-standard data. The labeling indicators include at least whether the arms are straight and the center of gravity is correct.

The terminal sends the action images marked by professionals to the gesture detection module 400 through the second data transmission port of the processor. The posture detection module 400 receives the action image with annotation information, and uses the CPR posture angle data and center-of-gravity matching angle data of the arms conforming to the arm straightening and its index, the center of gravity matching angle and its index as standard data. Preferably, when the motion image collected by the acquisition end is a professional group composed of professionals, the posture angle data of both arms and the center-of-gravity matching angle data of the CPR posture that meet each index are the standard data of the professional group. The professional group normative data serves as a data set for standard development of cardiopulmonary resuscitation. The posture detection module 400 receives the action image with tag information, and regards the CPR posture angle data and center-of-gravity matching angle data of arms that do not conform to arm straightening and its index, center of gravity matching angle and its index as irregular data.

The posture detection module 400 makes statistics on the filtered professional group specification data.

Specifically, as shown in Figure 4, a total of 28,800 sets of human skeleton point coordinate data were collected from the professional group normative data set. A total of 7,200 sets of human skeleton point coordinate data were collected from the non-professional group dataset. In FIGS. 5 and 6 , the thick curve represents the actual distribution curve 70 . The thin curve represents the normal distribution curve 71 . It can be seen that the posture angle data of both arms and the matching angle data of the center of gravity conform to the skewed distribution. The 5% quantile is taken as the normal value range for the left and right arm posture angle data, and the 95% quantile is taken as the normal value range for the center of gravity matching angle data.

Specifically, the measurement data are described by mean ± standard deviation, and the comparison of means between groups is carried out by independent sample t test. Because the arm angle is a unilaterally skewed distribution data, the 5% to 10% percentile is used to calculate the reasonable range boundary value. Similarly, the center of gravity matching angle range is a unilaterally skewed distribution data, and the 90% to 95% percentile is taken Calculate the reasonable range boundary value. All statistical analyzes will be performed at a two-sided 0.05 level of significance.

As shown in Figure 7, in the case of taking the 5% percentile, the reasonable range of the left arm posture angle is 169.24-180 degrees, and the reasonable range of the right arm posture angle is 168.49-180 degrees. In the case of taking the 95% percentile, the reasonable range of the center of gravity matching angle is 0-18.46 degrees. This is also the standard threshold value of the cardiopulmonary resuscitation compression posture obtained by the present invention.

Preferably, when the CPR compression posture standard threshold of the present invention is applied to CPR action monitoring or quality control, the CPR action detection model or quality control model formed by it can be continuously optimized.

As shown in Figure 10, in the posture detection module 400, the CPR action data and standard threshold data that meet the standard threshold of cardiopulmonary resuscitation compression posture and the manual labeling index are used as the standard data, and the posture detection model is constructed based on the machine learning model, which can also be used as the posture Quality control model.

For the pose detection model or pose quality control model, the steps can also be implemented:

S5: Model optimization.

The steps of model optimization include at least:

S51: Establish a 3D moving point model through standard actions of professionals; convert the 3D moving point model into a second 2DCPR action with the same acquisition angle as the collected first 2DCPR action; combine the first detection result of the first 2DCPR action with the second Compare the second detection result of the two 2DCPR actions; if the first detection result is consistent with the second detection result, the posture detection model or the posture quality control model does not need to be optimized. If the first detection result is inconsistent with the second detection result, the posture detection model or the posture quality control model needs to be optimized.

It should be noted that the above specific embodiments are exemplary, and those skilled in the art can come up with various solutions inspired by the disclosure of the present invention, and these solutions also belong to the scope of the disclosure of the present invention and fall within the scope of this disclosure. within the scope of protection of the invention. Those skilled in the art should understand that the description and drawings of the present invention are illustrative rather than limiting to the claims. The protection scope of the present invention is defined by the claims and their equivalents. The description of the present invention contains a number of inventive concepts, such as "preferably", "according to a preferred embodiment" or "optionally" all indicate that the corresponding paragraph discloses an independent concept, and the applicant reserves the right to propose a division based on each inventive concept right to apply.

Claims (10)

1. A method for determining a cardiopulmonary resuscitation compression posture standard threshold, characterized in that the method at least includes: Using the first optical component and the second optical component with different collection angles to simultaneously collect CPR actions, Based on the first motion data collected by the first optical component and the second motion data collected by the second optical component, extract the bone point data of the human body and at least calculate the posture angle data of the arms and the center of gravity matching related to the CPR motion angle data, The unilaterally skewed distribution law of the double-arm posture angle data and the center-of-gravity matching angle data of several CPR action specifications is selected to determine the reasonable range of the double-arm posture angle data and the reasonable range of the center-of-gravity matching angle data. 2. the determination method of cardiopulmonary resuscitation press posture standard threshold value according to claim 1, is characterized in that, Based on the unilateral skewed distribution law of the posture angle data of the arms, the 5% percentile is selected to determine the reasonable range of the posture angle data of the arms. Based on the one-sided skewed distribution law of the center of gravity matching angle data, the 95% percentile is selected to determine the reasonable range of the center of gravity matching angle data. 3. The method for determining the standard threshold of cardiopulmonary resuscitation compression posture according to claim 1 or 2, wherein the collection angle deviation range of the first optical assembly and the second optical assembly at different collection angles is 30-90° Spend. 4. The method for determining the standard threshold of cardiopulmonary resuscitation compression posture according to any one of claims 1 to 3, wherein the collection angle deviation range of the first optical assembly and the second optical assembly at different collection angles is 45 degrees, where, The first optical component collects the first movement data of the CPR movement in a first angular direction, and the first angular direction faces the front of the body of the operator of the CPR movement, The second optical assembly collects second motion data of a CPR motion in a second angular direction. 5. The method for determining the standard threshold of cardiopulmonary resuscitation compression posture according to any one of claims 1 to 4, wherein the posture angle data of both arms and the matching angle data of the center of gravity are obtained based on the connection of human skeleton points. The formed line segment is analyzed; among them, The right arm posture angle refers to the angle formed by the line connecting the bone points of the right shoulder, right elbow joint and right wrist, The left arm posture angle refers to the angle formed by the line connecting the bone points of the left shoulder, left elbow joint and left wrist, The matching angle of the center of gravity refers to the angle between the moving direction of the center of gravity of the CPR operator and the normal vector of the plane. 6. The method for determining the standard threshold of cardiopulmonary resuscitation compression posture according to any one of claims 1 to 5, wherein the method of selecting a number of CPR action specifications for both arm posture angle data and center of gravity matching angle data at least includes: Select the arms posture angle data and center of gravity matching angle data with qualified confidence; At least two professionals marked the CPR action index, and selected the CPR action's double-arm posture angle data and center-of-gravity matching angle data for distribution statistics. 7. The method for determining the standard threshold of cardiopulmonary resuscitation compression posture according to any one of claims 1 to 6, wherein the method further comprises: The marking content of the indicators for the CPR action by professionals at least includes: arm straightening and its indicators, center of gravity matching angle and its indicators. 8. The method for determining the compression posture standard threshold for cardiopulmonary resuscitation according to any one of claims 1 to 7, wherein the method further comprises: performing data preprocessing on the first motion data collected by the first optical component and the second motion data collected by the second optical component before extracting the bone point data, The data preprocessing method includes data missing value and outlier value analysis, data cleaning, feature selection and/or data transformation. 9. The method for determining the standard threshold of cardiopulmonary resuscitation compression posture according to any one of claims 1 to 7, wherein the posture angle data of both arms comprises posture angle data of the left arm and posture angle data of the right arm, The reasonable range of the left arm posture angle data is 169.24°-180°, and the reasonable range of the right arm posture angle data is 168.49-180°, The reasonable range of center of gravity matching angle data is 0-18.46°. 10. A processor capable of running the method for determining the standard threshold of cardiopulmonary resuscitation compression posture, characterized in that the processor is configured to: receiving the first motion data collected by the first optical component and the second motion data collected by the second optical component at different collection angles, Extracting the skeleton point data of the human body based on the first motion data and the second motion data and at least calculating the posture angle data of the arms and the center-of-gravity matching angle data related to the CPR motion, The unilaterally skewed distribution law of the double-arm posture angle data and the center-of-gravity matching angle data of several CPR action specifications is selected to determine the reasonable range of the double-arm posture angle data and the reasonable range of the center-of-gravity matching angle data.

Images