Suction Detection Method and Device for Ventricular Assist Device

Abstract

The present application provides a method and device for detecting aspiration of a ventricular assist device. The method includes: obtaining a target pumping flow curve, which is the pumping flow curve of the ventricular assist device during the first cycle of operation; extracting a target feature set from the target pumping flow curve; inputting the target feature set into a target classification tree model to obtain a classification result; and when the classification result indicates aspiration, calculating the degree of aspiration of the ventricular assist device. The present application uses a classification tree model to judge aspiration of the pumping flow curve of the ventricular assist device, which can quickly detect whether the current ventricular assist device has aspiration, improve the accuracy of aspiration detection, and further judge the degree of aspiration when there is aspiration, enabling timely reminder to medical staff to take corresponding solutions and improving the safety of patients.

Description

Technical Field

[0001] This application relates to the technical field of medical devices, and particularly to a method and device for detecting aspiration of a ventricular assist device. Background Art

[0002] Currently, ventricular assist devices (VADs) have become an important means for treating end-stage heart failure diseases. They are artificial mechanical devices that draw fluid from the venous system or the heart and directly introduce it into the arterial system, partially or fully replacing the work of the ventricles, and can solve the problem of shortage of heart donors. To meet the circulatory needs of patients, developing a suitable pump control system to regulate the pumping flow through the ventricular assist device by controlling the pump speed is an important challenge faced by the increasing use of these devices. However, when the speed of the ventricular assist device exceeds what the user needs, it may cause aspiration problems in the ventricular assist device and damage the patient's heart. Summary of the Invention

[0003] Embodiments of this application provide a method and device for detecting aspiration of a ventricular assist device, which can timely and accurately detect whether there is aspiration in the ventricular assist device and the degree of aspiration.

[0004] In a first aspect, embodiments of this application provide a method for detecting aspiration of a ventricular assist device, the method comprising:

[0005] Obtain a target pumping flow curve, where the target pumping flow curve is the pumping flow curve of the ventricular assist device during a first cycle of operation;

[0006] Extract a target feature set of the target pumping flow curve;

[0007] Input the target feature set into a target classification tree model to obtain a classification result;

[0008] When the classification result is aspiration, calculate the degree of aspiration of the ventricular assist device.

[0009] In a second aspect, a control unit of a ventricular assist device provided by embodiments of this application, the control unit comprising one or more processors, and the one or more processors are configured to:

[0010] Obtain a target pumping flow curve, where the target pumping flow curve is the pumping flow curve of the ventricular assist device during a first cycle of operation;

[0011] Extract a target feature set of the target pumping flow curve;

[0012] Input the target feature set into a target classification tree model to obtain a classification result;

[0013] When the classification result is suction, calculate the suction degree of the ventricular assist device.

[0014] In a third aspect, an embodiment of the present application provides a ventricular assist device, which includes:

[0015] A housing;

[0016] An impeller disposed in the housing;

[0017] A motor that drives the impeller to rotate in a suspended manner;

[0018] A control unit connected to the motor, and the control unit is used to execute the instructions of the steps in the method described in the first aspect above.

[0019] In a fourth aspect, an embodiment of the present application provides a medical device, which includes a processor, a memory, a communication interface, and one or more programs. The one or more programs are stored in the memory and are configured to be executed by the processor. The programs include instructions for executing some or all of the steps described in the method described in the first aspect above.

[0020] In a fifth aspect, an embodiment of the present application provides a computer-readable storage medium that stores a computer program for electronic data exchange, wherein the computer program causes a computer to execute some or all of the steps described in the method described in the first aspect above.

[0021] In a sixth aspect, an embodiment of the present application provides a computer program product, wherein the computer program product includes a non-transitory computer-readable storage medium storing a computer program, and the computer program is operable to cause a computer to execute some or all of the steps described in the method described in the first aspect of the embodiments of the present application. The computer program product can be a software installation package.

[0022] The technical solution provided by the present application is to obtain a target pumping flow curve, which is the pumping flow curve of the ventricular assist device during the first cycle of operation; extract the target feature set of the target pumping flow curve; input the target feature set into the target classification tree model to obtain a classification result; when the classification result is suction, calculate the suction degree of the ventricular assist device. The present application uses a classification tree model to judge suction for the pumping flow curve of the ventricular assist device, which can quickly detect whether the current ventricular assist device has suction, improve the accuracy of suction detection, and further judge the degree of suction when there is suction, and can timely remind medical staff to take corresponding solutions, improving the safety of patients. Description of the Drawings

[0023] To more clearly illustrate the technical solutions in the embodiments of the present application, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings in the following description are some embodiments of the present application. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0024] Figure 1 It is a schematic diagram of a ventricular assist system provided by an embodiment of the present application;

[0025] Figure 2 It is a schematic structural diagram of a ventricular assist device provided by an embodiment of the present application;

[0026] Figure 3 It is a schematic diagram of a classification tree model provided by Embodiment 1 of the present application;

[0027] Figure 4 It is a schematic flowchart of a suction detection method for a ventricular assist device provided by Embodiment 2 of the present application;

[0028] Figure 5 It is a partial schematic diagram of a classification tree model provided by Embodiment 2 of the present application;

[0029] Figure 6 It is a schematic structural diagram of a medical device provided by an embodiment of the present application. Detailed implementation manners

[0030] For those in the technical field to better understand the technical solutions of the present application, the following clearly and completely describes the technical solutions in the embodiments of the present application in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only some embodiments of the present application, rather than all embodiments. Based on the description of the embodiments of the present application, all other embodiments obtained by those skilled in the art without creative efforts fall within the scope protected by the present application.

[0031] The terms "first", "second", etc. in the specification, claims and drawings of the present application are used to distinguish different objects, rather than to describe a specific order. In addition, the terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, software, product or device that includes a series of steps or units is not limited to the listed steps or units, but also includes unlisted steps or units, or other steps or units inherent to these processes, methods, products or devices.

[0032] References to "embodiments" in this specification mean that a particular feature, structure, or characteristic described in connection with the embodiments can be included in at least one embodiment of the present application. The phrase occurs in various places in the specification and is not necessarily referring to the same embodiment, nor is it an independent or alternative embodiment mutually exclusive of other embodiments. Those skilled in the art will understand explicitly and implicitly that the embodiments described herein can be combined with other embodiments.

[0033] The medical device involved in the present application can be a ventricular assist device, such as an implantable ventricular assist device. The ventricular assist device can be used for the left heart, right heart, or both hearts. The ventricular assist device can include at least one blood pump, and the blood pump can be a magnetic levitation pump.

[0034] The "rotational speed" in the present application refers to the rotational speed of a motor or an electric machine, which is associated with the rotational speed of the rotor or impeller of the ventricular assist device and can be defined as revolutions per minute. "Flow rate", "fluid flow rate", "pumping flow rate" refer to the volume of fluid transported through the ventricular assist device per unit time, which can be estimated and measured in liters per minute.

[0035] Patients with heart failure can have a ventricular assist device implanted to assist the heart in performing the blood pumping function. The ventricular assist device can be disposed in the patient's left ventricle to pump blood from the patient's left ventricle to the aorta to solve the patient's left heart failure problem; the ventricular assist device can also be disposed in the patient's right ventricle to pump blood from the patient's right ventricle to the pulmonary artery to solve the patient's right heart failure problem.

[0036] The ventricular assist system of the present application can include an LVAD (left ventricular assist device), an RVAD (right ventricular assist device), or a BIVAD (biventricular assist device). These systems not only include the ventricular assist device implanted in the patient during operation, but also generally include a controller disposed outside the patient and connected to the ventricular assist device through a percutaneous line (drive system). A part of the drive system extends outside the patient between the controller and the puncture site, and another part extends inside the patient between the puncture site and the ventricular assist device. The controller can include, for example, an integrated battery (rechargeable battery) or can be connected to a battery such that the implanted ventricular assist device can be powered by the controller through a line passing through the skin. The ventricular assist device generally includes a motor having a stator and a rotor with blades. The motor of the ventricular assist device can generally be driven by the power delivered by the controller. For example, a current is generated in the windings of the stator, which causes the rotor and its blades to rotate in order to pump the patient's blood.

[0037] Please refer to Figure 1 , Figure 1A ventricular assist system provided by an embodiment of the present application. The ventricular assist system includes a ventricular assist device 100, a controller 200, and a percutaneous cable 300 that connects the ventricular assist device 100 to the controller 200. The proximal end of the percutaneous cable 300 passes through the skin and extends into the ventricular assist device 100 to transmit the power, information, and control signals for the operation of the ventricular assist device 100.

[0038] The ventricular assist device 100 can be implanted in the body. For example, it can be attached to the heart via a ventricular connection assembly (such as a top ring, ventricular cuff, ventricular sleeve). The ventricular connection assembly can be sutured to the heart and connected to the ventricular assist device 100. The other end of the ventricular assist device 100 can be connected to the ascending aorta or pulmonary artery via an outlet tube and / or an artificial blood vessel connected to the outlet tube. In this way, the ventricular assist device 100 can effectively transfer blood from the weakened heart and pump it into the aorta or pulmonary artery, so as to circulate to the remaining part of the patient's vascular system and provide ventricular assist function for the patient.

[0039] The controller 200 is used to monitor the ventricular assist device 100, and it can implement functions such as control, data display, fault detection, alarm, and data recording of the ventricular assist device 100. For example, the controller 200 can have a touch screen display for displaying the operation data of the ventricular assist device 100, patient information, information of the ventricular assist device 100, etc. Further, the user can set the operation parameters of the ventricular assist device 100 through the controller 200.

[0040] Please refer to Figure 2 , Figure 2 FIG. is a schematic structural diagram of a ventricular assist device 100 proposed by an embodiment of the present application. The ventricular assist device 100 includes a housing assembly having an inlet tube 50, an impeller 20 for pushing fluid, and a motor 30 for driving the impeller 20 to rotate in a suspended manner. The housing assembly includes a first housing and a second housing connected to the first housing. The first housing and the second housing enclose a chamber 10. The housing assembly is also respectively provided with a fluid inlet 14 and a fluid outlet 15 communicating with the chamber 10, and the fluid inlet 14 is provided on the first housing. The impeller 20 can rotate in a suspended manner within the chamber 10. The rotation of the impeller 20 can generate a centrifugal force for transporting fluid, so that the fluid can enter the chamber 10 from the fluid inlet 14 and be output from the fluid outlet 15. Among them, the suspended rotation of the impeller 20 means that the impeller 20 does not contact the chamber wall of the chamber 10 during rotation.

[0041] The second housing includes a first side wall 11, the first housing includes a second side wall 12, and the motor 30 includes a stator 31 and a rotor 32 which are arranged on both sides of the first side wall 11. Among them, the stator 31 is fixed on the outer side of the first side wall 11 relative to the chamber 10, and correspondingly, the rotor 32 is located inside the chamber 10. Further, the rotor 32 is fixedly connected to the impeller 20. When the stator 31 drives the rotor 32 to rotate inside the chamber 10, the impeller 20 also rotates synchronously with the rotor 32 inside the chamber 10.

[0042] The ventricular assist device 100 further includes a control unit 33. The control unit 33 is electrically connected to the stator 31, and the control unit 33 can control the rotation speed and suspension height of the rotor 32 by adjusting the current flowing through the stator 31.

[0043] Taking left ventricular assist as an example, the inlet pipe 50 is fixed at the apex position. During the implantation process or operation process of the ventricular assist device 100, the orifice of the inlet pipe 50 may be too close to or adjacent to the heart tissue, resulting in a suction event. When the fluid inlet 14 interacts with the heart tissue, causing partial or complete blockage of the inlet pipe 50, a suction event may occur. Continuous suction may damage the patient's heart, impair the function of the ventricular assist device 100, and cause insufficient perfusion of the patient.

[0044] Based on this, the present application proposes a method for detecting suction of the ventricular assist device 100. By inputting the pumping flow curve of the ventricular assist device 100 into a pre-trained classification tree model, it is determined whether there is suction in the current ventricular assist device 100 according to the output classification result. Furthermore, when the classification result is suction, the suction degree is further judged, so that it can be quickly detected whether the ventricular assist device 100 has suction, the accuracy of suction detection is improved, and the suction degree is judged, and corresponding treatment can be made in time according to the suction degree.

[0045] Combined with the above description, the method provided by the present application will be described from the model training side and the model application side below:

[0046] The method for training the classification tree model provided by the embodiments of the present application involves the processing of natural language and computer vision, and can be specifically applied to data processing methods such as data training, machine learning, and deep learning. It performs symbolic and formal intelligent information modeling, extraction, preprocessing, training, etc. on the training data (such as the pumping flow of the ventricular assist device in the present application), and finally obtains a trained classification tree model.

[0047] It should be noted that the method for training the classification tree model provided by the embodiments of the present application and the method for using the classification tree model to determine whether there is suction are inventions generated based on the same concept, and can also be understood as two parts of a system, or two stages of an overall process: such as the model training stage and the model application stage.

[0048] Example 1

[0049] This embodiment of the present application describes the structural characteristics of the classification tree model. For example, as Figure 3 shown, the inputs of the classification tree model are the first feature P1, the second feature P2, the third feature P3, and the fourth feature P4. The leaf nodes of the tree are the classification results output by the model, and the classification results are divided into suction and normal. Suction is represented by y = 1, and normal is represented by y = 0. The nodes of the tree are any one of the first feature P1, the second feature P2, the third feature P3, and the fourth feature P4. Each layer constructs classification rules using the first feature P1, the second feature P2, the third feature P3, and the fourth feature P4 as splitting points, and the classification rules make the classification results output by the classification tree model close to the true categories of the data.

[0050] The first feature P1 is used to measure the symmetry degree of the data and is a statistic describing the symmetry of the data distribution. The fourth feature P4 is used to measure the distribution shape of the data. In statistics, the fourth feature can be represented by kurtosis, and kurtosis minus 3 is used to measure the degree of peakedness compared with the normal distribution. If the data distribution is uniform or more dispersed in the center and lacks extreme values, its kurtosis value is low; if the data is more inclined to be concentrated near the mean and has more extreme values at the same time, its kurtosis value is high.

[0051] Among them, the first feature P1 can be calculated through the first formula, and the fourth feature P4 can be calculated through the second formula. The first formula is:

[0052]

[0053] The second formula is:

[0054]

[0055] Among them, is the data sampling point, is the average value of n data sampling points, and S is the standard deviation of n data sampling points.

[0056] The second feature P2 is the ratio of the low-frequency energy to the fundamental frequency energy in the spectrum of the data, and the third feature P3 is the ratio of the high-frequency energy to the fundamental frequency energy in the spectrum of the data. In the spectrum data PF(w), where the frequency range (0, a) is the low-frequency range, the frequency range (a, b) is the fundamental frequency range, and the frequency range (b, is the high-frequency range. By summing the spectral data PF(w) in the fundamental frequency range, low-frequency range, and high-frequency range respectively, the low-frequency energy, fundamental frequency energy, and high-frequency energy can be obtained. The low-frequency energy is expressed as N1 = PF(0) + … + PF(a), the fundamental frequency energy is expressed as N1 = PF(a) + … + PF(b), and the high-frequency energy is expressed as N1 = PF(b) + … + PF( ). Therefore, the second feature P2 = N1 / N2, and the third feature P3 = N3 / N3.

[0057] Input the extracted first feature P1, second feature P2, third feature P3, and fourth feature P4 into the classification tree model, and classify them layer by layer according to the classification rules, and finally output the classification result.

[0058] Exemplarily, the classification tree model can be a classification and regression tree model, a decision tree model, a random forest model, etc.

[0059] The classification and regression tree model consists of multiple nodes, and the number of nodes is positively correlated with the complexity of the model. Nodes include a root node, several intermediate nodes, and multiple leaf nodes (the root node is the starting node of the decision tree, representing the initial splitting point of the entire dataset. The leaf node is the terminal node of the decision tree, representing the final classification result and no further splitting. The intermediate node is a non-terminal node in the decision tree, responsible for further splitting the dataset according to features until the leaf node is finally reached). Each node represents a decision condition for a feature, and each non-leaf node has two child nodes, representing the result of the data being assigned to two different subsets according to the judgment of a certain feature value.

[0060] At each node, a feature and a certain threshold of the feature are selected to divide the dataset into two subsets. The basis for selecting the feature and a certain threshold of the feature is to measure the quality of each split by evaluating the Gini index. This criterion is used to evaluate whether the subsets after splitting are more "pure", that is, the more concentrated the class distribution of the samples in the subset, the more conducive to improving the classification accuracy.

[0061] The calculation formula of the Gini coefficient is: , where pi is the proportion of samples belonging to the i-th class in the dataset, and c is the number of classes. The smaller the Gini index, the higher the purity of the dataset. An intermediate node will be divided into two branches, and the Gini coefficients of the left and right branches are calculated respectively, and then the weighted average of these two Gini coefficients is used as the Gini coefficient of the intermediate node. The final selected feature and a certain threshold of the feature are used as the basis for the intermediate node because the Gini coefficient calculated by them is the smallest.

[0062] For example, assume that when the first feature P1 > 0.234, the left branch divided includes 60 groups of feature arrays, among which 40 are normal and 20 are aspirated; the right branch includes 80 groups of feature arrays, among which 50 are normal and 30 are aspirated. Then the Gini coefficient of the left branch is equal to 1 - (40 / 60)^2 - (20 / 60)^2 = 0.44444; the Gini coefficient of the right branch is equal to 1 - (50 / 80)^2 - (30 / 80)^2 = 0.46875; the Gini coefficient of the intermediate node is the weighted average of the two: 0.44444 (60 / 140) + 0.46875 (60 / 140) = 0.39136.

[0063] To select the best splitting point, the model calculates the Gini index for all possible splitting points of each feature and selects the feature and threshold that maximize the purity.

[0064] Once the feature and threshold are selected, all samples are assigned to the two branches according to this condition. This process continues recursively until a leaf node is reached. The leaf node will output a class label. In a classification task, the class label of the leaf node is the majority class of the samples within that leaf node. That is, among all the samples that reach this leaf node, the class with the most occurrences of the class label will be used as the prediction result.

[0065] Example Two

[0066] Please refer to Figure 4 , Figure 4 which is a schematic flow diagram of a method for detecting aspiration of a ventricular assist device provided in an embodiment of the present application, and is applied to the ventricular assist device 100 as shown in Figures 1-2 . As shown in Figure 4 , this method includes the following steps.

[0067] S410. Obtain a target pumping flow curve, where the target pumping flow curve is the pumping flow curve of the first cycle of operation of the ventricular assist device.

[0068] During the operation of the ventricular assist device 100 in a user's body, the pumping flow rate through the ventricular assist device 100 depends on the work done by the ventricular assist device 100 to overcome resistance and pump blood from the left ventricle to the aorta. The work done by the ventricular assist device 100 can be quantified as the magnitude of the current supplied to the motor 30 (more specifically, the stator 31), that is, the motor current corresponds to the amount of current delivered to the motor 30 of the ventricular assist device 100 when the ventricular assist device 100 is operating in the user. During different stages of the cardiac cycle of the user's heart, the load on the motor 30 changes. When the pressure difference in the user's heart changes, the motor current will also change to keep the rotor 32 speed constant. For example, when the blood flow rate into the aorta increases (such as during systole), the current required by the motor 30 will increase. Therefore, the change in the motor current can thus help characterize heart performance. That is to say, during the operation of the ventricular assist device 100, the ventricular assist device 100 has a current-flow characteristic curve, where the larger the current, the more work the ventricular assist device 100 does, that is, the larger the pumping flow rate of the ventricular assist device 100.

[0069] The current of the ventricular assist device 100 can be measured by a set phase current detection circuit or any other suitable means (such as a current sensor). This current-flow characteristic curve can be pre-stored in the control unit 33. Before the ventricular assist device 100 leaves the factory, it can be placed in a test system to test the relationship curve of the pumping flow rate changing with the current at different speeds of the ventricular assist device 100, and then store this current-flow characteristic curve in the control unit 33. The control unit 33 can store the detected current in real time.

[0070] When the ventricular assist device 100 operates at a target speed, after the control unit 33 obtains the current curve, it uses the pre-stored current-flow characteristic curve to estimate the corresponding flow curve of this current curve to obtain the target pumping flow curve.

[0071] Among them, the first period is a multiple of the patient's cardiac cycle, such as 3 times, 5 times, 10 times, 30 times, etc. of the cardiac cycle.

[0072] S420. Extract the target feature set of the target pumping flow curve.

[0073] Among them, the target feature set includes a first feature, a second feature, a third feature, and a fourth feature. The first feature is used to measure the symmetry degree of the target pumping flow curve. The second feature is the ratio of the low frequency to the fundamental frequency of the target pumping flow curve in the frequency domain. The third feature is the ratio of the high frequency to the fundamental frequency of the target pumping flow curve in the frequency domain. The fourth feature is used to measure the distribution shape of the target pumping flow curve.

[0074] In order to determine whether the ventricular assist device 100 is pumping, the control unit 33 can determine based on the characteristics represented by the target pumping flow curve. When the ventricular assist device 100 is operating normally, the waveform of its pumping flow curve is generally a sine wave waveform. When the ventricular assist device 100 is pumping, the waveform of the pumping flow curve of the ventricular assist device 100 will change. For example, if the pumping occurs during the systolic period of the cardiac cycle, the pumping flow during the systolic period of the cardiac cycle will decrease sharply or even decrease to 0. If the pumping occurs during the diastolic period of the cardiac cycle, the pumping flow during the systolic period of the cardiac cycle will continue to remain at a low value or suddenly decrease to a smaller pumping flow or even decrease to 0. Therefore, when the ventricular assist device 100 is pumping, its pumping flow curve will change in the time domain and frequency domain. Then, by extracting the characteristics of the pumping flow curve in the time domain and frequency domain, it can be determined whether the pumping flow curve is abnormal based on the characteristics, thereby determining whether the ventricular assist device 100 is pumping.

[0075] Optionally, the extraction of the target feature set of the target pumping flow curve includes: obtaining n pumping flow sampling points from the target pumping flow curve, where n is a positive integer; calculating the standard deviation and the average flow of the n pumping flow sampling points; substituting the standard deviation, the average flow and the n pumping flow sampling points into a first formula to calculate the first feature; substituting the average flow and the n pumping flow sampling points into a second formula to calculate the fourth feature; performing a Fourier transform on the target pumping flow to obtain fundamental frequency energy, low-frequency energy and high-frequency energy; calculating the ratio of the low-frequency energy to the fundamental frequency energy to obtain the second feature; calculating the ratio of the high-frequency energy to the fundamental frequency energy to obtain the third feature.

[0076] When the ventricular assist device 100 is operating normally, its pumping flow curve is roughly symmetrically distributed. However, when the ventricular assist device 100 is pumping, its pumping flow rate will drop suddenly, causing its pumping flow curve to be symmetrical within the cardiac cycle. When the ventricular assist device 100 is operating normally, the pumping flow curve is roughly symmetrically distributed, and the first feature P1 is approximately equal to 0; when the ventricular assist device 100 is pumping, the pumping flow curve is asymmetric, and the first feature P1 is less than 0. When the ventricular assist device 100 is operating normally, its pumping flow curve is relatively evenly distributed, and the value of its feature four is small; when the ventricular assist device 100 is pumping, more extremely low values ​​will appear due to the pumping, making the value of feature four larger.

[0077] Aortic valve regurgitation, abnormal position of the ventricular assist device 100, abnormal patient heart rate, aspiration of the ventricular assist device 100, abnormal rotation speed (too high or too low) of the ventricular assist device 100, etc. may all cause abnormal pumping flow of the ventricular assist device 100 in the time domain. It has a large error to judge whether there is aspiration in the ventricular assist device 100 only through the abnormality of the pumping flow curve in the time domain. Therefore, in this application, the Fourier transform is performed on the pumping flow curve to obtain its spectrum data PF(w), and through spectrum analysis of the spectrum data, it can be judged whether there is aspiration in the frequency domain according to the analysis results.

[0078] When the ventricular assist device 100 operates normally, its pumping flow curve is mainly the fundamental frequency in the frequency domain, its low-frequency energy is much lower than the fundamental frequency energy, and its high-frequency energy is also much lower than the fundamental frequency energy. The change in the pumping flow during aspiration will be manifested as clutter of abnormal frequencies in the frequency domain, and the clutter of abnormal frequencies may be low-frequency or high-frequency. Therefore, when the ventricular assist device 100 aspirates, the low-frequency energy or high-frequency energy will increase, so that the second feature P2 or the third feature P3 will increase.

[0079] The control unit 33 can extract the first feature P1, the second feature P2, the third feature P3, and the fourth feature P4 from the target pumping flow curve respectively. Specifically, the control unit 33 collects n pumping flow sampling points from the pumping flow curve according to the acquisition frequency, and then substitutes the n pumping flow sampling points into the first formula to calculate the first feature P1 of the target pumping flow curve, and substitutes the n pumping flow sampling points into the second formula to calculate the fourth feature P4 of the target pumping flow curve. At the same time, perform Fourier transform on the target pumping flow curve to obtain the spectrum data of the target pumping flow curve, and calculate the third feature P3 and the fourth feature P4 according to the spectrum data.

[0080] S430. Input the target feature set into the target classification tree model to obtain a classification result.

[0081] The control unit 33 inputs the first feature P1, the second feature P2, the third feature P3, and the fourth feature P4 extracted from the target pumping flow curve into the target classification tree model to obtain the classification result of the target pumping flow curve, and the classification result includes aspiration and normal. If the classification result output by the target classification tree model is 1, it means that the current ventricular assist device 100 aspirates; if the classification result output by the target classification tree model is 0, it means that the current ventricular assist device 100 operates normally.

[0082] To better illustrate the classification process of the classification tree model, a small part of the classification tree is taken as an example for illustration, such as Figure 5As shown, at the root node of the classification tree, if the first feature P1 is less than 0.14173, it enters the left branch; if the first feature P1 is greater than or equal to 0.14173, it enters the right branch. At the child node 1 in the left branch, if the fourth feature P4 is less than 1.74075, its classification result is y = 0; if the fourth feature P4 is greater than or equal to 1.74075, it enters child node 2. If the second feature P2 is less than 0.426645, its classification result is y = 1; if the second feature P2 is greater than or equal to 0.426645, its classification result is y = 0. At the child node 1 in the right branch, if the third feature P3 is less than 0.148827, its classification result y = 0; if the third feature P3 is greater than or equal to 0.148827, its classification result y = 1. Assume that the first feature P1, the second feature P2, the third feature P3, and the fourth feature P4 of the target pumping flow curve are 0.13877, 0.41452, 0.14523, and 1.80124 respectively, then its classification result is y = 1, indicating that the current ventricular assist device has suction.

[0083] S440. When the classification result is suction, calculate the suction degree of the ventricular assist device.

[0084] After determining that the ventricular assist device 100 has suction, for the convenience of medical staff to propose solutions in a timely manner, the control unit 33 can also determine the suction degree of the ventricular assist device 100.

[0085] Optionally, the step of calculating the suction degree of the ventricular assist device when the classification result is suction includes: obtaining a target average flow rate, where the target average flow rate is the average pumping flow rate during the second cycle of operation of the ventricular assist device, and the second cycle is earlier than the first cycle; obtaining the minimum pumping flow rate and the maximum pumping flow rate of the target pumping flow curve; calculating a discrimination threshold according to the target average flow rate, the minimum pumping flow rate, and the maximum pumping flow rate; and determining the suction degree of the ventricular assist device according to the discrimination threshold.

[0086] In the embodiment of the present application, to further determine the suction degree of the ventricular assist device 100, the control unit 33 can use the sudden drop degree of the pumping flow rate of the ventricular assist device 100 as a discrimination basis to determine the suction degree of the ventricular assist device 100.

[0087] The control unit 33 obtains the pumping flow rate curve in the second period and calculates the average value of the pumping flow rate curve in the second period. The second period is the period when the ventricular assist device 100 operates normally, that is, the pumping flow rate curve in the second period is a normal pumping flow rate curve. When the ventricular assist device 100 has suction, the minimum value or the maximum and minimum values of its pumping flow rate will change. By comparing it with the average value during normal operation, the degree of sudden drop in the pumping flow rate can be evaluated, and then the degree of suction can be determined according to the degree of sudden drop.

[0088] Specifically, the control unit 33 calculates the first difference between the target average flow rate and the minimum value in the target pumping flow rate curve, and the second difference between the target average flow rate and the maximum value in the target pumping flow rate curve, and then takes the ratio of the first difference to the second difference as the discrimination threshold.

[0089] Among them, determining the degree of suction of the ventricular assist device 100 according to the discrimination threshold includes: if the discrimination threshold is greater than the first value, it is determined that the ventricular assist device 100 has severe suction; if the discrimination threshold is greater than or equal to the second value and less than or equal to the first value, it is determined that the ventricular assist device 100 has moderate suction; if the discrimination threshold is greater than the third value and less than the second value, it is determined that the ventricular assist device 100 has mild suction.

[0090] When the ventricular assist device 100 operates normally, the average pumping flow rate is a value with very little fluctuation. Therefore, the target average flow rate can be approximated as a fixed value. The more severe the suction of the ventricular assist device 100, the smaller the minimum value in the target pumping flow rate curve, resulting in a larger discrimination threshold. When the calculated discrimination threshold is greater than the first value, it is considered that the current ventricular assist device 100 has severe suction; when the discrimination threshold is less than or equal to the first value and greater than or equal to the second value, it is considered that the current ventricular assist device 100 has moderate suction; when the discrimination threshold is less than the second value and greater than the third value, it is considered that the current ventricular assist device 100 has mild suction.

[0091] Among them, the first value, the second value, and the third value can be set according to the accuracy of suction detection. For example, the first value, the second value, and the third value can be set to 1.4, 1.2, and 1 respectively.

[0092] In the embodiment of the present application, the control unit 33 can quickly detect whether the current ventricular assist device 100 has suction by using a classification tree model, improving the accuracy of suction detection, and further determining the degree of suction when there is suction, which can timely remind medical staff to take corresponding solutions and improve the safety of patients.

[0093] Embodiment III

[0094] Embodiment 3 of the present application provides a method for training a classification tree model, which can be the classification tree model in Embodiment 1. This training method can train the classification tree model to be trained so that the trained classification tree model can accurately determine whether there is suction in the ventricular assist device according to the pumping flow rate curve.

[0095] In the embodiment of the present application, before training the classification tree model, the method further includes: obtaining a training data set, where the training data set includes a plurality of first pumping flow rate curves, and the first pumping flow rate curve is the pumping flow rate curve when the ventricular assist device is sucking or operating normally; marking the classification label of each first pumping flow rate curve; calculating a first ratio and a second ratio according to the classification label, where the first ratio is the proportion of the first pumping flow rate curves with the classification label of suction in the training data set, and the second ratio is the proportion of the first pumping flow rate curves with the classification label of normal in the training data set.

[0096] The plurality of first pumping flow rate curves include the pumping flow rate curve when the ventricular assist device is operating normally and the pumping flow rate curve when the ventricular assist device is sucking. Each first pumping flow rate curve is the pumping flow rate curve within the first cycle of the operation of the ventricular assist device 100. After obtaining the training data set, the classification label of each first pumping flow rate curve in the training data set is manually marked. The classification labels are divided into suction and normal to distinguish whether the first pumping flow rate curve is a normal pumping flow rate curve or a sucking pumping flow rate curve. After marking the classification labels of the plurality of first pumping flow rate curves, the proportion of the pumping flow rate curves when the ventricular assist device 100 is sucking and the proportion of the pumping flow rate curves when the ventricular assist device 100 is normal in the plurality of first pumping flow rate curves can be calculated respectively.

[0097] The control unit 33 trains the classification tree model to be trained, uses the plurality of first pumping flow rate curves as the input of the classification tree model to be trained, uses the first ratio and the second ratio as the Gini coefficients of the output categories of the classification tree model, and continuously optimizes the splitting points in the classification tree model to obtain the final target classification tree model.

[0098] Optionally, training the target classification tree model specifically includes: extracting the target feature set of the plurality of first pumping flow rate curves; inputting each target feature set into the classification tree model to be trained to obtain the classification result of each first pumping flow rate curve; determining the splitting point of each layer in the classification tree model to be trained according to the classification result, the first ratio, and the second ratio to obtain the target classification tree model.

[0099] Extract the first feature P1, the second feature P2, the third feature P3, and the fourth feature P4 from each first pumping flow rate curve. First, sample n pumping flow rate sampling points from the first pumping flow rate curve according to the adoption frequency, and then calculate the standard variance S and the average value of the first pumping flow rate curve based on these n pumping flow rate sampling points. . Then substitute the standard variance S and the average value into the above first formula to calculate the first feature P1, and substitute the average value into the above second formula to calculate the fourth feature P4. At the same time, perform Fourier transform on each first pumping flow rate curve to obtain the corresponding spectrum data, perform spectrum analysis on the spectrum data, calculate the low-frequency energy N1, the fundamental frequency energy N2, and the high-frequency energy N3 corresponding to the low frequency, fundamental frequency, and high frequency in each spectrum data, so as to calculate the second feature P2 = N1 / N2 and the third feature P3 = N3 / N2 of each first pumping flow rate curve.

[0100] Construct a classification tree based on the four features of the first feature P1, the second feature P2, the third feature P3, and the fourth feature P4. Specifically: divide the data into categories y = 1 and y = 0, where y = 1 represents that the classification result is suction, and y = 0 represents that the classification result is normal. Use these four features as the splitting points for each layer of the classification tree, traverse these four features, and try all possible splitting points. Then input the first feature P1, the second feature P2, the third feature P3, and the fourth feature P4 of multiple first pumping flow rate curves into the classification tree formed by each possible splitting point respectively to obtain the classification results of different splitting points. By calculating the Gini coefficients of the two groups of data of y = 1 and y = 0 for the multiple first pumping flow rate curves obtained after classification through each splitting point, use the feature and splitting point with the smallest calculated Gini coefficient as the splitting condition for this layer. During the splitting process, if the Gini coefficients of the two groups of data of y = 1 and y = 0 calculated after splitting this layer differ greatly from the first ratio and the second ratio of these multiple first pumping flow rate curves or their error is greater than the preset error, continue to split the child nodes of this layer node. In this way, layer by layer splitting is performed until the optimal splitting point that meets the preset error is found layer by layer, so as to construct the classification rules of the classification tree and obtain the final classification tree model.

[0101] It should be noted that the training process of the classification tree model in the embodiments of the present application is the construction process of the classification tree. By continuously optimizing the splitting points of each layer in the classification tree through the training data set, the classification rules of the classification tree are constructed, so that the error between the classification result after classifying the pumping flow rate curve according to this classification tree model and the actual result is within the preset error.

[0102] It can be seen that in the training method of the classification tree model provided by the embodiments of the present application, during the training process, by selecting appropriate features, collecting more data samples, improving data quality, and using cross-validation, the accuracy of classification can be improved.

[0103] The above mainly introduces the solutions of the embodiments of the present application from the perspective of the execution process of the method side. It can be understood that in order for a network device to implement the above functions, it includes the corresponding hardware structure and / or software module for executing each function. Those skilled in the art should easily realize that, in combination with the units and algorithm steps of each example described in the embodiments provided herein, the present application can be implemented in the form of hardware or a combination of hardware and computer software. Whether a certain function is executed in the way of hardware or computer software driving hardware depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of the present application.

[0104] Exemplarily, the present application provides a control unit for a ventricular assist device. The control unit includes one or more processors, and the one or more processors are used to: obtain a target pumping flow curve, where the target pumping flow curve is the pumping flow curve of the ventricular assist device during the first cycle of operation; extract a target feature set of the target pumping flow curve;

[0105] input the target feature set into a target classification tree model to obtain a classification result; and calculate the suction degree of the ventricular assist device when the classification result is suction.

[0106] Exemplarily, the present application also provides a ventricular assist device, and the ventricular assist device includes:

[0107] a housing;

[0108] an impeller disposed in the housing;

[0109] a motor for driving the impeller to rotate in a suspended manner;

[0110] a control unit connected to the motor, and the control unit is used for some or all of the steps described in the above method.

[0111] Exemplarily, the present application also provides a medical device, and the medical device includes the above-mentioned control unit or ventricular assist device.

[0112] Among them, the control unit of each of the above solutions has the function of implementing the corresponding steps executed by the medical device in the above method; the function can be implemented by hardware or by hardware executing the corresponding software.

[0113] In an embodiment of the present application, the control unit may also be a chip or a chip system, for example: a system on chip (SoC).

[0114] Please refer to Figure 6 , Figure 6 FIG. is a schematic structural diagram of a medical device provided by an embodiment of the present application. The medical device includes: one or more processors, one or more memories, one or more communication interfaces, and one or more programs; the one or more programs are stored in the memory and are configured to be executed by the one or more processors.

[0115] The above program includes instructions for performing the following steps:

[0116] Obtain a target pumping flow curve, where the target pumping flow curve is the pumping flow curve of the first cycle of operation of the ventricular assist device;

[0117] Extract a target feature set of the target pumping flow curve;

[0118] Input the target feature set into a target classification tree model to obtain a classification result;

[0119] When the classification result is suction, calculate the suction degree of the ventricular assist device.

[0120] Wherein, all relevant contents of each scenario involved in the above method embodiment can be cited in the function description of the corresponding functional module, and will not be elaborated here.

[0121] It should be understood that the above memory may include a read-only memory and a random access memory, and provide instructions and data to the processor. A part of the memory may also include a non-volatile random access memory. For example, the memory may also store information about the device type.

[0122] In an embodiment of the present application, the processor of the above device may be a central processing unit (CPU), and the processor may also be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0123] It should be understood that the "at least one" involved in the embodiments of the present application refers to one or more, and the "multiple" refers to two or more. "And / or" describes the association relationship of associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist simultaneously, and B exists alone, where A and B can be singular or plural. The character " / " generally represents an "or" relationship between the associated objects before and after. "At least one (item)" or its similar expression refers to any combination of these items, including any combination of single item (item) or plural items (items). For example, at least one (item) of a, b, or c can represent: a, b, c, a - b, a - c, b - c, or a - b - c, where a, b, and c can be single or multiple.

[0124] Moreover, unless otherwise stated, the ordinal numbers such as "first" and "second" mentioned in the embodiments of the present application are used to distinguish multiple objects, and are not used to limit the order, time sequence, priority, or importance of multiple objects. For example, the first information and the second information are only used to distinguish different information, rather than indicating differences in the content, priority, sending order, or importance of these two types of information.

[0125] In the implementation process, each step of the above method can be completed by the integrated logic circuit in the hardware of the processor or the instructions in the form of software. The steps of the method disclosed in combination with the embodiments of the present application can be directly embodied as being executed and completed by the hardware processor, or executed and completed by the combination of the hardware and software units in the processor. The software unit can be located in a mature storage medium in the art such as random access memory, flash memory, read-only memory, programmable read-only memory, or electrically erasable programmable memory, register, etc. This storage medium is located in the memory, and the processor executes the instructions in the memory and combines its hardware to complete the steps of the above method. To avoid repetition, it will not be described in detail here.

[0126] The embodiments of the present application also provide a computer storage medium, wherein the computer storage medium stores a computer program for electronic data exchange, and the computer program enables the computer to execute part or all of the steps of any method recorded in the above method embodiments.

[0127] The embodiments of the present application also provide a computer program product. The above computer program product includes a non-transitory computer-readable storage medium storing a computer program, and the above computer program is operable to enable the computer to execute part or all of the steps of any method recorded in the above method embodiments. The computer program product can be a software installation package.

[0128] It should be noted that, for the foregoing method embodiments, for the sake of simple description, they are all expressed as a series of action combinations. However, those skilled in the art should understand that this application is not limited by the described action sequence, because according to this application, certain steps can be performed in other sequences or simultaneously. Secondly, those skilled in the art should also understand that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily essential to this application.

[0129] In the above embodiments, the descriptions of the respective embodiments have their own focuses. For the parts not detailed in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0130] In the several embodiments provided by this application, it should be understood that the disclosed device can be implemented in other ways. For example, the device embodiments described above are only illustrative. For example, the above division of units is only a logical function division. In actual implementation, there can be other division methods. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed couplings or direct couplings or communication connections to each other can be through some interfaces. The indirect couplings or communication connections of devices or units can be in electrical or other forms.

[0131] The units described as separate components above may or may not be physically separated. The components displayed as units may or may not be physical units, that is, they can be located in one place, or they can be distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of the embodiments of this application.

[0132] In addition, in each embodiment of this application, the functional units can be integrated in a processing unit, or each unit can exist physically alone, or two or more units can be integrated in one unit. The above integrated units can be implemented in the form of hardware or in the form of software functional units.

[0133] If the above integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable memory. Based on such an understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a memory and includes several instructions for causing a computer device (which can be a personal computer, a server, or a TRP, etc.) to execute all or part of the steps of the methods in various embodiments of this application. The aforementioned memory includes: various media such as USB flash drives, read-only memories (ROM, Read-Only Memory), random access memories (RAM, Random Access Memory), external hard drives, magnetic disks, or optical discs that can store program codes.

[0134] Those of ordinary skill in the art can understand that all or part of the steps in the various methods of the above embodiments can be completed by instructing relevant hardware through a program. This program can be stored in a computer-readable memory, and the memory can include: flash drives, ROM, RAM, magnetic disks, or optical discs, etc.

[0135] The above has introduced the embodiments of this application in detail. Specific examples are used in this article to elaborate on the principles and implementation manners of this application. The description of the above embodiments is only used to help understand the method and its core idea of this application; at the same time, for those of ordinary skill in the art, according to the idea of this application, there will be changes in the specific implementation manners and application scopes. In summary, the content of this specification should not be construed as a limitation to this application.

Claims

1. A method for detecting aspiration of a ventricular assist device, characterized in that: The method comprises: Acquire a target pumping flow curve, where the target pumping flow curve is a pumping flow curve of the first cycle of the ventricular assist device; Extracting target feature sets of the target pumping flow curve in the time domain and the frequency domain, wherein the target feature set in the time domain includes a first feature and a fourth feature; the target feature set in the frequency domain includes a second feature and a third feature, wherein the first feature is used to measure the degree of symmetry of the target pumping flow curve, the second feature is the ratio of the low frequency to the fundamental frequency of the target pumping flow curve in the frequency domain, the third feature is the ratio of the high frequency to the fundamental frequency of the target pumping flow curve in the frequency domain, and the fourth feature is used to measure the distribution shape of the target pumping flow curve; Inputting the target feature set into a target classification tree model to obtain a classification result; When the classification result is suction, calculating the suction degree of the ventricular assist device; Among them, calculating the pumping degree of the ventricular assist device includes: obtaining a target average flow rate, the target average flow rate is the average pumping flow rate in the second cycle of the ventricular assist device operation, the second cycle is earlier than the first cycle; respectively calculating a first difference and a second difference, the first difference is the difference between the target average flow rate and the minimum value in the target pumping flow curve, and the second difference is the difference between the target average flow rate and the maximum value in the target pumping flow curve; using the ratio of the first difference to the second difference as a discrimination threshold; if the discrimination threshold is greater than the first value, determining that the ventricular assist device is severely pumping; if the discrimination threshold is greater than or equal to the second value and less than or equal to the first value, determining that the ventricular assist device is moderately pumping; if the discrimination threshold is greater than a third value and less than the second value, determining that the ventricular assist device is lightly pumping.

2. The method according to claim 1, characterized in that The step of extracting the target feature set of the target pumping flow curve in the time domain and the frequency domain comprises: obtaining n pumping flow sampling points from the target pumping flow curve, wherein n is a positive integer; calculating the standard deviation and the average flow of the n pumping flow sampling points; substituting the standard deviation, the average flow and the n pumping flow sampling points into a first formula to calculate the first feature; substituting the average flow and the n pumping flow sampling points into a second formula to calculate the fourth feature; performing Fourier transformation on the target pumping flow to obtain fundamental frequency energy, low frequency energy and high frequency energy; calculating the ratio of the low frequency energy to the fundamental frequency energy to obtain the second feature; calculating the ratio of the high frequency energy to the fundamental frequency energy to obtain the third feature; The first formula is: ; The second formula is: ; Said is the pumping flow sampling point, i is 1, 2, ..., n, is the average flow rate of the n pumping flow sampling points, and S is the standard deviation of the n pumping flow sampling points.

3. The method according to claim 1, characterized in that The method further comprises: Acquire a training data set, wherein the training data set includes a plurality of first pumping flow curves, wherein the first pumping flow curve is a pumping flow curve when the ventricular assist device is pumping or operating normally; labeling each of the first pumping flow rate curves with a classification label; A first ratio and a second ratio are calculated according to the classification label, wherein the first ratio is the proportion of the first pumping flow curve whose classification label is suction in the training data set, and the second ratio is the proportion of the first pumping flow curve whose classification label is normal in the training data set.

4. The method according to claim 3, characterized in that: The method further includes: training the target classification tree model, specifically including: extracting the target feature set of the plurality of first pumping flow rate curves; Inputting each of the target feature sets into a classification tree model to be trained to obtain the classification result of each of the first pumping flow rate curves; The segmentation points of each layer in the classification tree model to be trained are determined according to the classification result, the first ratio and the second ratio to obtain the target classification tree model.

5. The method according to claim 1, characterized in that The first value, the second value, and the third value are determined by a puff detection accuracy.

6. A control unit for a ventricular assist device, characterized in that: The control unit comprises one or more processors, wherein the one or more processors are configured to: Acquire a target pumping flow curve, where the target pumping flow curve is a pumping flow curve of the first cycle of the ventricular assist device; Extracting a target feature set of the target pumping flow rate curve in the time domain and the frequency domain, wherein the target feature set in the time domain includes a first feature and a fourth feature; The target feature set in the frequency domain includes a second feature and a third feature, the first feature is used to measure the symmetry of the target pumping flow curve, the second feature is the ratio of the low frequency to the fundamental frequency of the target pumping flow curve in the frequency domain, the third feature is the ratio of the high frequency to the fundamental frequency of the target pumping flow curve in the frequency domain, and the fourth feature is used to measure the distribution shape of the target pumping flow curve; Inputting the target feature set into a target classification tree model to obtain a classification result; When the classification result is suction, calculating the suction degree of the ventricular assist device; Among them, calculating the pumping degree of the ventricular assist device includes: obtaining a target average flow rate, the target average flow rate is the average pumping flow rate in the second cycle of the ventricular assist device operation, the second cycle is earlier than the first cycle; respectively calculating a first difference and a second difference, the first difference is the difference between the target average flow rate and the minimum value in the target pumping flow curve, and the second difference is the difference between the target average flow rate and the maximum value in the target pumping flow curve; using the ratio of the first difference to the second difference as a discrimination threshold; if the discrimination threshold is greater than the first value, determining that the ventricular assist device is severely pumping; if the discrimination threshold is greater than or equal to the second value and less than or equal to the first value, determining that the ventricular assist device is moderately pumping; if the discrimination threshold is greater than a third value and less than the second value, determining that the ventricular assist device is lightly pumping.

7. A ventricular assist device, characterized in that: The ventricular assist device comprises: case; an impeller disposed in the housing; A motor driving the impeller to rotate in suspension; A control unit connected to the motor, the control unit being used to execute instructions of the steps in the method according to any one of claims 1-5.

8. A medical device, characterized in that: The method comprises a processor, a memory and a communication interface, wherein the memory stores one or more programs, and the one or more programs are executed by the processor, and the one or more programs include instructions for executing the steps in the method according to any one of claims 1 to 5.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program for electronic data exchange, wherein the computer program enables a computer to execute the steps of the method according to any one of claims 1 to 5.

Images