Ventricular assist device and control unit therefor, medical device

Abstract

The application provides a ventricular assist device, a control unit thereof and a medical device. The control unit acquires a target pumping flow curve, which is a pumping flow curve of the ventricular assist device within a target time length. A plurality of target characteristics are extracted from the target pumping flow curve, and the plurality of target characteristics are used to measure characteristics of the target pumping flow curve in the time domain and the frequency domain. The plurality of target characteristics are input into a target suction detection model, and a target detection result is output. The application extracts characteristics in the time domain and the frequency domain from a single pumping flow curve, and combines a machine learning model to detect the state of the ventricular assist device, so that suction detection of the ventricular assist device can be realized in real time and accurately, and user safety is improved.

Description

Technical Field

[0001] This application relates to the field of medical device technology, and in particular to a ventricular assist device and its control unit, and a medical device. Background Technology

[0002] Currently, ventricular assist devices (VADs) have become an important means of treating end-stage heart failure. These are artificial mechanical devices that draw fluid from the venous system or heart directly into the arterial system, partially or completely replacing the work of the ventricles, thus addressing the problem of heart donor shortages. Developing suitable pump control systems to meet patients' circulatory needs, and regulating the flow rate through the VAD by controlling the pump speed, is a significant challenge arising from the increasing use of these devices. However, when the VAD's speed exceeds the user's requirements, it may cause suction problems, potentially damaging the patient's heart. Summary of the Invention

[0003] This application provides a ventricular assist device and its control unit, as well as a medical device, which can detect in a timely and accurate manner whether there is aspiration in the ventricular assist device without relying on external equipment, so as to prevent irreversible damage to the user.

[0004] In a first aspect, embodiments of this application provide a control unit for a ventricular assist device, the control unit being configured to perform the following steps: Obtain the target pumping flow rate curve, which is the pumping flow rate curve of the ventricular assist device within a target duration; Multiple target features are extracted from the target pumping flow rate curve, and these multiple target features are used to measure the characteristics of the target pumping flow rate curve in the time domain and frequency domain. The multiple target features are input into the target suction detection model, and the target detection results are output.

[0005] Secondly, embodiments of this application provide a ventricular assist device, the ventricular assist device comprising: impeller; A motor that drives the impeller to rotate; A control unit connected to the motor is configured to perform the following steps: Obtain the target pumping flow rate curve, which is the pumping flow rate curve of the ventricular assist device within a target duration; Multiple target features are extracted from the target pumping flow rate curve, and these multiple target features are used to measure the characteristics of the target pumping flow rate curve in the time domain and frequency domain. The multiple target features are input into the target suction detection model, and the target detection results are output.

[0006] Thirdly, embodiments of this application provide a medical device, which includes the control unit described in the first aspect above.

[0007] Fourthly, embodiments of this application provide a medical device, the medical device including a processor, a memory, a communication interface, and one or more programs, the one or more programs being stored in the memory and configured to be executed by the processor, the programs including instructions for performing the execution steps of the control unit described above.

[0008] The technical solution provided in this application involves a control unit acquiring a target pumping flow rate curve, which is the pumping flow rate curve of the ventricular assist device within a target duration. Multiple target features are extracted from the target pumping flow rate curve, and these features are used to measure the characteristics of the target pumping flow rate curve in the time and frequency domains. These multiple target features are input into a target aspiration detection model, and the target detection result is output. This application extracts time and frequency domain features from a single pumping flow rate curve and combines this with a machine learning model to detect the state of the ventricular assist device, enabling real-time and accurate aspiration detection of the ventricular assist device and improving user safety. Attached Figure Description

[0009] To more clearly illustrate the technical solutions in the embodiments of this application, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0010] Figure 1 This is a schematic diagram of the structure of a ventricular assist device provided in an embodiment of this application; Figure 2 This is a schematic diagram of a ventricular assist device placed in the left ventricle according to an embodiment of this application; Figure 3 This is a schematic diagram of a control unit performing suction detection according to an embodiment of this application; Figure 4 This is a schematic diagram of the structure of a medical device provided in an embodiment of this application. Detailed Implementation

[0011] To help those skilled in the art better understand the technical solutions of this application, the technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the description of the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0012] The terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish different objects, not to describe a specific order. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, software, product, or apparatus that includes a series of steps or units is not limited to the listed steps or units, but also includes steps or units not listed, or other steps or units inherent to such processes, methods, products, or apparatus.

[0013] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0014] It should be noted that in this application, the terms "proximal" or "proximal" refer to the end or side closer to the surgeon; and "distal" or "distal" refer to the end or side farther from the surgeon.

[0015] Please see Figures 1-2 , Figure 1 This is a schematic diagram of the structure of a ventricular assist device 100 provided in an embodiment of this application. Figure 2This is a schematic diagram illustrating a ventricular assist device 100 located in the normal position of a patient's heart 120, as provided in an embodiment of this application. The ventricular assist device 100 can operate within the patient's heart, partially within the heart, outside the heart, partially outside the heart, partially outside the vascular system, or in any other suitable location within the vascular system. The ventricular assist device 100 can be percutaneously inserted into the aorta 124 via the femoral artery 122 and passes through the aorta 124 into the left ventricle 128. For example, the ventricular assist device 100 can be percutaneously inserted into the aorta 124 via the axillary artery 123 and passes through the aorta 124 into the left ventricle. In other embodiments, the ventricular assist device 100 can also be directly inserted into the aorta 124 and passes through the aorta 124 into the left ventricle 128. During operation, the ventricular assist device 100 pumps blood from the left ventricle 128 into the aorta 124.

[0016] The ventricular assist device 100 includes a cannula 10. The cannula 10 has a proximal end and a distal end, the distal end of the cannula 10 having a fluid inlet 101 and the proximal end of the cannula 10 having a fluid outlet 102, through which blood flows in from the fluid inlet 101 and out from the fluid outlet 102 via the cannula 10.

[0017] The ventricular assist device 100 includes an impeller (not shown). The impeller is at least partially located at the proximal end of the cannula 10, such as at the fluid inlet 101 of the cannula 10. Alternatively, the impeller is at least partially located at the distal end of the cannula 10, such as at the fluid outlet 102 of the cannula 10. This is such that when the ventricular assist device 100 is in operation, it drives the impeller to rotate to pump blood from the left ventricle 128 to the aorta 124.

[0018] The ventricular assist device 100 may include a motor (not shown in the figure), which may be located inside or outside the ventricular assist device 100. This embodiment of the application illustrates the example where the motor is located inside the ventricular assist device 100. For instance, the motor is housed in a motor housing 201, with the distal end of the motor housing 201 connected to the proximal end of the cannula 10. The motor drives the drive shaft to rotate, thereby rotating the impeller and realizing the pumping function of the ventricular assist device 100.

[0019] The ventricular assist device 100 includes a catheter 30, the distal end of which is connected to the proximal end of a motor housing 201, through which a drive cable extends. As an example, the catheter 30 may accommodate an electrical connection between the ventricular assist device 100 and an external controller. As an example, the ventricular assist device 100 also includes a distal component 110, such as a pigtail cannula, extending distally away from the distal end of the cannula 10.

[0020] When the ventricular assist device 100 is used on a patient's left heart, the ventricular assist device 100 can be considered to be in the desired position when it is positioned such that the cannula 10 extends across the patient's aortic valve 126, the distal end of the cannula 10 is located in the patient's left ventricle 128, the proximal end of the cannula 10 is located in the patient's aorta 124, and the distal component 110 is just abutting against or at a predetermined distance from the inner wall of the patient's left ventricle 128.

[0021] It should be noted that the ventricular assist device 100 in this application is not limited to use in the left ventricle, but can also be used in the right ventricle. When the ventricular assist device 100 is used in the right ventricle, the distal end of the cannula 10 has a fluid outlet 101 and the proximal end of the cannula 10 has a fluid inlet 102. Blood flows in from the fluid inlet 102 and flows out from the fluid outlet 101 via the cannula 10. The ventricular assist device 100 is considered to be in the desired position, i.e., the ventricular assist device 100 is in normal position, when the cannula 10 extends across the pulmonary valve, the distal end of the cannula 10 is located in the patient's right ventricle, the proximal end of the cannula 10 is located in the patient's pulmonary artery, and the distal component 110 is just abutting the ventricular wall of the patient's right ventricle. The following description uses the ventricular assist device 100 used in the left ventricle as an example.

[0022] The ventricular assist device 100 also includes pressure sensors disposed on the outer surface of the cannula 10 and positioned corresponding to the fluid outlet 102 and / or fluid inlet 101, for detecting pressure at the fluid outlet 102 and / or the fluid inlet 101 of the ventricular assist device 100. When the ventricular assist device 100 is correctly positioned across the aorta 124, the top (outer surface) of the pressure sensor at the fluid outlet 102 is exposed to the aorta 124, thereby allowing measurement of aortic pressure; the top (outer surface) of the pressure sensor at the fluid inlet 101 is exposed to the left ventricle 128, thereby allowing measurement of left ventricular pressure.

[0023] The ventricular assist device 100 also includes a control unit, which can be used to perform any of the embodiments, aspects, and methods of this application. The control unit may be located inside or outside the ventricular assist device 100. The control unit is used to detect relevant parameters of the ventricular assist device 100 and the patient, and to control the operation of the ventricular assist device 100. For example, the control unit supplies current to a motor through one or more wires and detects the current through a current detection circuit (such as a phase current detection circuit); controls the rotational speed of the ventricular assist device 100 according to received instructions; receives feedback signals from pressure sensors to identify the position of the ventricular assist device 100, and so on.

[0024] The ventricular assist device 100 is implanted into the patient's heart via a guidewire and positioned as desired. When the ventricular assist device 100 is in the left ventricle, its desired position is such that the ventricular assist device crosses the aortic valve 126, with the fluid outlet 102 completely within the aorta 124 and the fluid inlet 101 completely within the left ventricle 128. When the ventricular assist device 100 is in the right ventricle, the fluid outlet 102 is completely within the pulmonary artery, and the fluid inlet 101 is completely within the right atrium. After the procedure, the patient requires ventricular assist device 100 support for a period of time. During this time, excessive pumping by the ventricular assist device 100 can lead to insufficient preload, resulting in decreased left ventricular pressure and collapse. Ultimately, this can cause myocardial tissue to be aspirated, leading to aspiration of the ventricular assist device 100. Aspiration can lead to a series of serious consequences, such as a sudden drop in the pump flow rate of the ventricular assist device leading to insufficient perfusion of the whole body organs, patients often experience severe hypotension, which may develop into malignant ventricular arrhythmias, and the increased motor load due to the blockage of the fluid inlet may damage the ventricular assist device, thus endangering the user's safety.

[0025] Based on this, the control unit in this application extracts a set of multiple features with strong physical meaning and discriminative ability from a single pump flow signal, preprocesses these features and inputs them into a pre-trained machine learning model, and outputs the predicted detection results, so as to realize rapid, non-invasive and accurate detection of whether the current ventricular assist device 100 is aspirating, which can be universally applicable to users with different degrees of heart failure and different physiological conditions.

[0026] Based on the above description, this application will now be described from the perspective of method examples.

[0027] Please see Figure 3 , Figure 3 This application provides a schematic diagram of a control unit executing a suction detection process, applicable to, for example... Figures 1-2 The ventricular assist device 100 is shown. (As shown in the image) Figure 3 As shown, the control unit performs the following steps.

[0028] S310. Obtain the target pumping flow rate curve, wherein the target pumping flow rate curve is the pumping flow rate curve of the ventricular assist device within a target duration.

[0029] During operation of the ventricular assist device 100 within the user's body, the pumping flow rate through the ventricular assist device 100 depends on the work that the ventricular assist device 100 must do to overcome resistance and pump blood from the left ventricle 128 to the aorta 124. The amount of work done by the ventricular assist device 100 can be quantified as the amount of current required to supply the motor (more specifically, the stator), i.e., the motor current corresponds to the amount of current delivered to the motor of the ventricular assist device 100 when it is operating within the user. The motor load varies during different phases of the user's cardiac cycle. When the pressure differential in the user's heart changes, the motor current also changes to maintain a constant rotor speed. For example, when the flow rate of blood into the aorta 124 increases (such as during cardiac contraction), the current required by the motor will increase. Therefore, changes in the motor current can thus help characterize cardiac performance. In other words, during the operation of the ventricular assist device 100, the ventricular assist device 100 has a current-flow characteristic curve, in which the greater the current, the more work the ventricular assist device 100 does, that is, the greater the pumping flow of the ventricular assist device 100.

[0030] The current of the ventricular assist device 100 can be measured by a 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. Before the ventricular assist device 100 leaves the factory, it can be placed in a testing system to test the relationship between the pumping flow rate and current at different speeds, and then the current-flow characteristic curve can be stored in the control unit. The control unit can store the detected current in real time.

[0031] When the ventricular assist device 100 is running at the target speed, after the control unit obtains the current curve, it uses the pre-stored current-flow characteristic curve to estimate the pumping flow curve corresponding to the current curve, and obtains the target pumping flow curve.

[0032] The target rotational speed is within the allowable range of the ventricular assist device 100. For example, if the rotational speed range is 23,000 RPM to 46,000 RPM, the target rotational speed can be set to 36,000 RPM. The target duration is a multiple of the patient's cardiac cycle, such as 3, 5, 10, or 30 times the cardiac cycle. For example, the target duration could be 5 seconds, 10 seconds, 15 seconds, or 20 seconds.

[0033] S320. Extract multiple target features from the target pumping flow rate curve, wherein the multiple target features are used to measure the characteristics of the target pumping flow rate curve in the time domain and frequency domain.

[0034] To determine whether the ventricular assist device 100 is aspirating, the control unit can make this judgment based on the characteristics represented by the target pump flow rate curve. When the ventricular assist device 100 is operating normally, its pump flow rate curve generally exhibits a sinusoidal waveform. When aspiration occurs, the waveform of the ventricular assist device 100's pump flow rate curve changes. For example, during aspiration, the pump flow rate of the ventricular assist device 100 may decrease sharply or even drop to zero; in extreme cases, backflow may occur. Therefore, by extracting the characteristics of the target pump flow rate curve in the time and frequency domains, the current aspiration detection result of the ventricular assist device 100 can be accurately determined.

[0035] Optionally, the extraction of multiple target features from the target pumping flow curve includes: calculating the zero-crossing rate asymmetry, negative pulse exponent, and kurtosis of local extrema of the target pumping flow curve; the zero-crossing rate asymmetry is used to measure the balance of fluctuations in the positive and negative directions of the target pumping flow curve; the negative pulse exponent is used to measure the intensity of negative blood flow pulses generated by the ventricular assist device; and the kurtosis of local extrema is used to measure the steepness of the amplitude distribution of local maxima and local minima of the target pumping flow curve; performing a Fourier transform on the target pumping flow curve to obtain a flow spectrum; and calculating the target energy percentage based on the flow spectrum, where the target energy percentage is the ratio of the energy in the target frequency band to the total energy.

[0036] Because of its small size and light weight, the interventional ventricular assist device 100 is more unstable during operation. Therefore, compared with implantable ventricular assist devices, the pump flow curve of the interventional ventricular assist device 100 has more spikes and its waveform is more irregular. By extracting features in the time and frequency domains that are strongly correlated with the state of the ventricular assist device 100 during aspiration (zero-crossing rate asymmetry, negative pulse exponent, kurtosis of local extrema, and target energy percentage), the current detection state of the ventricular assist device 100 can be identified from these features.

[0037] During the operation of the ventricular assist device 100, the control unit can set a sliding window, the duration of which can be set to a target duration. The control unit uses the pumping flow rate curve of the ventricular assist device 100 within the sliding window as the target pumping flow rate curve.

[0038] Zero-crossing asymmetry measures the balance of signal curve fluctuations in both positive and negative directions. Since aspiration in the ventricular assist device 100 causes asymmetric distortion in its pump flow curve, zero-crossing asymmetry is used as a characteristic of the pump flow curve of the ventricular assist device 100 during aspiration. The line corresponding to the mean of the pump flow curve within this time window is used as the horizontal axis (zero value). Zero-crossing asymmetry is defined as the number of positive zero-crossings N+ (i.e., the number of times the signal changes from negative to positive after crossing zero) and the number of negative zero-crossings N in the pump flow curve. The ratio of the number of times a signal changes from a positive value to a negative value (i.e., the number of times it crosses zero from a positive value to a negative value). Under normal conditions, the zero-crossing asymmetry is close to 1, but it deviates significantly during the pumping state.

[0039] The negative impulse index refers to the amplitude of flow fluctuations in the flow rate curve that are less than the average flow rate. It can be used to measure the intensity of abnormal negative blood flow pulses that occur during cardiac diastole, which is a direct manifestation of momentary reverse flow or a sudden drop in flow rate caused by ventricular suction. First, the diastolic time interval is estimated based on heart rate. Within the diastolic phase of each cardiac cycle, the minimum point of the flow rate signal is located. The amplitude of the negative pulse at that point ,in The average value of the pump flow rate curve within the current time window is used to calculate the average amplitude of all diastolic negative pulses within that time window. After normalization, the negative pulse index of the pump flow rate curve is obtained. When the ventricular assist device 100 performs aspiration, the value of the negative pulse index increases significantly.

[0040] The kurtosis of local extrema can be used to measure the steepness of the amplitude distribution of local maxima (peaks) and minima (troughs) sequences in a flow signal, reflecting the difference in amplitude distribution between physiological pulsations and abnormal fluctuations in the pump flow curve caused by suction. All local maxima sequences {Pi} and local minima sequences {Vi} are detected and extracted within a time window. The merged sequence E = {Pi} ∪ {Vi} is then used to calculate the sample kurtosis of the sequence.

[0041] Where N is the length of E. and These represent the mean and standard deviation of E, respectively. The peak-valley distribution corresponding to physiological pulsations is relatively concentrated (with moderate kurtosis), while suction can cause abnormal sharp peaks or deep valleys, leading to an increase in the absolute value of the kurtosis of local extreme values.

[0042] The target energy percentage measures the concentration of energy in the flow rate signal curve within the dominant frequency band centered on heart rate. Under normal physiological pulsation, the energy of the pumping flow rate curve is highly concentrated in the dominant frequency and its harmonics; as suction approaches, the regular pulsation weakens, and the energy diffuses into a wider frequency band; when suction occurs, the flow rate of the pumping flow rate curve drops sharply and many spikes appear, and these abnormal frequency noises reduce the concentration in the dominant frequency band. The dominant frequency band is [f...]. HR Δf, f HR +Δf], where Δf is the preset bandwidth (e.g., set to 0.5 Hz), f HR This represents the current heart rate frequency. The target energy percentage is the ratio of the energy in the main frequency band to the total energy. A higher target energy percentage indicates a lower probability of aspiration occurring.

[0043] After acquiring the target pumping flow rate curve, the control unit calculates characteristics such as zero-crossing rate asymmetry, negative pulse exponent, and kurtosis of local extrema in the time domain from the target pumping flow rate curve. Then, a Fourier transform is performed on the target pumping flow rate curve to obtain the flow rate spectrum. The target energy percentage in this flow rate spectrum is then calculated.

[0044] Furthermore, to ensure the accuracy of the detection results, the control unit can preprocess the signal. For example, the target pumping flow rate curve is passed through a Butterworth low-pass filter with a cutoff frequency of 20Hz to eliminate high-frequency measurement noise. Then, a fixed-length sliding window (e.g., 5s) is used to analyze the filtered signal curve to ensure real-time performance. Within each time window, the four characteristics of the target pumping flow rate curve (zero-crossing rate asymmetry, negative pulse exponent, distribution kurtosis of local extrema, and target energy proportion in the frequency domain) are calculated to obtain the four-dimensional feature vector of the target pumping flow rate curve.

[0045] S330. Input the multiple target features into the target suction detection model and output the target detection result.

[0046] The control unit inputs the zero-crossing rate asymmetry, negative pulse exponent, and kurtosis distribution characteristics of local extrema calculated from the target pumping flow rate curve in the time domain, as well as the target energy proportion characteristics in the frequency domain, into the target aspiration detection model, and outputs the target detection results. The target detection results include normal state, aspiration state, and approximate aspiration state. The normal state is the state when the ventricular assist device 100 is operating normally; the aspiration state is the state when the ventricular assist device 100 is aspirating; and the approximate aspiration state is the state n seconds before the ventricular assist device 100 aspirates. This target aspiration detection model is a support vector machine classification model.

[0047] Approaching aspiration is a critical warning state. In this state, the ventricular assist device 100 exhibits abnormal pumping flow and pressure (e.g., ventricular pressure is lower than normal, pumping flow is lower than normal), but the ventricular wall is not yet fully adhered, and blood flow is not interrupted. This application detects whether the ventricular assist device 100 is in a near-aspiration state, allowing for proactive intervention before danger occurs, achieving preventative protection, and avoiding serious consequences such as blood flow interruption, cardiac damage, hemolysis, thrombosis, and shock. Simultaneously, it maximizes the assist efficiency of the ventricular assist device 100.

[0048] In one possible example, the training method of the target aspiration detection model includes: acquiring k training datasets, each training dataset comprising r training data groups, each training data group comprising multiple first pump flow curves, multiple second pump flow curves, and multiple third pump flow curves, wherein the first pump flow curves are the pump flow curves when the ventricular assist device is in normal operation, the second pump flow curves are the pump flow curves when the ventricular assist device is in aspiration mode, and the third pump flow curves are the pump flow curves in the n seconds prior to the ventricular assist device determining that aspiration has occurred, where k and r are positive integers; using r-1 The target twitching detection model is obtained by training the r sets of training data with the training data set. A validation data set is then used to determine whether the target twitching detection model is valid. The validation data set consists of the remaining training data sets excluding the r-1 training data sets. If the target twitching detection model fails validation, k = k + 1, and the above steps are repeated until the target twitching detection model is valid or n is greater than a preset value. If the target twitching detection model is valid but is in the target condition, n = n + 1, and the above steps are repeated until the target twitching detection model is valid or n is greater than a preset value.

[0049] In this application, the aspiration detection model to be trained is trained using time-domain features and frequency-domain features calculated from the pump flow curve. By quantifying the correlation between multiple features and the aspiration state of the ventricular assist device 100, interpretable prediction of aspiration detection is achieved.

[0050] In practical applications, the sample size for aspiration is relatively small, and there is no specific limit to the acquisition time of the pump flow rate curve during the approximation of aspiration. If the acquisition time of the pump flow rate curve during the approximation of aspiration is too long, the pump flow rate under normal conditions will also be acquired, thereby increasing interference features and reducing the accuracy of the aspiration detection model. If the time is too short, there will be fewer features in the pump flow rate curve during the aspiration state, and the aspiration detection model may not be able to learn the core correlation between the approximation of the ventricular assist device 100 and multiple features in the pump curve, resulting in lower accuracy of the aspiration detection model.

[0051] To improve the detection accuracy of the target aspiration detection model, this application employs a cross-validation strategy for model training and validation. Multiple rounds of cross-validation are used to ensure the stability of the aspiration detection model under different data distribution scenarios, avoid overfitting, and verify the generalization ability of the aspiration detection model. Then, during model training and cross-validation, the performance of the aspiration detection model is evaluated by calculating the sensitivity dimension of the ventricular assist device 100 under different states, verifying its predictive ability for aspiration detection results. Furthermore, when the performance of the aspiration detection model does not meet the requirements, the training data is further optimized by increasing the amount of training data or the features carried in the training data, allowing the model to learn more about the correlation between the aspiration state of the ventricular assist device 100 and multiple features in the pumping flow curve, ultimately obtaining a high-precision and high-reliability target aspiration detection model.

[0052] The training dataset includes r training data sets. Each training data set contains multiple first pumping flow rate curves under normal conditions, second pumping flow rate curves under suction conditions, and third pumping flow rate curves under approximation suction conditions. Both j and r are integers greater than 1. The control unit first collects the pumping flow rate curve for the first n seconds before suction occurs as the third pumping flow rate curve. If the trained target suction detection model fails to meet the requirements, the collection time of the third pumping flow rate curve is increased, i.e., n is increased. For the initial training dataset, n can take a small value, such as 2s, 3s, 4s, etc. Then, the training model for suction detection is trained using r-1 training data sets from the training dataset. The remaining training data set is used as the validation data set to validate the performance of the trained suction detection model. This validation data set consists of the 1st, 2nd, 3rd, ..., rth training data sets. The r-1 training data sets consist of the 2nd to rth, 1st + 3rd to rth, 1st + 2nd + 4th to rth, ..., 1st to r-1th training data sets. This process is repeated r times for training and r times for validation. If the trained target suction detection model passes validation, it indicates that the trained suction detection model has the ability to accurately identify whether suction has occurred in the ventricular assist device 100. If the performance of the target suction detection model does not meet the requirements, the training dataset can be further optimized, such as by increasing the sample size of training data under different states or optimizing the granularity of suction acquisition. Then, the above "training-validation" process is re-executed using the optimized training dataset. This process is repeated iteratively until the slurry detection model passes validation. Finally, the validated slurry detection model is determined as the target slurry detection model.

[0053] Optionally, training the slurring detection model to be trained using r-1 sets of training data to obtain the target slurring detection model includes: determining the j-th training data set as the j-th validation data set, where j is a positive integer; training the slurring detection model to be trained using the remaining r-1 sets of training data excluding the j-th validation data set to obtain the target slurring detection model; inputting the j-th validation data set into the target slurring detection model and outputting the j-th detection result; calculating the j-th normal sensitivity, the j-th slurring sensitivity, and the j-th approximation slurring sensitivity based on the j-th detection result and the actual result, respectively; setting j = j + 1, and repeating the above steps until j = r.

[0054] This application uses an r-fold cross-validation loop to train and validate the target aspiration detection model. First, the first training data set is used as the validation data set, and the remaining r-1 training data sets are used as training data. The aspiration detection model to be trained is trained using these r-1 training data sets. After training, the trained aspiration detection model is validated using the validation data sets. This validation data set is input into the trained aspiration detection model to obtain the predicted detection result for each validation data set. Based on the predicted and actual detection results, the sensitivity of each validation data set is calculated under normal, aspiration, and near-aspiration states. Sensitivity represents the ability of the current aspiration detection model to identify whether the ventricular assist device 100 is in aspiration state; its value is between 0 and 1. The closer the sensitivity is to 1, the stronger the recognition ability of the current aspiration detection model and the fewer false negatives. Then, the second training data set is used as the validation data set, and the remaining r-1 training data sets are used as training data. The slurring detection model is trained using r-1 training datasets. After training, the model is validated using validation datasets to obtain the prediction detection results for each validation dataset. The sensitivity of each validation dataset is then calculated for the normal, slurring, and approximate slurring states. Following this method, the i-th training dataset from the r training datasets is selected for validation. The remaining training datasets are used to train the slurring detection model, and the i-th training dataset is used as the validation dataset to validate the trained slurring detection model. The sensitivity of each validation dataset is calculated for the normal, slurring, and approximate slurring states, until i=r. This yields the sensitivity of each of the r validation datasets for the normal, slurring, and approximate slurring states.

[0055] The process of training the slurring detection model using the training data set and validating the trained slurring detection model using the validation data set after training is as described above. Figure 3The application of the target suction detection model is the same. Features such as zero-crossing rate asymmetry, negative impulse exponent, and kurtosis of local extrema in the time domain, as well as target energy proportion features in the frequency domain, are extracted from the first, second, and third pump flow rate curves in the training or validation data sets. These features are then input into the suction detection model to be trained or into the trained suction detection model for validation. Detailed specifications will not be elaborated here.

[0056] Optionally, the step of using the verification data set to determine whether the target aspiration detection model has passed verification includes: calculating the overall average sensitivity and the approximate aspiration average sensitivity, wherein the approximate aspiration average sensitivity is the average value of the j-th approximate aspiration sensitivity, and the overall average sensitivity is the average of the approximate aspiration average sensitivity, the normal average sensitivity, and the aspiration aspiration sensitivity; if the overall average sensitivity is less than a first value, it is determined that the target aspiration detection model has failed verification.

[0057] The target aspiration detection model is verified as qualified but is in the target condition when: the overall average sensitivity is greater than or equal to the first value, and the approximation aspiration average sensitivity is less than the first value.

[0058] This application ensures that each training data set is validated through this method. A support vector machine model is fitted to the training data set, and performance is evaluated on the validation data set. Finally, the average of the r validation results is taken as the generalization performance metric of the slurring detection model. Specifically, the average sensitivity of the r validation data sets under normal, slurring, and approximate slurring states is calculated, and then the overall average sensitivity of the three states under normal, slurring, and approximate slurring states is calculated.

[0059] If the overall average sensitivity is less than the first value, it indicates that the accuracy of the current aspiration detection model has not yet met the requirements and needs further optimization. This means that the current aspiration detection model has not learned the core correlation between the different states of the ventricular assist device and multiple features in its pump flow curve. The amount of sample data needs to be increased. That is, let k=k+1, and supplement the sample data of the pump flow curves of the ventricular assist device 100 in three different states. Ensure that the effective sample size of each state covers the fluctuations of the pump flow curve of the ventricular assist device 100 and the differences in different experimental conditions, so that the target aspiration detection model can learn the core correlation. Then, the target aspiration detection model can be retrained based on the supplemented training dataset. If the overall average sensitivity is greater than or equal to the first value and the approximate aspiration average sensitivity is less than the first value, it indicates that the current aspiration detection model's detection accuracy under approximate aspiration conditions has not yet met the requirements. The acquisition time of the third pump flow curve in the current training dataset needs further optimization. That is, let n=n+1, increase the acquisition time of the third pump flow curve under approximate conditions to cover more features under approximate aspiration, and then retrain the target aspiration detection model based on the optimized training dataset. After optimizing the training dataset, for each training dataset, train and validate the aspiration detection model to be trained according to the above method, and calculate the overall average sensitivity and approximate aspiration average sensitivity for each training dataset. If both the overall average sensitivity and the approximate aspiration average sensitivity are greater than the first value, it indicates that the currently trained aspiration detection model has passed validation, the loop can be ended, and the current aspiration detection model can be used as the target aspiration detection model for subsequent detection of the ventricular assist device 100 aspiration state. Alternatively, when the acquisition time n of the third pump flow rate curve reaches the preset value, it indicates that the acquisition time in the current third pump flow rate curve has reached the maximum value. Increasing n will greatly increase the interference of the pump flow rate characteristics under normal conditions. If the suction detection model is still not verified as qualified at this time, the loop will end.

[0060] The first value and the preset value can be set according to the application scenario or actual needs. This application does not limit the specific value of the first value. The following explanation uses a first value of 0.8 and a preset value of 15 as an example.

[0061] For example, a training dataset is obtained, which includes 10 training data sets. Each training data set contains 50 first pump flow rate curves under normal conditions, 50 second pump flow rate curves under suction conditions, and 50 third pump flow rate curves under approximate suction conditions (n ​​is 3). First, the first training data set is used as the validation data set. The second to tenth training data sets are input into the suction detection model to be trained, resulting in a trained suction detection model. The first training data set is then used as the validation data set and input into the trained suction detection model, outputting 150 predicted detection results. Based on these 150 predicted detection results and 150 actual detection results, the sensitivity under normal conditions, the sensitivity under suction conditions, and the sensitivity under approximate suction conditions in the first validation data set are calculated. The second training data set is then used as the validation data set. The first, third, through tenth training data sets are input into the previously trained slurring detection model for further training, resulting in a trained slurring detection model. This second training data set is then used as the validation data set and input into the trained slurring detection model, outputting 150 predicted detection results. Based on these 150 predicted and 150 actual detection results, the sensitivity under normal conditions, slurring conditions, and approximate slurring conditions in the second validation data set are calculated. Following this method, the slurring detection model is trained again, and the sensitivity under normal, slurring, and approximate slurring conditions is calculated for each validation data set, resulting in 10 sensitivity values ​​for these conditions. The average values ​​of these 10 sensitivity values ​​are then calculated, and finally, the average of these three average sensitivities is calculated. If the calculated overall average sensitivity is ≤0.8, the current training dataset is considered too small. The training dataset is increased from one to two, each containing 10 training data sets. Each training data set still includes 50 first-stage pumping flow curves under normal conditions, 50 second-stage pumping flow curves under suction conditions, and 50 third-stage pumping flow curves under approximation suction conditions (n ​​is 3). Training and validation are then performed again using the above method. If the calculated overall average sensitivity is >0.8 and the approximation suction average sensitivity is ≤0.8, the acquisition time for the third-stage pumping flow curves in the current training dataset is considered too short. This means the suction detection model has not fully learned the core correlation between multiple features in the pumping flow curves under approximation suction conditions. The acquisition time n needs to be increased; therefore, n is increased from 3s to 4s, and the 50 third-stage pumping flow curves in the training data set are reacquired. After obtaining the new third-stage pumping flow curves, training and validation are then performed again using the above method.

[0062] The aspiration detection model is trained using the method described above until the overall average sensitivity is >0.8 and the approximate aspiration average sensitivity is >0.8. At this point, the trained aspiration detection model is considered to have reached the required accuracy, and the loop ends. The aspiration detection model at the end of the loop is used as the target aspiration detection model. Using this target aspiration detection model, the ventricular assist device can be accurately detected in real time whether it is in a normal state, aspiration state, or approximate aspiration state, thereby improving user safety. If, after multiple loops, the overall average sensitivity and the approximate aspiration average sensitivity are still not both >0.8 when n is greater than 15, the loop also ends, and an alarm is triggered, indicating that the aspiration detection model training has failed.

[0063] As can be seen, the control unit of this application acquires the target pumping flow curve, which is the pumping flow curve of the ventricular assist device within a target duration; it extracts multiple target features from the target pumping flow curve, and these multiple target features are used to measure the characteristics of the target pumping flow curve in the time and frequency domains; the multiple target features are input into the target aspiration detection model, and the target detection result is output. This application extracts time and frequency domain features from a single pumping flow curve and combines them with a machine learning model to detect the state of the ventricular assist device, enabling real-time and accurate aspiration detection of the ventricular assist device and improving user safety.

[0064] The above primarily describes the solutions of the embodiments of this application from the perspective of the method execution process. It is understood that, in order to achieve the above functions, the network device includes corresponding hardware structures and / or software modules for executing each function. Those skilled in the art should readily recognize that, in conjunction with the units and algorithm steps of the various examples described in the embodiments provided herein, this application can be implemented in hardware or a combination of hardware and computer software. Whether a function is executed in hardware or by computer software driving hardware depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0065] For example, this application embodiment proposes a ventricular assist device, the ventricular assist device comprising: impeller; A motor that drives the impeller to rotate; A control unit connected to the motor is configured to perform the following steps: Obtain the target pumping flow rate curve, which is the pumping flow rate curve of the ventricular assist device within a target duration; Multiple target features are extracted from the target pumping flow rate curve, and these multiple target features are used to measure the characteristics of the target pumping flow rate curve in the time domain and frequency domain. The multiple target features are input into the target suction detection model, and the target detection results are output.

[0066] For example, an embodiment of this application provides a medical device, which includes the control unit described above.

[0067] For example, an embodiment of this 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 configured to be executed by the processor. The programs include instructions for performing some or all of the steps performed by the control unit.

[0068] For example, embodiments of this application provide a computer-readable storage medium that stores a computer program for electronic data interchange, wherein the computer program causes a computer to perform some or all of the steps performed by a control unit.

[0069] The control unit of each of the above schemes has the function of implementing the corresponding steps executed by the control unit; the function can be implemented by hardware or by hardware executing corresponding software.

[0070] In embodiments of this application, the control unit may also be a chip or a chip system, such as a system on chip (SoC).

[0071] Please see Figure 4 , Figure 4 This is a schematic diagram of the structure of a medical device provided in an embodiment of this 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 memories and configured to be executed by the one or more processors.

[0072] The above procedure includes instructions for performing the following steps: Obtain the target pumping flow rate curve, which is the pumping flow rate curve of the ventricular assist device within a target duration; Multiple target features are extracted from the target pumping flow rate curve, and these multiple target features are used to measure the characteristics of the target pumping flow rate curve in the time domain and frequency domain. The multiple target features are input into the target suction detection model, and the target detection results are output.

[0073] All relevant content in each scenario involved in the above method embodiments can be referenced from the functional description of the corresponding functional module, and will not be repeated here.

[0074] It should be understood that the aforementioned memory may include read-only memory and random access memory, and provides instructions and data to the processor. A portion of the memory may also include non-volatile random access memory. For example, the memory may also store information about the device type.

[0075] In the embodiments of this application, the processor of the above-described device may be a central processing unit (CPU), or it may 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 any conventional processor, etc.

[0076] It should be understood that "at least one" in the embodiments of this application refers to one or more, and "more than one" refers to two or more. "And / or" describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A alone, A and B simultaneously, or B alone, where A and B can be singular or plural. The character " / " generally indicates that the preceding and following related objects are in an "or" relationship. "At least one of the following" or similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one of a, b, or c can represent: a, b, c, ab, ac, bc, or abc, where a, b, and c can be single or multiple.

[0077] Furthermore, unless otherwise stated, the ordinal numbers such as "first" and "second" mentioned in the embodiments of this application are used to distinguish multiple objects and are not used to limit the order, timing, priority, or importance of multiple objects. For example, "first information" and "second information" are only used to distinguish different information and do not indicate differences in the content, priority, sending order, or importance of these two types of information.

[0078] In implementation, the steps executed by the aforementioned control unit can be accomplished through integrated logic circuits in the processor's hardware or through software instructions. The steps executed by the control unit disclosed in the embodiments of this application can be directly manifested as execution by the hardware processor, or as a combination of hardware and software units within the processor. The software units can reside in readily available random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, or other mature storage media in the art. This storage medium is located in memory, and the processor executes the instructions in the memory, combining them with its hardware to complete the steps executed by the aforementioned control unit. To avoid repetition, further details are omitted here.

[0079] This application also provides a computer storage medium that stores a computer program for electronic data interchange, which causes a computer to perform some or all of the steps executed by a control unit.

[0080] This application also provides a computer program product, which includes a non-transitory computer-readable storage medium storing a computer program operable to cause a computer to perform some or all of the steps executed by a control unit. This computer program product can be a software installation package.

[0081] It should be noted that, for the sake of simplicity, the foregoing embodiments are all described as a series of actions. However, those skilled in the art should understand that this application is not limited to the described order of actions, as some steps may be performed in other orders or simultaneously according to this application. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are preferred embodiments, and the actions and modules involved are not necessarily essential to this application.

[0082] In the above embodiments, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.

[0083] In the several embodiments provided in this application, it should be understood that the disclosed apparatus can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of the units described above is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical or other forms.

[0084] The units described above as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of the embodiments of this application, depending on actual needs.

[0085] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0086] If the aforementioned integrated units are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage device (CMD). Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the 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 to cause a computer device (which may be a personal computer, server, or TRP, etc.) to execute all or part of the steps performed by the control unit of the various embodiments of this application. The aforementioned memory includes various media capable of storing program code, such as USB flash drives, read-only memory (ROM), random access memory (RAM), portable hard drives, magnetic disks, or optical disks.

[0087] Those skilled in the art will understand that all or part of the steps performed by the control unit in the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage device, which may include a flash drive, ROM, RAM, magnetic disk, or optical disk, etc.

[0088] The embodiments of this application have been described in detail above. Specific examples have been used to illustrate the principles and implementation methods of this application. The description of the above embodiments is only for the purpose of helping to understand the method and core ideas of this application. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this application. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A control unit for a ventricular assist device, characterized in that, The control unit is used to perform the following steps: Obtain the target pumping flow rate curve, which is the pumping flow rate curve of the ventricular assist device within a target duration; Multiple target features are extracted from the target pumping flow rate curve, and these multiple target features are used to measure the characteristics of the target pumping flow rate curve in the time domain and frequency domain. The multiple target features are input into the target suction detection model, and the target detection results are output.

2. The method according to claim 1, characterized in that, The extraction of multiple target features from the target pumping flow rate curve includes: The zero-crossing rate asymmetry, negative pulse exponent, and kurtosis of local extrema of the target pumping flow curve are calculated. The zero-crossing rate asymmetry is used to measure the balance of fluctuations in the positive and negative directions of the target pumping flow curve. The negative pulse exponent is used to measure the intensity of negative blood flow pulses generated by the ventricular assist device. The kurtosis of local extrema is used to measure the steepness of the amplitude distribution of the local maximum and local minimum sequences of the target pumping flow curve. Perform a Fourier transform on the target pump flow rate curve to obtain the flow rate spectrum. The target energy percentage is calculated based on the flow spectrum diagram, where the target energy percentage is the ratio of the energy in the target frequency band to the total energy.

3. The control unit according to claim 2, characterized in that, In terms of training the target aspiration detection model, the control unit is specifically used for: Obtain k training datasets, each training dataset comprising r training data groups, each training data group comprising multiple first pump flow curves, multiple second pump flow curves, and multiple third pump flow curves. The first pump flow curve is the pump flow curve when the ventricular assist device is in normal state, the second pump flow curve is the pump flow curve when the ventricular assist device is in aspiration state, and the third pump flow curve is the pump flow curve in the n seconds before the ventricular assist device determines that aspiration has occurred. k and r are positive integers. The target slurring detection model is obtained by training the slurring detection model to be trained using r-1 sets of training data, and the target slurring detection model is qualified by using a set of validation data. The set of validation data consists of the remaining training data sets other than the r-1 sets of training data. If the target aspiration detection model fails the verification, let k = k + 1 and repeat the above steps until the target aspiration detection model passes the verification or n is greater than the preset value. If the target aspiration detection model is verified as qualified but is in the target condition, then let n = n + 1, and repeat the above steps until the target aspiration detection model is verified as qualified or n is greater than the preset value.

4. The control unit according to claim 3, characterized in that, In the process of training the spitting detection model to obtain the target spitting detection model using r-1 sets of training data, the control unit is specifically used for: The j-th training data group is determined as the j-th validation data group, where j is a positive integer; The target suction detection model is obtained by training the remaining r-1 training data sets excluding the j-th verification data set. Input the j-th verification data set into the target aspiration detection model and output the j-th detection result; Calculate the j-th normal sensitivity, j-th aspiration sensitivity, and j-th approximation aspiration sensitivity based on the j-th detection result and the actual result, respectively. Let j = j + 1, and repeat the above steps until j = r.

5. The control unit according to claim 4, characterized in that, In determining whether the target aspiration detection model has passed verification using the verification data set, the control unit is specifically used for: Calculate the overall average sensitivity and the approximate aspiration average sensitivity, wherein the approximate aspiration average sensitivity is the average value of the j-th approximate aspiration sensitivity, and the overall average sensitivity is the average of the approximate aspiration average sensitivity, the normal average sensitivity, and the aspiration aspiration sensitivity. If the overall average sensitivity is less than the first value, the target aspiration detection model is deemed unqualified.

6. The control unit according to claim 5, characterized in that, The target aspiration detection model is verified as qualified but is in the target condition if the overall average sensitivity is greater than or equal to the first value and the approximation aspiration average sensitivity is less than the first value.

7. The control unit according to any one of claims 1-6, characterized in that, The target suction detection model is a support vector machine classification model.

8. A ventricular assist device, characterized in that, The ventricular assist device includes: impeller; A motor that drives the impeller to rotate; A control unit connected to the motor is configured to perform the following steps: Obtain the target pumping flow rate curve, which is the pumping flow rate curve of the ventricular assist device within a target duration; Multiple target features are extracted from the target pumping flow rate curve, and these multiple target features are used to measure the characteristics of the target pumping flow rate curve in the time domain and frequency domain. The multiple target features are input into the target suction detection model, and the target detection results are output.

9. A medical device, characterized in that, The medical device includes a control unit as described in any one of claims 1-7.

10. A medical device, characterized in that, It includes 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, the one or more programs including instructions for performing the steps of the control unit as described in any one of claims 1-7.

Images