A method and device for detecting suction events of a ventricular catheter pump

Abstract

An embodiment of the present invention provides a method and device for detecting aspiration events of a ventricular catheter pump, which relates to the technical field of medical devices. The above method includes: during the operation of the ventricular catheter pump, obtaining the rotational speed data and current data of the ventricular catheter pump, and obtaining the flow rate data of the blood passing through the ventricular catheter pump; for each type of target data, based on the data sequence order of the target data of this type, extracting aspiration features from the target data of this type, wherein the types of the target data include: rotational speed data type, current data type, and blood flow rate data type; based on the extracted aspiration features of the target data, predicting whether an aspiration event occurs in the ventricular catheter pump. When applying the solution provided in this embodiment, it is possible to detect whether an aspiration event occurs in the ventricular catheter pump.

Description

Technical Field

[0001] The present invention relates to the technical field of medical devices, and in particular, to a method and device for detecting suction events of a ventricular catheter pump. Background Art

[0002] A ventricular catheter pump is a miniaturized intravascular axial flow pump used to support a patient's blood circulation system. Taking the left ventricular catheter pump as an example, the left ventricular catheter pump can be implanted into the left ventricle of a patient. When the left ventricular catheter pump is in a normal operating state, the left ventricular catheter pump can transport blood from the inlet area located in the left ventricle to the ascending aorta outlet through a catheter. The left ventricular catheter pump can assist in increasing cardiac output, raising aortic pressure and coronary perfusion pressure, and improving mean arterial pressure and coronary blood flow.

[0003] However, due to the influence of in-vivo environmental factors, suction events are likely to occur when the ventricular catheter pump operates in a patient. When a suction event occurs in the ventricular catheter pump, it causes damage to the heart. Therefore, there is an urgent need for a detection scheme to detect whether a suction event occurs in the ventricular catheter pump. Summary of the Invention

[0004] The purpose of the embodiments of the present invention is to provide a method and device for detecting suction events of a ventricular catheter pump to detect whether a suction event occurs in the ventricular catheter pump. The specific technical solutions are as follows:

[0005] In a first aspect, the embodiments of the present invention provide a method for detecting suction events of a ventricular catheter pump, the method including:

[0006] During the operation of the ventricular catheter pump, obtain the rotational speed data and current data of the ventricular catheter pump, and obtain the flow rate data of the blood passing through the ventricular catheter pump;

[0007] For each type of target data, based on the data sequence order of the target data of this type, extract suction characteristics from the target data of this type, where the types of the target data include: rotational speed data type, current data type, and blood flow rate data type;

[0008] Based on the suction characteristics of the extracted target data, predict whether a suction event occurs in the ventricular catheter pump.

[0009] In an embodiment of the present invention, the suction characteristics of each type of target data are extracted in the following manner:

[0010] Extract suction characteristics from the target data in the forward order of the data sequence of the target data, and determine the extracted characteristics as the first sub-characteristics;

[0011] Extract the aspiration features of the target data in reverse order according to the data sequence of the target data, and determine the extracted features as the second sub-features;

[0012] Based on the first sub-features and the second sub-features, determine the aspiration features of the target data.

[0013] In one embodiment of the present invention, the above-mentioned extraction of the aspiration features of the target data in the forward order of the data sequence of the target data and determining the extracted features as the first sub-features includes:

[0014] Extract the aspiration features of each data included in the target data;

[0015] In the forward order of the data sequence, for each data in the target data except the first data, based on the aspiration features of the data obtained by extraction, update the first type of aspiration features updated by the previous data of the data, where the first data is: the data included in the target data that is in the first position in the forward order of the data sequence, and the first type of aspiration features is: the features obtained by extracting the aspiration features of the first data;

[0016] Determine the first sub-features as the first type of aspiration sub-features updated by the second data included in the target data, where the second data is: the data included in the target data that is in the last position in the forward order of the data sequence.

[0017] In one embodiment of the present invention, the above-mentioned extraction of the aspiration features of the target data in reverse order according to the data sequence of the target data and determining the extracted features as the second sub-features includes:

[0018] Extract the aspiration features of each data included in the target data;

[0019] In the reverse order of the data sequence, for each data in the target data except the third data, based on the aspiration features of the data obtained by extraction, update the second type of aspiration features updated by the previous data of the data, where the third data is: the data included in the target data that is in the first position in the reverse order of the data sequence, and the second type of aspiration features is: the features obtained by extracting the aspiration features of the third data;

[0020] Determine the second sub-features as the second type of aspiration sub-features updated by the fourth data included in the target data, where the fourth data is: the data included in the target data that is in the last position in the reverse order of the data sequence.

[0021] In one embodiment of the present invention, the above-mentioned prediction of whether the ventricular catheter pump has an aspiration event based on the extracted aspiration features of the target data includes:

[0022] Perform feature fusion on the suction characteristics of each type of target data;

[0023] Calculate the matching degree between the fused suction characteristics and the preset suction characteristics;

[0024] Determine whether a suction event occurs in the ventricular catheter pump according to the calculated matching degree.

[0025] In a second aspect, an embodiment of the present invention provides a suction event detection device for a ventricular catheter pump, and the device includes:

[0026] A data acquisition module, configured to acquire the rotational speed data and current data of the ventricular catheter pump during the operation of the ventricular catheter pump, and acquire the flow rate data of the blood passing through the ventricular catheter pump;

[0027] A feature extraction module, configured to perform suction feature extraction on each type of target data based on the data sequence order of the target data of this type, where the types of the target data include: rotational speed data type, current data type, and blood flow rate data type;

[0028] An event detection module, configured to predict whether a suction event occurs in the ventricular catheter pump based on the suction characteristics of the extracted target data.

[0029] In an embodiment of the present invention, the above feature extraction module includes:

[0030] A first feature extraction unit, configured to perform suction feature extraction on the target data in the forward order of the data sequence of the target data, and determine the extracted feature as the first sub-feature;

[0031] A second feature extraction unit, configured to perform suction feature extraction on the target data in the reverse order of the data sequence of the target data, and determine the extracted feature as the second sub-feature;

[0032] A feature determination unit, configured to determine the suction feature of the target data based on the first sub-feature and the second sub-feature.

[0033] In one embodiment of the present invention, the above-mentioned first feature extraction unit is specifically configured to extract suction features for each data included in the target data; in the forward order of the data sequence, for each data in the target data except the first data, based on the extracted suction features of this data, update the first type of suction features updated by the previous data of this data, where the first data is the data that is located in the first position in the forward order of the data sequence included in the target data, and the first type of suction features is the features obtained by extracting suction features for the first data; determine the first sub-features updated by the second data included in the target data as the first sub-features, where the second data is the data that is located in the last position in the forward order of the data sequence included in the target data.

[0034] In one embodiment of the present invention, the above-mentioned second feature extraction unit is specifically configured to extract suction features for each data included in the target data; in the reverse order of the data sequence, for each data in the target data except the third data, based on the extracted suction features of this data, update the second type of suction features updated by the previous data of this data, where the third data is the data that is located in the first position in the reverse order of the data sequence included in the target data, and the second type of suction features is the features obtained by extracting suction features for the third data; determine the second sub-features updated by the fourth data included in the target data as the second sub-features, where the fourth data is the data that is located in the last position in the reverse order of the data sequence included in the target data.

[0035] In one embodiment of the present invention, the above-mentioned event detection module is specifically configured to perform feature fusion on the suction features of each type of target data; calculate the matching degree between the fused suction features and the preset suction features; and determine whether the ventricular catheter pump has a suction event according to the calculated matching degree.

[0036] In a third aspect, an embodiment of the present invention provides an electronic device, including a processor, a communication interface, a memory, and a communication bus, where the processor, the communication interface, and the memory complete communication with each other through the communication bus;

[0037] The memory is used to store a computer program;

[0038] The processor is configured to implement the method steps described in the first aspect when executing the program stored on the memory.

[0039] In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium, where a computer program is stored in the computer-readable storage medium, and when the computer program is executed by a processor, the method steps described in the first aspect are implemented.

[0040] As can be seen from the above, when applying the solution provided by the embodiments of the present invention, since the detection of whether a suction event occurs in the ventricular catheter pump is based on the rotational speed data, current data of the ventricular catheter pump, and the flow rate data of the blood passing through the ventricular catheter pump, and since the rotational speed, current, and the flow rate of the blood passing through the ventricular catheter pump all affect whether a suction event will occur in the ventricular catheter pump, therefore, by integrating three different types of detection indicators, the detection can be carried out comprehensively and accurately, improving the accuracy of the detection result.

[0041] In addition, since the extraction of the suction characteristics of the target data is based on the data sequence order of the target data, and the above data sequence order can reflect the relationship information between the data included in each type of target data, therefore, when extracting the suction characteristics, in addition to considering the target data itself, the relationship information between the data included in the target data is also considered. By integrating the above two types of information, the suction characteristics of the target data can be extracted more accurately, thereby improving the accuracy of the detection result.

[0042] Of course, when implementing any product or method of the present invention, it is not necessarily required to achieve all the above-mentioned advantages simultaneously. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings in the following description are only some embodiments of the present application, and those of ordinary skill in the art can also obtain other embodiments according to these drawings.

[0044] Figure 1 It is a schematic flowchart of the first method for detecting a suction event of a ventricular catheter pump provided by an embodiment of the present invention;

[0045] Figure 2 It is a schematic flowchart of the second method for detecting a suction event of a ventricular catheter pump provided by an embodiment of the present invention;

[0046] Figure 3 It is a schematic diagram of the first feature update process provided by an embodiment of the present invention;

[0047] Figure 4 It is a schematic diagram of the second feature update process provided by an embodiment of the present invention;

[0048] Figure 5 It is a schematic structural diagram of a device for detecting a suction event of a ventricular catheter pump provided by an embodiment of the present invention;

[0049] Figure 6 It is a schematic structural diagram of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0050] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art based on this application belong to the scope of protection of the present invention.

[0051] First, before specifically describing the solutions provided in the embodiments of the present invention, the application scenarios and the execution entities of the embodiments of the present invention are introduced.

[0052] The application scenario of the embodiments of the present invention is: the application scenario where a ventricular catheter pump operates in a patient's body. The above ventricular catheter pump can be a left ventricular catheter pump.

[0053] The execution entity of the embodiments of the present invention is: a ventricular assist device. The above ventricular assist device is used to collect the operation parameter data of the ventricular catheter pump and control the operation of the ventricular catheter pump.

[0054] Of course, the execution entity of the embodiments of the present invention can also be: an electronic device that performs data interaction with the ventricular assist device connected to the ventricular catheter pump. The above electronic device can be a server, a cloud server, etc.

[0055] The following specifically describes the suction event detection method for the ventricular catheter pump provided in the embodiments of the present invention.

[0056] See Figure 1 , Figure 1 which is a schematic flow chart of the first suction event detection method for the ventricular catheter pump provided in the embodiments of the present invention. The above method includes the following steps S101 - S103.

[0057] Step S101: During the operation of the ventricular catheter pump, obtain the rotation speed data and current data of the ventricular catheter pump, and obtain the flow rate data of the blood passing through the ventricular catheter pump.

[0058] The above operation process of the ventricular catheter pump can be the operation process when the ventricular catheter pump operates in a patient's body during actual application, or the operation process when the ventricular catheter pump operates in a simulation environment during simulation application. The above simulation environment can be a heart simulation environment, etc.

[0059] The rotation speed data refers to the rotation speed of the ventricular catheter pump, and the current data refers to the motor current of the ventricular catheter pump.

[0060] For the convenience of description below, the rotation speed data, current data, and flow rate data are uniformly named target data.

[0061] The above-mentioned target data can be data at each unit moment within a cardiac cycle, where the cardiac cycle refers to the time period between the start times of two adjacent heartbeats, and the above-mentioned unit moment can be 1 ms, 10 ms, etc.

[0062] During the operation of the ventricular catheter pump, the ventricular assist device can monitor the operation of the ventricular catheter pump in real time. Based on this, when the execution entity of this embodiment is the ventricular assist device, the above-mentioned target data can be directly obtained; when the execution entity of this embodiment is the server, the ventricular assist device can send the above-mentioned target data to the server.

[0063] Step S102: For each type of target data, based on the data sequence order of this type of target data, extract the suction characteristics of this type of target data.

[0064] The types of the above-mentioned target data include: rotational speed data type, current data type, and blood flow data type.

[0065] The data sequence order of the target data refers to: the sequence order formed according to the generation time of each data included in the target data. For example: taking the target data as flow data, the obtained flow data includes the flow data generated at each unit moment within the cardiac cycle, and the above-mentioned data sequence order is the sequence order formed according to the generation time of each flow data.

[0066] Since the suction characteristics of the target data are extracted based on the above-mentioned data sequence order, and since the data sequence order can reflect the information of the data adjacent to each target data, therefore, when extracting the suction characteristics, in addition to considering the target data itself, the adjacent data of the target data is also considered. By integrating these two types of information, the suction characteristics of the target data can be more accurately extracted.

[0067] When extracting the suction characteristics, in one implementation, the feature extraction algorithm can be first used to extract the features of each data included in the target data, and then, according to the data sequence order of the target data, perform feature fusion on each extracted suction feature, and determine the feature located in the first preset layer among the fused features as the suction feature of the target data. The above-mentioned feature extraction algorithm can be the principal component analysis method, the linear discriminant analysis method, etc. The above-mentioned first preset layer refers to: the feature layer corresponding to the suction feature determined in advance. For example, the above-mentioned first preset layer can be the fifth layer.

[0068] Other ways to extract the suction characteristics can be seen in the subsequent Figure 2 corresponding embodiments, which will not be elaborated here.

[0069] Step S103: Based on the extracted suction characteristics of the target data, predict whether a suction event occurs in the ventricular catheter pump.

[0070] When predicting whether a suction event occurs in a ventricular catheter pump, the prediction can be carried out in the following two implementation manners:

[0071] In the first implementation manner, the suction characteristics of each type of target data can be subjected to feature fusion; the matching degree between the fused suction characteristics and the preset suction characteristics can be calculated; and whether a suction event occurs in the ventricular catheter pump can be determined according to the calculated matching degree.

[0072] When performing feature fusion, the suction characteristics of each type of target data can be weighted and summed according to the preset weight of each type of target data, and the calculated features can be used as the fused suction characteristics.

[0073] The above-mentioned preset suction characteristics refer to the characteristics of the suction event determined in advance.

[0074] When calculating the matching degree, the distance between the preset suction characteristics and the fused suction characteristics can be calculated. The above-mentioned distance can be an Euclidean distance, a cosine distance, etc., and according to the corresponding relationship between the preset distance and the matching degree, the calculated distance can be converted into a matching degree.

[0075] When determining whether a suction event occurs in the ventricular catheter pump, in one implementation manner, the relationship between the calculated matching degree and the preset matching degree threshold is judged. If the matching degree is greater than or equal to the preset matching degree threshold, it is determined that a suction event occurs in the ventricular catheter pump; if the matching degree is less than the preset matching degree threshold, it is determined that no suction event occurs in the ventricular catheter pump.

[0076] Since the fused suction characteristics are obtained by feature fusion of the suction characteristics of each target data, and the fused suction characteristics comprehensively reflect the suction characteristics of each target data, therefore, by using the fused suction characteristics, it is possible to more accurately detect whether a suction event occurs in the ventricular catheter pump.

[0077] In the second implementation manner, the matching degree between the suction characteristics of each type of target data and the preset suction characteristics can be calculated, and whether a suction event occurs in the ventricular catheter pump can be determined based on the matching degree corresponding to each type of target data.

[0078] When calculating the matching degree, the distance between the preset suction characteristics and the suction characteristics of the target data can be calculated. The above-mentioned distance can be an Euclidean distance, a cosine distance, etc., and according to the corresponding relationship between the preset distance and the matching degree, the calculated distance can be converted into a matching degree.

[0079] When determining whether a suction event occurs in a ventricular catheter pump, in one implementation, the number of target data with a matching degree greater than a preset matching degree threshold can be determined. If the determined number is greater than or equal to a preset number threshold, it is determined that a suction event occurs in the ventricular catheter pump. If the determined number is less than the preset number threshold, it is determined that no suction event occurs in the ventricular catheter pump.

[0080] For example: the preset number threshold is 2, and the target data with a matching degree greater than the preset matching degree threshold includes: rotational speed data and current data. That is, the number of target data with a matching degree greater than the preset matching degree threshold is 2. Since the above number is equal to the preset number threshold 2, it can be determined that a suction event occurs in the ventricular catheter pump.

[0081] As can be seen from the above, when applying the solution provided in this embodiment, since it is based on the rotational speed data, current data of the ventricular catheter pump, and the flow rate data of the blood passing through the ventricular catheter pump to detect whether a suction event occurs in the ventricular catheter pump, and since the rotational speed, current of the ventricular catheter pump, and the flow rate of the blood passing through the ventricular catheter pump all affect whether a suction event will occur in the ventricular catheter pump, therefore, by combining three different types of detection indicators, it can be comprehensively and accurately detected, improving the accuracy of the detection result.

[0082] In addition, since the suction characteristics of the target data are extracted based on the data sequence order of the target data, and the above data sequence order can reflect the relationship information between the data included in each type of target data, therefore, when performing suction characteristic extraction, in addition to considering the target data itself, the relationship information between the data included in the target data is also considered. By combining the above two types of information, the suction characteristics of the target data can be more accurately extracted, thereby improving the accuracy of the detection result.

[0083] In the foregoing Figure 1 In the corresponding embodiment, in addition to the implementation methods mentioned above, a deep learning method can also be combined to detect the suction event of the ventricular catheter pump. Based on this, in one embodiment of the present invention, during the operation of the ventricular catheter pump, the rotational speed data and current data of the ventricular catheter pump are obtained, and the flow rate data of the blood passing through the ventricular catheter pump is obtained. The target data is input into the detection model, and the detection result indicating whether a suction event occurs in the ventricular catheter pump output by the detection model is obtained.

[0084] The above detection model is a neural network model obtained by training an initial neural network model with the sample target data of the sample ventricular catheter pump as the training sample, the actual result indicating whether a suction event occurs in the sample ventricular catheter pump as the training benchmark, and is used to detect whether a suction event occurs in the ventricular catheter pump.

[0085] The above sample target data includes: during the operation of the sample ventricular catheter pump, the rotational speed data and current data of the sample ventricular catheter pump, and the flow rate data of the blood passing through the sample ventricular catheter pump.

[0086] The above initial neural network model mainly includes a convolutional layer and an LSTM (Long Short-Term Memory) layer. Among them, the convolutional layer is used to perform preliminary feature extraction on the input data, and the LSTM layer is used to further accurately extract features based on the data sequence order of the input data.

[0087] When training the above initial neural network model, after the sample target data is input into the model, first perform time-frequency transformation on the sample target data to facilitate more accurate extraction of suction features in the subsequent process; then, the convolutional layer performs feature extraction on the sample target data after time-frequency transformation. In this process, batch normalization can be performed. Batch normalization can accelerate training and has a regularization feature to prevent overflow; secondly, reallocate the features output by the convolutional layer into a sequence and enter the bidirectional LSTM layer and the unidirectional LSTM layer. The LSTM layer performs feature extraction based on the sequence, thereby combining memory with data to improve the accuracy of prediction; finally, the features output by the LSTM layer enter the fully connected layer. The fully connected layer maps the features and uses the sigmoid binary activation function to obtain the detection result indicating whether a suction event occurs in the sample ventricular catheter pump. Compare the matching degree between the detection result and the training benchmark, and adjust the model parameters of the initial neural network model based on the matching degree until the model converges to obtain the detection model.

[0088] In the foregoing Figure 1 In step S102 of the corresponding embodiment, in addition to the feature extraction method mentioned above for suction feature extraction, it can also be implemented according to the following Figure 2 corresponding steps S202 - S204 of the embodiment.

[0089] See Figure 2 , Figure 2 is a schematic flowchart of the second method for detecting suction events of the ventricular catheter pump provided by the embodiment of the present invention. On the basis of the foregoing Figure 1 corresponding embodiment, Figure 2 the corresponding embodiment includes the following steps S201 - S205.

[0090] Step S201: During the operation of the ventricular catheter pump, obtain the rotational speed data and current data of the ventricular catheter pump, and obtain the flow rate data of the blood passing through the ventricular catheter pump.

[0091] The above step S201 is the same as step S101 in the foregoing Figure 1 corresponding embodiment and will not be elaborated here.

[0092] The following steps S202 - S204 are for extracting suction features for each type of target data.

[0093] Step S202: Extract suction features from the target data in the forward order of the data sequence of the target data, and determine the extracted features as the first sub - features.

[0094] The forward order of the data sequence means: the data sequence order formed according to the forward time order of the generation time of each data included in the target data. The forward order of the data sequence can reflect the relationship between each data included in the target data and its adjacent data from the perspective of the forward time order.

[0095] For example, the generation time of each data included in the target data is shown in Table 1 below.

[0096] Table 1

[0097] D1 D2 D3 D4 D5 D6 13:00 13:01 13:02 13:03 13:04 13:05

[0098] In the above Table 1, the first - row data represents each data included in the target data, and the second row represents the generation time of the corresponding each data. On this basis, the forward time order of the generation time of the above data is: 13:00, 13:01, 13:02, 13:03, 13:04, 13:05. Therefore, the forward order of the data sequence of the target data is: D1, D2, D3, D4, D5, D6.

[0099] In determining the first sub - features, in one implementation, the feature extraction algorithm can be first used to extract features from each data included in the target data, and then, according to the forward order of the data sequence of the target data, feature fusion is performed on each extracted feature, and the feature located in the third preset layer among the fused features is determined as the first sub - feature. The above - mentioned feature extraction algorithm can be the principal component analysis method, the linear discriminant analysis method, etc. The above - mentioned third preset layer means: the feature layer corresponding to the suction features determined in advance. For example, the above - mentioned third preset layer can be the third layer.

[0100] Other implementations for determining the first sub - features can be seen in the subsequent embodiments and will not be elaborated here.

[0101] Step S203: Extract suction features from the target data in the reverse order of the data sequence of the target data, and determine the extracted features as the second sub - features.

[0102] The reverse order of the data sequence means: the order of the data sequence formed according to the reverse chronological order of the generation time of each data included in the target data. The reverse order of the data sequence can reflect the relationship between each data included in the target data and the adjacent data from the perspective of the reverse chronological order. Continuing with Table 1 above, the reverse order of the data sequence of the target data is: D6, D5, D4, D3, D2, D1.

[0103] In one implementation manner for determining the second sub-feature, first, a feature extraction algorithm can be used to extract features from each data included in the target data, and then, according to the reverse order of the data sequence of the target data, feature fusion is performed on each extracted feature, and the feature located at the fourth preset layer in the fused features is determined as the second sub-feature. The above feature extraction algorithm can be the principal component analysis method, the linear discriminant analysis method, etc. The above fourth preset layer refers to: the feature layer corresponding to the extracted feature determined in advance. For example, the above fourth preset layer can be the second layer, the fourth layer, etc.

[0104] For other implementation manners of determining the second sub-feature, reference can be made to the subsequent embodiments, which will not be elaborated here.

[0105] Step S204: Determine the extraction feature of the target data based on the first sub-feature and the second sub-feature.

[0106] In one implementation manner, feature fusion can be performed on the first sub-feature and the second sub-feature, such as feature addition, feature splicing, and the fused feature is determined as the extraction feature of the target data.

[0107] Step S205: Predict whether a suction event occurs in the ventricular catheter pump based on the extraction feature of the target data obtained.

[0108] The above step S205 is the same as step S103 in the corresponding Figure 1 embodiment and will not be elaborated again.

[0109] As can be seen from the above, since the first sub-feature is the pumping feature determined according to the forward order of the data sequence, and the second sub-feature is the pumping feature determined according to the reverse order of the data sequence, the forward order of the data sequence can reflect the relationship between each data included in the target data and its adjacent data from the perspective of the forward order of time, and the reverse order of the data sequence can reflect the relationship between each data included in the target data and its adjacent data from the perspective of the reverse order of time. Also, since the pumping feature is determined based on the above first sub-feature and second sub-feature, the pumping feature can both reflect the relationship between each data included in the target data and its adjacent data from the perspective of the forward order of time and reflect the relationship between each data included in the target data and its adjacent data from the perspective of the reverse order of time. Therefore, the information reflected by the above pumping feature is more comprehensive and rich, which makes the accuracy of the extracted pumping feature relatively high, and further improves the accuracy of the detection result.

[0110] The foregoing Figure 2 In the corresponding embodiment step S202, in addition to extracting the first sub-feature by using the foregoing implementation method, it can also be implemented according to the following steps A1-A3.

[0111] Step A1: Extract the pumping feature for each data in the target data.

[0112] When extracting the pumping feature, in one implementation, a feature extraction algorithm can be used to extract features for each data included in the target data, and the feature located at the fifth preset layer among the extracted features is determined as the pumping feature of each data included in the target data. The above feature extraction algorithm can be the principal component analysis method, the linear discriminant analysis method, etc. The above second preset layer refers to the feature layer corresponding to the preset pumping feature. For example, the above fifth preset layer can be the second layer, the third layer, etc.

[0113] Step A2: According to the forward order of the data sequence, for each data in the target data except the first data, based on the extracted pumping feature of this data, update the first type of pumping feature updated for the data preceding this data.

[0114] The above first data is: the data included in the target data that is in the first position in the forward order of the data sequence. Continuing with the foregoing Table 1, the target data arranged in the forward order of the data sequence is: D1, D2,..., D6. The data included in the target data that is in the first position in the forward order of the data sequence is D1. Therefore, D1 is the first data.

[0115] The above first type of pumping feature is: the feature obtained by extracting the pumping feature for the first data.

[0116] When updating the first type of pumping feature, the first type of pumping feature is updated in sequence according to the forward order of the data sequence of each data included in the target data. The following is combined with Figure 3 to illustrate the above process.

[0117] Figure 3 FIG. 6 is a schematic diagram of the first feature update process provided by an embodiment of the present invention. Figure 3 Continuing to use the target data shown in Table 1 above, Figure 3 each rectangular box in FIG. 6 represents each data included in the target data. Among them, D1 is the first data, D2-D6 are each data in the target data except the first data, vector d1 is the first type of pumping sub-feature, and vectors d2-d6 are the pumping sub-features corresponding to D2-D6 respectively.

[0118] When performing feature update, first, for the data D2 located in the second place. The previous data of D2 is D1. When performing feature update, based on the pumping feature d2 of D2, the first type of pumping sub-feature d1 is updated. In this case, the first type of pumping sub-feature has been updated to the first type of pumping sub-feature updated by D2.

[0119] Secondly, for the data D3 located in the third place. The previous data of D3 is D2. When performing feature update, based on the pumping feature d3 of D3, the first type of pumping sub-feature updated by D2 is updated. In this case, the first type of pumping sub-feature has been updated to the first type of pumping sub-feature updated by D3.

[0120] In the above manner, until the first type of pumping sub-feature updated by the sixth data D6 is obtained, the above feature update process ends.

[0121] In one implementation manner, when performing feature update, feature fusion can be performed on two types of features, and the fused feature is determined as the updated first type of pumping feature. The above feature fusion method can include feature addition, feature splicing, etc.

[0122] Step A3: Determine the first type of pumping sub-feature updated by the second data included in the target data as the first sub-feature.

[0123] The above second data is: the data located at the last place in the forward order of the data sequence among the data included in the target data. Continuing to use Table 1 above, the target data arranged in the forward order of the data sequence is: D1, D2,..., D6. The data located at the last place in the forward order of the data sequence among the data included in the target data is D6. Therefore, D6 is the second data.

[0124] Taking Figure 3 as an example, Figure 3The middle vector d6 is the first type of pumping sub-feature updated by the data D6, so the first sub-feature is the vector d6.

[0125] Since it is in the forward order of the data sequence, based on the pumping features of each data included in the extracted target data, the first type of pumping feature is iteratively updated, so that the obtained first sub-feature comprehensively reflects the information of each data.

[0126] The aforementioned Figure 2 In the corresponding step S203 of the embodiment, in addition to extracting the second sub-feature in the aforementioned manner, it can also be implemented according to the following steps B1 - B3.

[0127] Step B1: Extract the pumping feature for each data included in the target data.

[0128] When extracting the pumping feature, in one implementation, a feature extraction algorithm can be used to extract the feature for each data included in the target data, and the feature located in the sixth preset layer among the extracted features is determined as the pumping feature of each data included in the target data. The above feature extraction algorithm can be the principal component analysis method, the linear discriminant analysis method, etc. The above second preset layer refers to: the feature layer corresponding to the preset pumping feature. For example, the above sixth preset layer can be the second layer, the fourth layer, etc.

[0129] Step B2: In the reverse order of the data sequence, for each data in the target data except the third data, based on the pumping feature of this data extracted, update the second type of pumping feature updated by the previous data of this data.

[0130] The above third data is: the data that is in the first position in the reverse order of the data sequence among the data included in the target data. Continuing with the aforementioned Table 1, the target data arranged in the reverse order of the data sequence is: D6, D5,..., D1, and the data that is in the first position in the reverse order of the data sequence among the data included in the target data is D6. Therefore, D6 is the third data.

[0131] The above second type of pumping feature is: the feature obtained by extracting the pumping feature of the third data.

[0132] When updating the second type of pumping feature, the second type of pumping feature is updated in turn according to the reverse order of the data sequence of each data included in the target data. The following combines Figure 4 To illustrate the above process.

[0133] Figure 4 It is a schematic diagram of the second feature update process provided by the embodiment of the present invention. Figure 4 Continuing with the target data shown in the aforementioned Table 1. Figure 4Each rectangular box represents each piece of data included in the target data. Among them, D6 is the third data, D5 - D1 are each piece of data in the target data except the third data, vector t1 is the second type of pumping sub - feature, and vectors t2 - t6 are the pumping sub - features corresponding to D5 - D1 respectively.

[0134] When performing feature update, first, for the data D5 located in the second position. The previous data of D5 is D6. When performing feature update, based on the pumping feature t2 of D5, the second type of pumping sub - feature t1 is updated. In this case, the second type of pumping sub - feature has been updated to the second type of pumping sub - feature updated by D5.

[0135] Secondly, for the data D4 located in the third position. The previous data of D4 is D5. When performing feature update, based on the pumping feature t3 of D4, the second type of pumping sub - feature updated by D5 is updated. In this case, the second type of pumping sub - feature has been updated to the second type of pumping sub - feature updated by D4.

[0136] In the above - mentioned manner, until the second type of pumping sub - feature updated by the sixth - position data D1 is obtained, the above - mentioned feature update process ends.

[0137] In one implementation, when performing feature update, feature fusion can be performed on two types of features, and the fused feature is determined as the updated second type of pumping feature. The above - mentioned feature fusion methods can include feature addition, feature splicing, etc.

[0138] Step B3: Determine the second - type pumping sub - feature updated by the fourth data included in the target data as the second sub - feature.

[0139] The above - mentioned fourth data is: the data located at the last position in the reverse order of the data sequence among the data included in the target data. Continuing with Table 1 mentioned above, the target data arranged in the reverse order of the data sequence is: D6, D5, ……, D1. The data located at the last position in the reverse order of the data sequence among the data included in the target data is D1. Therefore, D1 is the fourth data.

[0140] Take Figure 4 as an example. Figure 4 Among them, vector t6 is the second - type pumping sub - feature updated by data D1. Therefore, the second sub - feature is vector t6.

[0141] Since it is in the reverse order of the data sequence, and based on the pumping features of each piece of data included in the target data extracted, the second - type pumping feature is iteratively updated, so that the obtained second sub - feature comprehensively reflects the information of each piece of data.

[0142] Corresponding to the above suction event detection method of the ventricular catheter pump, an embodiment of the present invention further provides a suction event detection device for a ventricular catheter pump.

[0143] See Figure 5 , Figure 5 which is a schematic structural diagram of a suction event detection device for a ventricular catheter pump provided by an embodiment of the present invention. The above device includes the following modules 501-503.

[0144] A data acquisition module 501, configured to acquire the rotation speed data and current data of the ventricular catheter pump during the operation of the ventricular catheter pump, and acquire the flow rate data of the blood passing through the ventricular catheter pump;

[0145] A feature extraction module 502, configured to perform suction feature extraction on each type of target data based on the data sequence order of the target data of this type, where the types of the target data include: rotation speed data type, current data type, and blood flow rate data type;

[0146] An event detection module 503, configured to predict whether a suction event occurs in the ventricular catheter pump based on the suction features of the extracted target data.

[0147] As can be seen from the above, when applying the solution provided by this embodiment, since the detection of whether a suction event occurs in the ventricular catheter pump is based on the rotation speed data, current data of the ventricular catheter pump, and the flow rate data of the blood passing through the ventricular catheter pump, and since the rotation speed, current, and flow rate of the blood passing through the ventricular catheter pump all affect whether a suction event will occur in the ventricular catheter pump, therefore, by integrating three different detection indexes, the detection can be carried out comprehensively and accurately, and the accuracy of the detection result is improved.

[0148] In addition, since the suction feature extraction of the target data is performed based on the data sequence order of the target data, the above data sequence order can reflect the relationship information between the data included in each type of target data. Therefore, when performing suction feature extraction, in addition to considering the target data itself, the relationship information between the data included in the target data is also considered. By integrating the above two types of information, the suction features of the target data can be extracted more accurately, thereby improving the accuracy of the detection result.

[0149] In an embodiment of the present invention, the above feature extraction module 502 includes:

[0150] A first feature extraction unit, configured to perform suction feature extraction on the target data in the forward order of the data sequence of the target data, and determine the extracted feature as the first sub-feature;

[0151] A second feature extraction unit, configured to extract suction features from the target data in reverse order of the data sequence of the target data, and determine the extracted features as second sub-features;

[0152] A feature determination unit, configured to determine the suction feature of the target data based on the first sub-feature and the second sub-feature.

[0153] As can be seen from the above, since the first sub-feature is the suction feature determined in the forward order of the data sequence, and the second sub-feature is the suction feature determined in the reverse order of the data sequence, the forward order of the data sequence can reflect the relationship between each data included in the target data and its adjacent data from the perspective of the forward time order, and the reverse order of the data sequence can reflect the relationship between each data included in the target data and its adjacent data from the perspective of the reverse time order. Also, since the suction feature is determined based on the above first sub-feature and second sub-feature, the suction feature can reflect both the relationship between each data included in the target data and its adjacent data from the perspective of the forward time order and the relationship between each data included in the target data and its adjacent data from the perspective of the reverse time order. Therefore, the information reflected by the above suction feature is more comprehensive and rich, so that the accuracy of the extracted suction feature is relatively high, and further the accuracy of the detection result is improved.

[0154] In an embodiment of the present invention, the above first feature extraction unit is specifically configured to extract suction features for each data included in the target data; in the forward order of the data sequence, for each data in the target data except the first data, based on the suction feature of the data extracted, update the first type of suction feature updated for the data preceding the data; where the first data is the data that is the first in the forward order of the data sequence among the data included in the target data, and the first type of suction feature is the feature obtained by extracting the suction feature for the first data; determine the first type of suction sub-feature updated for the second data included in the target data as the first sub-feature, where the second data is the data that is the last in the forward order of the data sequence among the data included in the target data.

[0155] Since it is in the forward order of the data sequence and the first type of suction feature is iteratively updated based on the suction feature of each data included in the target data extracted, the obtained first sub-feature comprehensively reflects the information of each data.

[0156] In one embodiment of the present invention, the above-mentioned second feature extraction unit is specifically configured to extract suction features for each piece of data included in the target data; in the reverse order of the data sequence, for each piece of data in the target data except the third data, based on the suction features of this piece of data extracted, update the second type of suction features updated by the previous piece of data located at this piece of data, where the third data is: the piece of data located first in the reverse order of the data sequence among the data included in the target data, and the second type of suction features is: the features obtained by extracting suction features for the third data; determine the second sub-features updated by the fourth data included in the target data as the second sub-features, where the fourth data is: the piece of data located last in the reverse order of the data sequence among the data included in the target data.

[0157] Since it is in the reverse order of the data sequence and iteratively updates the second type of suction features based on the suction features of each piece of data included in the extracted target data, the obtained second sub-features can comprehensively reflect the information of each piece of data.

[0158] In one embodiment of the present invention, the above-mentioned event detection module 503 is specifically configured to perform feature fusion on the suction features of each type of target data; calculate the matching degree between the fused suction features and the preset suction features; and determine whether the ventricular catheter pump has a suction event according to the calculated matching degree.

[0159] Since the fused suction features are obtained by performing feature fusion on the suction features of each target data, and the fused suction features comprehensively reflect the suction features of each target data, therefore, by using the fused suction features, it is possible to more accurately detect whether the ventricular catheter pump has a suction event.

[0160] Corresponding to the above-mentioned method for detecting the suction event of the ventricular catheter pump, an embodiment of the present invention further provides an electronic device.

[0161] See Figure 6 , Figure 6 which is a schematic structural diagram of an electronic device provided by an embodiment of the present invention, including a processor 601, a communication interface 602, a memory 603, and a communication bus 604. Among them, the processor 601, the communication interface 602, and the memory 603 communicate with each other through the communication bus 604.

[0162] The memory 603 is used to store computer programs.

[0163] When the processor 601 is used to execute the program stored on the memory 603, it implements the method for detecting the suction event of the ventricular catheter pump provided by the embodiment of the present invention.

[0164] The communication bus mentioned in the above-mentioned electronic device can be a Peripheral Component Interconnect (PCI) bus, an Extended Industry Standard Architecture (EISA) bus, etc. This communication bus can be divided into an address bus, a data bus, a control bus, etc. For the sake of convenience in representation, only a thick line is used in the figure, but it does not mean that there is only one bus or one type of bus.

[0165] The communication interface is used for communication between the above-mentioned electronic device and other devices.

[0166] The memory can include a Random Access Memory (RAM), and can also include a Non-Volatile Memory (NVM), such as at least one disk memory. Optionally, the memory can also be at least one storage device located far from the aforementioned processor.

[0167] The above-mentioned processor can be a general-purpose processor, including a Central Processing Unit (CPU), a Network Processor (NP), etc.; it can also be a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), a Field-Programmable Gate Array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components.

[0168] In another embodiment provided by the present invention, there is also provided a computer-readable storage medium, in which a computer program is stored, and when the computer program is executed by a processor, the suction event detection method of the ventricular catheter pump provided by the embodiment of the present invention is implemented.

[0169] In another embodiment provided by the present invention, there is also provided a computer program product containing instructions, and when it runs on a computer, it causes the computer to implement the suction event detection method of the ventricular catheter pump provided by the embodiment of the present invention when executed.

[0170] As can be seen from the above, when applying the solution provided in this embodiment, since the detection of whether a suction event occurs in the ventricular catheter pump is based on the rotational speed data, current data of the ventricular catheter pump, and the flow rate data of the blood passing through the ventricular catheter pump, and since the rotational speed, current, and the flow rate of the blood passing through the ventricular catheter pump all affect whether a suction event will occur in the ventricular catheter pump, therefore, by integrating three different types of detection indicators, it is possible to perform comprehensive and accurate detection, improving the accuracy of the detection result.

[0171] In addition, since the extraction of the suction characteristics of the target data is based on the data sequence order of the target data, and the above data sequence order can reflect the relationship information between the data included in each type of target data, therefore, when extracting the suction characteristics, in addition to considering the target data itself, the relationship information between the data included in the target data is also considered. By integrating the above two types of information, it is possible to more accurately extract the suction characteristics of the target data, thereby improving the accuracy of the detection result.

[0172] In the above embodiment, it can be implemented in whole or in part by software, hardware, firmware, or any combination thereof. When implemented using software, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the processes or functions described in the embodiments of the present invention are generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (such as coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (such as infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or data center that includes one or more integrated available media. The available medium can be a magnetic medium (for example, a floppy disk, a hard disk, a magnetic tape), an optical medium (for example, a DVD), or a semiconductor medium (for example, a solid state disk (SSD)).

[0173] It should be noted that in this article, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprising", "including" or any other variants thereof are intended to cover non-exclusive inclusion, such that a process, method, article or device comprising a series of elements not only includes those elements but also includes other elements not expressly listed, or elements inherent to such process, method, article or device. Without further limitation, an element defined by the statement "comprising an..." does not exclude the presence of additional identical elements in the process, method, article or device comprising the said element.

[0174] Each embodiment in this specification is described in a related manner. For the same and similar parts among the embodiments, reference can be made to each other. Each embodiment focuses on the differences from other embodiments. In particular, for the embodiments of the device, electronic device, and computer-readable storage medium, since they are basically similar to the method embodiments, the description is relatively simple. For the relevant parts, reference can be made to the partial description of the method embodiments.

[0175] The above are only the preferred embodiments of the present invention and are not intended to limit the protection scope of the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention are all included in the protection scope of the present invention.

Claims

1. A method for detecting suction events of a ventricular catheter pump, characterized in that, The method includes: During the operation of the ventricular catheter pump, obtaining the rotational speed data and current data of the ventricular catheter pump, and obtaining the blood flow data of the blood passing through the ventricular catheter pump; For each type of target data, based on the data sequence order of the target data of this type, extracting the suction characteristics of the target data of this type, where the types of the target data include: rotational speed data type, current data type, and blood flow data type; Based on the suction characteristics of the extracted target data, predicting whether a suction event occurs in the ventricular catheter pump; Extract the suction characteristics of each type of target data in the following manner: Extract the suction characteristics of the target data in the forward order of the data sequence of the target data, and determine the extracted characteristics as the first sub-characteristics; Extract the suction characteristics of the target data in the reverse order of the data sequence of the target data, and determine the extracted characteristics as the second sub-characteristics; Based on the first sub-characteristics and the second sub-characteristics, determine the suction characteristics of the target data; The step of extracting the suction characteristics of the target data in the forward order of the data sequence of the target data and determining the extracted characteristics as the first sub-characteristics includes: Extract the suction characteristics of each data included in the target data; In the forward order of the data sequence, for each data in the target data except the first data, based on the suction characteristics of the extracted data, update the first type of suction characteristics updated by the previous data of this data, where the first data is: the data included in the target data that is in the first position in the forward order of the data sequence, and the first type of suction characteristics is: the characteristics obtained by extracting the suction characteristics of the first data; Determine the first sub-characteristics as the first type of suction sub-characteristics updated by the second data included in the target data, where the second data is: the data included in the target data that is in the last position in the forward order of the data sequence.

2. The method according to claim 1, characterized in that, The step of extracting the suction characteristics of the target data in the reverse order of the data sequence of the target data and determining the extracted characteristics as the second sub-characteristics includes: Extract the suction characteristics of each data included in the target data; In the reverse order of the data sequence, for each data in the target data except the third data, based on the suction characteristics of the extracted data, update the second type of suction characteristics updated by the previous data of this data, where the third data is: the data included in the target data that is in the first position in the reverse order of the data sequence, and the second type of suction characteristics is: the characteristics obtained by extracting the suction characteristics of the third data; Determine the second sub-characteristics as the second type of suction sub-characteristics updated by the fourth data included in the target data, where the fourth data is: the data included in the target data that is in the last position in the reverse order of the data sequence.

3. The method according to claim 1 or 2, characterized in that, The step of predicting whether a suction event occurs in the ventricular catheter pump based on the suction characteristics of the extracted target data includes: Performing feature fusion on the suction characteristics of each type of target data; Calculating the matching degree between the fused suction characteristics and the preset suction characteristics; Determine whether a suction event occurs in the ventricular catheter pump according to the calculated matching degree.

4. An aspiration event detection device for a ventricular catheter pump, characterized in that, The device includes: A data acquisition module, configured to acquire the rotational speed data and current data of the ventricular catheter pump during the operation of the ventricular catheter pump, and acquire the flow rate data of the blood passing through the ventricular catheter pump; A feature extraction module, configured to perform suction feature extraction on each type of target data based on the data sequence order of the target data of this type, where the types of the target data include: rotational speed data type, current data type, and blood flow rate data type; An event detection module, configured to predict whether a suction event occurs in the ventricular catheter pump based on the suction features of the extracted target data; The feature extraction module includes: A first feature extraction unit, configured to perform suction feature extraction on the target data in the forward order of the data sequence of the target data, and determine the extracted features as first sub-features; A second feature extraction unit, configured to perform suction feature extraction on the target data in the reverse order of the data sequence of the target data, and determine the extracted features as second sub-features; A feature determination unit, configured to determine the suction features of the target data based on the first sub-features and the second sub-features; The first feature extraction unit is specifically configured to perform suction feature extraction on each data included in the target data; in the forward order of the data sequence, for each data in the target data except the first data, based on the suction features of the extracted data, update the first type of suction features updated by the previous data of this data, where the first data is: the data included in the target data that is in the first position in the forward order of the data sequence, and the first type of suction features is: the features obtained by performing suction feature extraction on the first data; determine the first type of suction sub-features updated by the second data included in the target data as the first sub-features, where the second data is: the data included in the target data that is in the last position in the forward order of the data sequence.

5. The device according to claim 4, characterized in that The second feature extraction unit is specifically configured to perform suction feature extraction on each data included in the target data; in the reverse order of the data sequence, for each data in the target data except the third data, based on the suction features of the extracted data, update the second type of suction features updated by the previous data of this data, where the third data is: the data included in the target data that is in the first position in the reverse order of the data sequence, and the second type of suction features is: the features obtained by performing suction feature extraction on the third data; determine the second type of suction sub-features updated by the fourth data included in the target data as the second sub-features, where the fourth data is: the data included in the target data that is in the last position in the reverse order of the data sequence.

6. An electronic device, characterized in that, It includes a processor, a communication interface, a memory, and a communication bus. Among them, the processor, the communication interface, and the memory complete mutual communication through the communication bus; The memory is used to store a computer program; A processor, when executing a program stored in a memory, implements the method steps described in any one of claims 1-3.

Images