WO2022242123A1

WO2022242123A1 - Mechanical ventilation man-machine asynchronous detection method and apparatus, and computer-readable storage medium

Info

Publication number: WO2022242123A1
Application number: PCT/CN2021/137606
Authority: WO
Inventors: 熊富海; 颜延; 谯小豪; 李慧慧; 王磊; 刘语诗; 陈达理; 吴选昆; 梁端; 王博; 曹修齐
Original assignee: 深圳先进技术研究院
Priority date: 2021-05-20
Filing date: 2021-12-13
Publication date: 2022-11-24
Also published as: CN113521460B; CN113521460A

Abstract

A mechanical ventilation man-machine asynchronous detection method and apparatus, and a computer-readable storage medium, relating to the technical field of mechanical ventilation of ventilators. The method comprises: acquiring ventilation data when a ventilator performs mechanical ventilation (S11); inputting the ventilation data into a preset auto-encoder to extract feature data of the ventilation data (S12); and inputting the feature data into a preset convolutional neural network to output a man-machine asynchronous state of the ventilator (S13). It is beneficial to improve the efficiency and accuracy of the mechanical ventilation man-machine asynchronous detection method.

Description

Mechanical ventilation human-computer asynchronous detection method, device and computer-readable storage medium

technical field

The present application relates to the technical field of ventilator-mechanical ventilation, in particular to a method, a device and a computer-readable storage medium for detecting asynchronous detection of mechanical ventilation.

Background technique

In the prior art, when detecting whether the air supply of the ventilator to the user is asynchronous (that is, not synchronous with the user's exhalation or inhalation), it is often necessary for experts in the field to analyze the characteristics of various asynchronous and normal conditions. The information is manually extracted and judged to determine whether the current gas supply and stop gas supply actions of the ventilator are synchronized with the patient's inhalation and exhalation actions respectively.

The defect of the prior art is that manual extraction of features and judgment of feature information to determine whether the air supply to the user by the ventilator is asynchronous, the efficiency is low, and the accuracy of manual processing of feature information to obtain detection results is low.

Contents of the invention

The application provides a mechanical ventilation human-machine asynchronous detection method, a device and a computer-readable storage medium to solve the technical problem of low efficiency and accuracy of the mechanical ventilation human-computer asynchronous detection method in the prior art.

In order to solve the above technical problems, the first technical solution provided by this application is: a mechanical ventilation man-machine asynchronous detection method, including: obtaining the ventilation data when the ventilator is performing mechanical ventilation; inputting the ventilation data into a preset autoencoder , to extract the characteristic data of the ventilation data; input the characteristic data into the preset convolutional neural network to output the human-machine asynchronous state of the ventilator.

The second technical solution provided by this application is: a mechanical ventilation human-machine asynchronous detection device, including: a memory and a processor; the memory is used to store program instructions, and the processor is used to execute the program instructions to realize the above-mentioned mechanical ventilation human-machine asynchronous detection method.

The third technical solution provided by the present application is: a computer-readable storage medium, the computer-readable storage medium stores program instructions, and when the program instructions are executed by a processor, the above-mentioned method for asynchronous detection of mechanical ventilation is implemented.

The human-machine asynchronous detection method of mechanical ventilation provided by this application obtains the ventilation data when the ventilator performs mechanical ventilation; inputs the ventilation data into the preset autoencoder to extract the characteristic data of the ventilation data; and inputs the characteristic data into the preset Convolutional neural network to output the human-machine asynchronous state of the ventilator. This application first extracts the characteristic data of the ventilation data based on the preset autoencoder, and then inputs the characteristic data into the preset convolutional neural network to identify the man-machine asynchronous state corresponding to the ventilation data, avoiding artificial The steps of extracting features or manually identifying the man-machine asynchronous state reduce the consumption of human resources and improve the efficiency and accuracy of the man-machine asynchronous detection method for mechanical ventilation.

Description of drawings

In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings that need to be used in the description of the embodiments will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present invention. For those skilled in the art, other drawings can also be obtained based on these drawings without creative effort.

Fig. 1 is a schematic flow chart of an embodiment of a mechanical ventilation man-machine asynchronous detection method of the present application;

Fig. 2 is a schematic structural diagram of an embodiment of the self-encoder of the present application;

Fig. 3 is a schematic structural diagram of an embodiment of the one-dimensional convolutional neural network of the present application;

Fig. 4 is a schematic flowchart of another embodiment of the mechanical ventilation man-machine asynchronous detection method of the present application;

Fig. 5 is a schematic flow chart of an embodiment of step S22 in the method for man-machine asynchronous detection of mechanical ventilation shown in Fig. 4;

Fig. 6 is a schematic diagram of the relationship between the training loss and the number of training iterations of an embodiment of the self-encoder of the present application;

Fig. 7 is a schematic diagram of the result of the confusion matrix of an embodiment of the mechanical ventilation man-machine asynchronous detection method of the present application;

Fig. 8 is a schematic structural diagram of an embodiment of the human-machine asynchronous detection device for mechanical ventilation of the present application;

FIG. 9 is a schematic structural diagram of an embodiment of a computer-readable storage medium of the present application.

Detailed ways

The following will clearly and completely describe the technical solutions in the embodiments of the application with reference to the drawings in the embodiments of the application. Apparently, the described embodiments are only some of the embodiments of the application, not all of them. Based on the embodiments in this application, all other embodiments obtained by persons of ordinary skill in the art without making creative efforts belong to the scope of protection of this application.

The terms "first" and "second" in this application are only used for descriptive purposes, and cannot be understood as indicating or implying relative importance or implicitly specifying the quantity of indicated technical features. In the description of the present application, "plurality" means at least two, such as two, three, etc., unless otherwise specifically defined. Furthermore, the terms "include" and "have", as well as any variations thereof, are intended to cover a non-exclusive inclusion. For example, a process, method, system, product or device comprising a series of steps or units is not limited to the listed steps or units, but optionally also includes unlisted steps or units, or optionally further includes For other steps or units inherent in these processes, methods, products or apparatuses.

The present application first proposes a method for detecting asynchronous mechanical ventilation, as shown in FIG. 1 , which is a schematic flowchart of an embodiment of the method for detecting asynchronous mechanical ventilation according to the present application.

It should be noted that the ventilator is an important life device for respiratory function support, and is widely used in the intensive care department of the hospital, general departments or families, and is an important auxiliary support device for people with respiratory dysfunction . The main function of the ventilator is to supply air to the patient when it detects that the patient needs to inhale, and to stop the supply of air to the patient when the patient needs to exhale. If the ventilator stops supplying air when the patient needs to inhale and supplies air to the patient when the patient needs to exhale, the ventilator can be considered to be asynchronous in the process of supplying air to the patient.

The human-machine asynchronous detection method for mechanical ventilation in this embodiment specifically includes the following steps:

Step S11: Obtain ventilation data when the ventilator performs mechanical ventilation.

In this embodiment, a preset sampling frequency (for example: 50 Hz) may be used to collect ventilation data of a ventilator undergoing mechanical ventilation. The ventilation data may at least include one of airflow velocity data, airflow channel pressure data and air flow data, wherein the airflow velocity data is used to describe the velocity of the airflow of the ventilator during mechanical ventilation, and the airflow channel pressure is used to describe the airflow velocity of the ventilator. The air pressure in the delivered airflow channel during mechanical ventilation, and the airflow data are used to describe the amount of airflow delivered by the ventilator during mechanical ventilation. For example, related sensors can be set in the air supply pipeline used by the ventilator to supply air to the patient, so as to obtain the airflow velocity data, airflow channel pressure data and air flow data of the ventilator during mechanical ventilation.

Step S12: Input the ventilation data into a preset autoencoder to extract characteristic data of the ventilation data.

In this embodiment, the ventilation data acquired in step S11 may be input into an autoencoder that has been trained in advance, so that the autoencoder can extract feature information from the ventilation data to generate feature data. The extraction of feature information based on autoencoders can not only save a lot of valuable human resources, but also avoid errors caused by human errors, which can greatly improve the efficiency and accuracy of feature information extraction.

Optionally, as shown in FIG. 2 , FIG. 2 is a schematic structural diagram of an embodiment of an autoencoder of the present application. Self-encoder 20 comprises: self-encoder input layer 201, first hidden layer 202, second hidden layer 203, third hidden layer 204, fourth hidden layer 205, fifth hidden layer 206 and self-encoder output layer 207;

The self-encoder input layer 201, the first hidden layer 202, the second hidden layer 203, the third hidden layer 204, the fourth hidden layer 205, the fifth hidden layer 206 and the self-encoder output layer 207 are sequentially connected, and the self-encoder input The layer 201 is used to receive ventilation data, the autoencoder output layer 207 is the output layer of the autoencoder during training, and the third hidden layer 204 is used to output feature data.

Step S13: Input the feature data into the preset convolutional neural network to output the human-machine asynchronous state of the ventilator.

In this embodiment, the feature data extracted in step S12 can be input into the convolutional neural network that has completed relevant training in advance, so that the convolutional neural network can identify the asynchronous state of man and machine through the processing of feature data, and then can pass human The Ventilator Asynchronous Status determines whether there is asynchronous air supply to the ventilator. The man-machine asynchronous state includes at least double-trigger state, invalid inspiratory effort state and normal state, and may also include other types of man-machine asynchronous states, which are not limited here. For example: In the first case, the patient’s breathing is relatively rapid, and the ventilator may capture the patient’s inspiratory movements twice in a row, which will then trigger the patient’s mechanical ventilation twice in a row, resulting in an oversupply situation. , the ventilator is in a dual-trigger state at this time; in the second case, the patient’s breathing force is weak, and the ventilator may always or often fail to capture the patient’s inspiratory action, and thus cannot trigger the patient’s mechanical ventilation action, thus In the third case, the ventilator normally supplies air when the patient needs to inhale, and stops supplying air when the patient needs to exhale. The machine is in normal condition. The man-machine asynchronous detection method for mechanical ventilation of the present application can quickly and accurately identify the abnormal state (double trigger state and invalid inspiratory effort state) or normal state in the above three situations.

Optionally, as shown in FIG. 3 , FIG. 3 is a schematic structural diagram of an embodiment of a one-dimensional convolutional neural network of the present application.

The convolutional neural network is a one-dimensional convolutional neural network 30;

One-dimensional convolutional neural network 30 includes: one-dimensional convolutional neural network input layer 301, first one-dimensional convolutional layer 302, first maximum pooling layer 303, second one-dimensional convolutional layer 304, second maximum pooling Layer 305, the third one-dimensional convolutional layer 306, the first fully connected layer 307, the second fully connected layer 308 and the one-dimensional convolutional neural network output layer 309;

One-dimensional convolution neural network input layer 301, first one-dimensional convolution layer 302, first maximum pooling layer 303, second one-dimensional convolution layer 304, second maximum pooling layer 305, third one-dimensional convolution Layer 306, the first fully connected layer 307, the second fully connected layer 308 and the one-dimensional convolutional neural network output layer 309 are sequentially connected, the one-dimensional convolutional neural network input layer 301 is used to receive feature data, and the one-dimensional convolutional neural network The output layer 309 is used to output the human-machine asynchronous state.

Optionally, as shown in FIG. 4 , FIG. 4 is a schematic flow chart of another embodiment of the method for detecting human-machine asynchrony in mechanical ventilation of the present application. The respiratory monitoring method of this embodiment specifically further includes the following steps:

Step S21: Obtain training data.

In this embodiment, based on the understanding of real respiratory events, a ventilator and a simulated lung can be used to simulate patients with acute respiratory distress syndrome (ARDS) facing different man-machines when using a ventilator. Respiratory events in the asynchronous state (including at least: double-trigger state, invalid inspiratory effort state and normal state), and then obtain simulated ventilation data when the ventilator is in different man-machine asynchronous states based on multiple respiratory events, and use the different Simulated ventilation data under respiratory events were used as training data.

Step S22: Perform preprocessing on the training data; wherein, the preprocessing includes at least marking the training data with man-machine asynchronous state.

Step S23: Input the preprocessed training data into the autoencoder to train the autoencoder.

In this embodiment, the training data can be preprocessed first (including at least marking the training data corresponding to the man-machine asynchronous state of each training data), and then the autoencoder is trained based on the preprocessed training data, and After the training is completed, the output data of the third hidden layer (middle layer) of the trained self-encoder is used as the feature data of the ventilation data.

Optionally, as shown in FIG. 4 , FIG. 4 is a schematic flow chart of another embodiment of the method for detecting human-machine asynchrony in mechanical ventilation of the present application. Step S22 may specifically include: dividing the training data into multiple data segments, and marking each data segment with a man-machine asynchronous state.

Step S23 may specifically include: respectively inputting data corresponding to each man-machine asynchronous state into the autoencoder, so as to train the autoencoder.

Further, as shown in FIG. 5 , FIG. 5 is a schematic flow chart of step S22 in the method for detecting the asynchronous detection of man-machine in mechanical ventilation shown in FIG. 4 . Step S22 may specifically include the following steps:

Step S221: Divide the training data into multiple data segments.

Step S222: Completing the sampling points for each data segment, so that the number of sampling points in each data segment is the same.

Step S223: Standardize the data of each data segment by using the mean value and standard deviation of the data of each sampling point in each data segment.

Step S224: Carry out man-machine asynchronous state labeling for each data segment after normalization processing.

In this embodiment, the filling of the above-mentioned sampling points may be to pad or truncate each piece of data so that the number of each data section is the same (same length). For example, each data section can be filled with zeros or deleted. The method of too many sampling points makes it have 100 sampling point data.

For example, in practice, the calculation formula of the longest respiratory cycle signal length is as follows:

maxLen=max(len(P ₁ ,P ₂ ,...,P _L ));

L=len _trainset +len _testset ;

In the formula, maxLen is the longest respiratory cycle signal length, len trainset is the data segment quantity of training data, len _testset is the data segment _quantity of test data, P _x is the data segment of a training data or test data, and test data is used for Autoencoders and Convolutional Neural Networks for testing purposes.

The longest respiratory cycle signal length can be calculated based on the above-mentioned longest respiratory cycle signal length calculation formula. If the longest respiratory cycle signal length is 100, each data segment can be filled with zeros or deleted. The method makes it have 100 sampling point data.

After completing the sampling points for each data segment, each data segment is standardized based on the mean value and standard deviation (or variance) of each sampling point data in each data segment. The training data trains the autoencoder and the convolutional neural network, which is conducive to further improving the efficiency and accuracy of the autoencoder and convolutional neural network in feature extraction and/or recognition of human-machine asynchronous states.

Specifically, the formula of the normalization process is as follows:

In the formula,

is the sampling point data after normalization processing, N _x is the sampling point data before normalization processing, μ is the mean value of the sampling point data in the data segment where the sampling point data is located, and σ is the standard deviation of the sampling point data in the data segment where the sampling point data is located .

In an actual application scenario, the training data and the ventilation data may include three types of data, namely air flow velocity data, air flow channel pressure data and air flow data, and may form three-dimensional data. For example, a sequence of data segments of the airflow velocity data of a breathing cycle is F _x =(f ₁ , f ₂ ,...,f _x ), and a sequence of data segments of airflow channel pressure data is P _x =(p ₁ ,p ₂ ,...,p _x ), a sequence of data segments of air flow data is V _x =(v ₁ ,v ₂ ,...,v _x ). Then, after normalization is performed on F _x , P _x , and V _x , N in the formula of the above normalization process may refer to any one of F, P, and V.

Self-encoder may comprise: self-encoder input layer, first hidden layer, second hidden layer, third hidden layer, fourth hidden layer, fifth hidden layer and self-encoder output layer; self-encoder input layer, second hidden layer The first hidden layer, the second hidden layer, the third hidden layer, the fourth hidden layer, the fifth hidden layer and the output layer of the autoencoder are sequentially connected. Among them, the number of neurons in the first hidden layer, the second hidden layer, the third hidden layer, the fourth hidden layer, and the fifth hidden layer are 128, 64, 32, 64, and 128 respectively. When training the autoencoder, the The training data is used as the input data of the input layer of the autoencoder and the output data of the output layer of the autoencoder at the same time for training. When the autoencoder is actually used, the data output by the third hidden layer can be used as the feature data extracted by the autoencoder based on the ventilation data, so as to extract the characteristics of the ventilation data as three-dimensional data, form one-dimensional feature data, and realize the feature The extraction and dimensionality reduction of data is conducive to the rapid processing of the feature data by the subsequent convolutional neural network.

The convolutional neural network is a one-dimensional convolutional neural network; the one-dimensional convolutional neural network includes: a one-dimensional convolutional neural network input layer, a first one-dimensional convolution layer, a first maximum pooling layer, a second one-dimensional convolution layer, the second maximum pooling layer, the third one-dimensional convolutional layer, the first fully connected layer, the second fully connected layer and the output layer of the one-dimensional convolutional neural network; the input layer of the one-dimensional convolutional neural network, the first one 1D convolutional layer, 1st max pooling layer, 2nd 1D convolutional layer, 2nd max pooling layer, 3rd 1D convolutional layer, 1st fully connected layer, 2nd fully connected layer and 1D convolution The output layers of the convolutional neural network are sequentially connected, wherein a dropout layer may also be included between the first fully connected layer and the second fully connected layer to reduce the risk of overfitting in the convolutional neural network. When training a one-dimensional convolutional neural network, the one-dimensional feature data extracted by the autoencoder based on the training data can be input into the input layer of the one-dimensional convolutional neural network, and the label information corresponding to the training data (used to label the training The information of the human-machine asynchronous state corresponding to the data) is used as the output of the output layer of the one-dimensional convolutional neural network for training. When using the one-dimensional convolutional neural network in practice, similar steps can be taken during training, and the one-dimensional feature data extracted by the autoencoder based on the ventilation data can be input into the input layer of the one-dimensional convolutional neural network to make the one-dimensional convolutional neural network The network outputs the corresponding man-machine asynchronous state (that is, the man-machine asynchronous state corresponding to the acquired ventilation data). Based on the above method, using a one-dimensional convolutional neural network to process one-dimensional data in the process of respiratory monitoring can greatly improve the efficiency of respiratory monitoring.

As shown in FIG. 6 , FIG. 6 is a schematic diagram of the relationship between the autoencoder training loss and the number of training iterations in the present application. A is the training loss curve of the airflow velocity data, B is the training loss curve of the airflow channel pressure data, and C is the training loss curve of the air flow data. As the number of training iterations in the training process increases, the training loss of the autoencoder becomes smaller and smaller. When the number of training iterations reaches more than one hundred times, the training loss has dropped to close to the convergence value, and the convergence Faster.

As shown in FIG. 7 , FIG. 7 is a schematic diagram of the confusion matrix result of the human-machine asynchronous detection method for mechanical ventilation of the present application. It can be seen that the human-machine asynchronous detection method for mechanical ventilation of the present application has an accurate detection probability of 100% for the double-trigger state or the invalid inspiratory effort state after training with a certain amount of training data, and the accurate detection probability for the normal state is 97%, there is a 3% probability that the normal state will be identified as a double-trigger state. It can be seen that the accuracy rate of the human-machine asynchronous detection method for mechanical ventilation in this application is extremely high, and the accuracy rate can continue to improve with the increase of training data.

The man-machine asynchronous detection method for mechanical ventilation provided by this application obtains the ventilation data when the ventilator performs mechanical ventilation; inputs the ventilation data into the preset autoencoder to extract the characteristic data of the ventilation data; and inputs the characteristic data into the preset Convolutional neural network to output the human-machine asynchronous state of the ventilator. This application first extracts the characteristic data of the ventilation data based on the preset autoencoder, and then inputs the characteristic data into the preset convolutional neural network to identify the man-machine asynchronous state corresponding to the ventilation data, avoiding artificial The steps of extracting features or manually identifying the man-machine asynchronous state reduce the consumption of human resources and improve the efficiency and accuracy of the man-machine asynchronous detection method for mechanical ventilation.

The present application further proposes a human-machine asynchronous detection device for mechanical ventilation, as shown in FIG. 8 , which is a schematic structural diagram of an embodiment of the human-computer asynchronous detection device for mechanical ventilation of the present application. The apparatus 80 for detecting human-machine asynchrony of mechanical ventilation in this embodiment includes: a processor 81 , a memory 82 and a bus 83 .

The processor 81 and the memory 82 are connected to the bus 83 respectively. The memory 82 stores program instructions, and the processor 81 is used to execute the program instructions to implement the method for detecting asynchronous mechanical ventilation in the above embodiments.

In this embodiment, the processor 81 may also be called a CPU (Central Processing Unit, central processing unit). The processor 81 may be an integrated circuit chip with signal processing capabilities. The processor 81 can also be a general-purpose processor, 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 . The general processor may be a microprocessor or the processor 81 may be any conventional processor or the like.

The asynchronous detection device for mechanical ventilation mentioned above may be a detection device independent of the ventilator, or a part of the ventilator, which is not limited here.

The present application further proposes a computer-readable storage medium, as shown in FIG. 9 , which is a structural schematic diagram of an embodiment of the computer-readable storage medium of the present application. The computer-readable storage medium 90 stores program instructions 91 thereon, When the program instruction 91 is executed by the processor (not shown in the figure), the method for detecting the asynchronous detection of mechanical ventilation in the above-mentioned embodiments is realized.

The computer-readable storage medium 90 in this embodiment may be, but not limited to, a USB flash drive, an SD card, a PD optical drive, a mobile hard disk, a large-capacity floppy drive, a flash memory, a multimedia memory card, a server, and the like.

In the description of this application, reference to the terms "one embodiment," "some embodiments," "example," "specific examples," or "some examples" means that specific features described in connection with that embodiment or example , structure, material or characteristic is included in at least one embodiment or example of the present application. In this specification, the schematic representations of the above terms are not necessarily directed to the same embodiment or example. Furthermore, the described specific features, structures, materials or characteristics may be combined in any suitable manner in any one or more embodiments or examples. In addition, those skilled in the art can combine and combine different embodiments or examples and features of different embodiments or examples described in this specification without conflicting with each other.

In addition, the terms "first" and "second" are used for descriptive purposes only, and cannot be interpreted as indicating or implying relative importance or implicitly specifying the quantity of indicated technical features. Thus, the features defined as "first" and "second" may explicitly or implicitly include at least one of these features. In the description of the present application, "plurality" means at least two, such as two, three, etc., unless otherwise specifically defined.

Any process or method descriptions in flowcharts or otherwise described herein may be understood to represent modules, segments or portions of code comprising one or more executable instructions for implementing specific logical functions or steps of the process , and the scope of preferred embodiments of the present application includes additional implementations in which functions may be performed out of the order shown or discussed, including in substantially simultaneous fashion or in reverse order depending on the functions involved, which shall It should be understood by those skilled in the art to which the embodiments of the present application belong.

The logic and/or steps represented in the flowcharts or otherwise described herein, for example, can be considered as a sequenced listing of executable instructions for implementing logical functions, can be embodied in any computer-readable medium, For the use of instruction execution systems, devices or equipment (which may be personal computers, servers, network equipment or other systems that can fetch instructions from instruction execution systems, devices or devices and execute instructions), or in combination with these instruction execution systems, devices or devices And use. For the purposes of this specification, a "computer-readable medium" may be any device that can contain, store, communicate, propagate or transmit a program for use in or in conjunction with an instruction execution system, device or device. More specific examples (non-exhaustive list) of computer-readable media include the following: electrical connection with one or more wires (electronic device), portable computer disk case (magnetic device), random access memory (RAM), Read Only Memory (ROM), Erasable and Editable Read Only Memory (EPROM or Flash Memory), Fiber Optic Devices, and Portable Compact Disc Read Only Memory (CDROM). In addition, the computer-readable medium may even be paper or other suitable medium on which the program can be printed, since the program can be read, for example, by optically scanning the paper or other medium, followed by editing, interpretation or other suitable processing if necessary. The program is processed electronically and stored in computer memory.

The above is only the implementation of the application, and does not limit the patent scope of the application. Any equivalent structure or equivalent process conversion made by using the specification and drawings of the application, or directly or indirectly used in other related technologies fields, are all included in the scope of patent protection of this application in the same way.

Claims

A mechanical ventilation man-machine asynchronous detection method, characterized in that it includes:

Obtain the ventilation data when the ventilator is performing mechanical ventilation;

inputting the ventilation data into a preset autoencoder to extract characteristic data of the ventilation data;

The feature data is input into a preset convolutional neural network to output the man-machine asynchronous state of the ventilator.
The mechanical ventilation human-machine asynchronous detection method according to claim 1, characterized in that,

The ventilation data includes at least one of air flow velocity data, air flow channel pressure data and air flow data.
The mechanical ventilation human-machine asynchronous detection method according to claim 1, characterized in that,

The man-machine asynchronous state includes at least one of a double trigger state, an invalid inspiratory effort state and a normal state.
The mechanical ventilation human-machine asynchronous detection method according to claim 1, characterized in that,

The method also includes:

get training data;

Preprocessing the training data; wherein, the preprocessing includes at least marking the training data with man-machine asynchronous state;

Inputting the preprocessed training data into the autoencoder to train the autoencoder.
The mechanical ventilation human-machine asynchronous detection method according to claim 4, characterized in that,

The preprocessing of the training data includes:

dividing the training data into a plurality of data segments;

Carrying out man-machine asynchronous status labeling for each of the data segments; and

The step of inputting the preprocessed training data into the autoencoder to train the autoencoder includes:

The data corresponding to each man-machine asynchronous state is input to the autoencoder to train the autoencoder.
The human-machine asynchronous detection method for mechanical ventilation according to claim 5, characterized in that,

After said dividing the training data into multiple data segments, it also includes:

Completing the sampling points for each of the data segments, so that the number of sampling points for each of the data segments is the same;

Using the mean value and standard deviation of the data of each sampling point in each of the data segments, the data of each of the data segments is standardized.
The human-machine asynchronous detection method according to any one of claims 1 to 6, wherein the autoencoder includes: an autoencoder input layer, a first hidden layer, a second hidden layer, and a third hidden layer. layer, the fourth hidden layer, the fifth hidden layer and the autoencoder output layer;

The autoencoder input layer, the first hidden layer, the second hidden layer, the third hidden layer, the fourth hidden layer, the fifth hidden layer and the autoencoder output layer connected in sequence, the autoencoder input layer is used to receive the ventilation data, the autoencoder output layer is the output layer of the autoencoder during training, and the third hidden layer is used to output the feature data.
The human-machine asynchronous detection method for mechanical ventilation according to any one of claims 1 to 6, wherein the convolutional neural network is a one-dimensional convolutional neural network;

The one-dimensional convolutional neural network comprises: one-dimensional convolutional neural network input layer, the first one-dimensional convolutional layer, the first maximum pooling layer, the second one-dimensional convolutional layer, the second maximum pooling layer, the first Three one-dimensional convolutional layer, first fully connected layer, second fully connected layer and one-dimensional convolutional neural network output layer;

The one-dimensional convolutional neural network input layer, the first one-dimensional convolutional layer, the first maximum pooling layer, the second one-dimensional convolutional layer, the second maximum pooling layer, the The third one-dimensional convolutional layer, the first fully connected layer, the second fully connected layer and the one-dimensional convolutional neural network output layer are sequentially connected, and the one-dimensional convolutional neural network input layer is used for The feature data is received, and the output layer of the one-dimensional convolutional neural network is used to output the asynchronous state of man-machine.
A mechanical ventilation human-machine asynchronous detection device, characterized in that it includes: a memory and a processor;

The memory is used to store program instructions, and the processor is used to execute the program instructions to implement the method according to any one of claims 1 to 8.
A computer-readable storage medium, wherein the computer-readable storage medium stores program instructions, and when the program instructions are executed by a processor, the method according to any one of claims 1 to 8 is implemented.