Method, apparatus, and electronic device for assisting in predicting latent atrial fibrillation

Abstract

The present invention provides a method, apparatus and electronic device for assisting in predicting latent atrial fibrillation, which are applied to the technical field of electromagnetic induction. The method includes: acquiring an initial electromagnetic echo signal; performing signal separation processing on the initial electromagnetic echo signal to obtain a target electromagnetic echo signal corresponding to the motion state of the heart region of the object to be measured within a predetermined time period; using the spatio-temporal encoder of the information detection model to perform first feature extraction processing on the target electromagnetic echo signal to generate a plurality of mechanical motion spatio-temporal features related to normal heartbeats and abnormal heartbeats in the heart region; using the feature extraction encoder of the information detection model to perform second feature extraction processing on the plurality of mechanical motion spatio-temporal features to generate a target mechanical motion time feature related to abnormal heartbeats in the heart region; and using the detection and evaluation sub-model of the information detection model to perform feature conversion mapping processing on the target mechanical motion time feature to obtain the target predicted heartbeat information of the object to be measured.

Description

Technical Field

[0001] The present invention relates to the technical field of electromagnetic induction, and more particularly to a method, device and electronic device for assisting in predicting latent atrial fibrillation. Background Art

[0002] Atrial fibrillation is a common type of arrhythmia and also a relatively common type of cardiovascular disease, and atrial fibrillation increases the risk of stroke or other serious problems. Electrocardiogram, as a common auxiliary detection image at present, is used to help diagnose whether a target has atrial fibrillation. However, due to the paroxysmal and concealed nature of latent atrial fibrillation, even when performing multiple electrocardiogram examinations on a target, there is still a high probability that the latent atrial fibrillation suffered by the target cannot be detected, resulting in an increased risk of serious problems.

[0003] In the process of implementing the above inventive concept, the inventors found that: in the related art, due to the paroxysmal and concealed nature of latent atrial fibrillation, it is impossible to directly obtain an accurate auxiliary image containing latent atrial fibrillation information, thus unable to timely assist relevant personnel in making predictive judgments. Moreover, when obtaining latent atrial fibrillation information, it is necessary to use a contact device to detect the target multiple times, resulting in technical problems of low processing efficiency and poor flexibility. Summary of the Invention

[0004] In view of the above problems, the present invention provides a method, device and electronic device for assisting in predicting latent atrial fibrillation.

[0005] According to a first aspect of the present invention, there is provided a method for assisting in predicting latent atrial fibrillation, including: obtaining an initial electromagnetic echo signal, where the initial electromagnetic echo signal represents a plurality of signals that are spatially aliased with each other and reflected by multiple body regions of a measured object within a predetermined time period; performing signal separation processing on the initial electromagnetic echo signal to obtain a target electromagnetic echo signal corresponding to the motion state of the heart region of the measured object within the predetermined time period; using a spatio-temporal encoder of an information detection model to perform a first feature extraction process on the target electromagnetic echo signal to generate a plurality of mechanical motion spatio-temporal features related to normal and abnormal heartbeats of the heart region; using a feature extraction encoder of the information detection model to perform a second feature extraction process on the plurality of mechanical motion spatio-temporal features to generate a target mechanical motion time feature related to abnormal heartbeats of the heart region; and using a detection and evaluation sub-model of the information detection model to perform a feature conversion mapping process on the target mechanical motion time feature to obtain target auxiliary prediction heartbeat information of the measured object.

[0006] According to an embodiment of the present invention, signal separation processing is performed on an initial electromagnetic echo signal to obtain a target electromagnetic echo signal corresponding to the motion state of the heart region of the object to be measured within a predetermined time period, including: performing beamforming processing on the initial electromagnetic echo signal to obtain a plurality of intermediate electromagnetic echo signals, wherein the intermediate electromagnetic echo signals represent a plurality of four-dimensional signals that are spatially separated from each other and reflected by a plurality of regions within a predetermined time period; extracting a three-dimensional electromagnetic echo signal corresponding to the heart region of the object to be measured from the plurality of intermediate electromagnetic echo signals; and performing filtering processing on the three-dimensional electromagnetic echo signal by using a second-order difference time-domain filtering algorithm to obtain the target electromagnetic echo signal.

[0007] According to an embodiment of the present invention, extracting a three-dimensional electromagnetic echo signal corresponding to the heart region of the object to be measured from the plurality of intermediate electromagnetic echo signals includes: performing superposition processing on the plurality of intermediate electromagnetic echo signals to obtain an intermediate electromagnetic echo superposition signal; performing processing on the intermediate electromagnetic echo superposition signal by using a constant false alarm rate algorithm to obtain a distance electromagnetic echo signal corresponding to the chest region of the object to be measured, wherein the distance electromagnetic echo signal is a three-dimensional signal; and performing clustering processing on the distance electromagnetic echo signal by using a density clustering algorithm to obtain a three-dimensional electromagnetic echo signal corresponding to the heart region of the object to be measured.

[0008] According to an embodiment of the present invention, performing first feature extraction processing on the target electromagnetic echo signal by using a spatio-temporal encoder of an information detection model to generate a plurality of mechanical motion spatio-temporal features related to normal and abnormal heartbeats of the heart region includes: based on a predetermined coding time period, splitting the target electromagnetic echo signal into target electromagnetic echo sub-signals of a plurality of time periods; and respectively inputting the target electromagnetic echo sub-signals of the plurality of time periods into the spatio-temporal encoder of the information detection model to perform first feature extraction processing, generating a plurality of mechanical motion spatio-temporal features related to normal and abnormal heartbeats of the heart region, wherein the plurality of mechanical motion spatio-temporal features correspond to the target electromagnetic echo sub-signals of the plurality of time periods.

[0009] According to an embodiment of the present invention, the detection and evaluation sub-model includes a first double-layer fully connected network and a second double-layer fully connected network, and both the first double-layer fully connected network and the second double-layer fully connected network include two linear layers; using the detection and evaluation sub-model of the information detection model to perform feature transformation mapping processing on the target mechanical motion time feature to obtain the target auxiliary predicted heartbeat information of the object to be measured, including: respectively inputting the target mechanical motion time feature into the first linear layer of the first double-layer fully connected network and the first linear layer of the second double-layer fully connected network for dimensionality reduction processing to obtain a first dimensionality-reduced mechanical motion state feature and a second dimensionality-reduced mechanical motion time feature; inputting the first dimensionality-reduced mechanical motion state feature into the second linear layer of the first double-layer fully connected network for feature transformation mapping processing to obtain the auxiliary atrial beating information of the object to be measured; inputting the second dimensionality-reduced mechanical motion time feature into the second linear layer of the second double-layer fully connected network for feature transformation mapping processing to obtain the auxiliary atrial fibrillation information of the object to be measured.

[0010] According to an embodiment of the present invention, the information detection model is trained by the following method, including: obtaining an initial model to be trained and a sample training data set, where the initial model to be trained includes an initial spatio-temporal encoder, a feature extraction encoder to be trained, and a detection and evaluation sub-model to be trained, and the sample training data set includes electromagnetic echo signal samples, recessive atrial fibrillation disease period information samples, atrial beating information samples, and atrial fibrillation information samples; based on the predetermined sample coding period information, splitting the electromagnetic echo signal samples into multiple electromagnetic echo signal sub-samples; inputting the multiple electromagnetic echo signal sub-samples into the initial spatio-temporal encoder of the initial model to be trained for the first feature extraction process to generate multiple training mechanical motion spatio-temporal features related to normal and abnormal heartbeats in the heart region; inputting the multiple mechanical motion spatio-temporal training features into the feature extraction encoder to be trained for the second feature extraction process to generate training mechanical motion time features related to abnormal heartbeats in the heart region; inputting the training mechanical motion time features into the detection and evaluation sub-model to be trained for feature transformation mapping processing to obtain training detection heartbeat information; based on the detection and evaluation loss function, obtaining a detection and evaluation loss value according to the training detection heartbeat information, recessive atrial fibrillation disease period information samples, atrial beating information samples, and atrial fibrillation information samples; adjusting the model parameters of the initial model according to the detection and evaluation loss value to obtain the trained information detection model.

[0011] According to an embodiment of the present invention, the initial spatio-temporal encoder is pre-trained, and the pre-training method includes the following operations: obtaining a basic spatio-temporal encoder to be trained; inputting a plurality of electromagnetic echo signal sub-samples into the basic spatio-temporal encoder to be trained for first feature extraction processing to generate a plurality of basic training mechanical motion spatio-temporal features related to normal heartbeats and abnormal heartbeats in the heart region; based on a preset contrast weight function for the recessive atrial fibrillation period, obtaining a contrast weight for the recessive atrial fibrillation period according to the recessive atrial fibrillation disease period information sample; based on a preset spatio-temporal loss function, performing regression contrast learning processing on the contrast weight for the recessive atrial fibrillation period and the plurality of basic training mechanical motion spatio-temporal features to obtain a spatio-temporal loss value; adjusting the contrast weight for the recessive atrial fibrillation period of the basic spatio-temporal encoder to be trained according to the spatio-temporal loss value to obtain the trained initial spatio-temporal encoder.

[0012] According to an embodiment of the present invention, the training for detecting heartbeat information includes training atrial beating information and training atrial fibrillation information; based on a detection evaluation loss function, according to the training for detecting heartbeat information, the recessive atrial fibrillation disease period information sample, the atrial beating information sample, and the atrial fibrillation information sample, obtaining the detection evaluation loss value includes a plurality of iteration rounds, and the operations in any t-th iteration round include: in the t-th iteration round, obtaining a cross-entropy loss value corresponding to the t-th iteration round according to the training atrial beating information and the atrial beating information sample in the t-th iteration round, where t≥1 and t is a positive integer; obtaining a mean square error loss value corresponding to the t-th iteration round according to the plurality of training atrial fibrillation information and the atrial fibrillation information sample in the t-th iteration round; determining a change rate of the atrial beating loss corresponding to the t-th iteration round according to the average cross-entropy loss value corresponding to the t-th iteration round and the average cross-entropy loss value corresponding to the (t - 1)-th iteration round; determining a change rate of the atrial fibrillation loss corresponding to the t-th iteration round according to the average mean square error loss value corresponding to the t-th iteration round and the average mean square error loss value corresponding to the (t - 1)-th iteration round; obtaining a change rate of the training loss corresponding to the t-th iteration round according to the change rate of the atrial beating loss corresponding to the t-th iteration round and the change rate of the atrial fibrillation loss corresponding to the t-th iteration round; based on a preset change rate function for the beating loss and a preset change rate function for the fibrillation loss, obtaining a weighted factor for the atrial beating loss and a weighted factor for the atrial fibrillation loss in the t-th iteration round according to the change rate of the training loss corresponding to the t-th iteration round, the change rate of the atrial beating loss corresponding to the t-th iteration round, and the change rate of the atrial fibrillation loss corresponding to the t-th iteration round; based on a detection evaluation function, obtaining the detection evaluation loss value in the t-th iteration round according to the cross-entropy loss value corresponding to the t-th iteration round, the mean square error loss value corresponding to the t-th iteration round, the weighted factor for the atrial beating loss in the t-th iteration round, and the weighted factor for the atrial fibrillation loss in the t-th iteration round.

[0013] The second aspect of the present invention provides a device for assisting in predicting latent atrial fibrillation, characterized by comprising: an acquisition module for acquiring an initial electromagnetic echo signal, wherein the initial electromagnetic echo signal represents a plurality of signals that are spatially aliased with each other and reflected by a plurality of body regions of the object to be measured within a predetermined time period; a separation module for performing signal separation processing on the initial electromagnetic echo signal to obtain a target electromagnetic echo signal corresponding to the motion state of the heart region of the object to be measured within a predetermined time period; a first extraction module for performing first feature extraction processing on the target electromagnetic echo signal by using the spatio-temporal encoder of the information detection model to generate a plurality of mechanical motion spatio-temporal features related to normal and abnormal heartbeats of the heart region; a second extraction module for performing second feature extraction processing on the plurality of mechanical motion spatio-temporal features by using the feature extraction encoder of the information detection model to generate target mechanical motion time features related to abnormal heartbeats of the heart region; and an obtaining module for performing feature transformation mapping processing on the target mechanical motion time features by using the detection and evaluation sub-model of the information detection model to obtain the target auxiliary prediction heartbeat information of the object to be measured.

[0014] The third aspect of the present invention provides an electronic device, comprising: one or more processors; a memory for storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors are caused to execute the above method.

[0015] The fourth aspect of the present invention further provides a computer-readable storage medium, having stored thereon executable instructions, which when executed by a processor cause the processor to execute the above method.

[0016] The fifth aspect of the present invention further provides a computer program product, comprising a computer program, which when executed by a processor implements the above method.

[0017] The method, device and electronic device for assisting in predicting latent atrial fibrillation according to the present invention obtain an initial electromagnetic echo signal, perform signal separation processing on the initial electromagnetic echo signal to obtain a target electromagnetic echo signal corresponding to the motion state of the heart region of the measured object within a predetermined time period, use the spatio-temporal encoder of the information detection model to perform first feature extraction processing on the target electromagnetic echo signal to generate a plurality of mechanical motion spatio-temporal features related to normal and abnormal heartbeats of the heart region, use the feature extraction encoder of the information detection model to perform second feature extraction processing on the plurality of mechanical motion spatio-temporal features to generate target mechanical motion time features related to abnormal heartbeats of the heart region, and use the detection and evaluation sub-model of the information detection model to perform feature conversion mapping processing on the target mechanical motion time features to obtain the target auxiliary prediction heartbeat information of the measured object. Based on the characteristic that latent atrial fibrillation can cause changes in the heart structure and there are still abnormal mechanical activities even after restoring sinus rhythm, all mechanical activities (abnormal activities and normal activities) related to the heart region are extracted from both the spatial and temporal dimensions by the pre-trained spatio-temporal encoder of the information detection model (latent atrial fibrillation detection model), and then the feature extraction encoder is used to further separate and extract all features including both spatial and temporal dimensions, so as to extract the time features of the heartbeat related to abnormal heartbeats, that is, related to latent atrial fibrillation. Based on the time features of the heartbeat related to latent atrial fibrillation, the measured object is predicted to obtain target auxiliary detection heartbeat information with high accuracy and high reliability. At the same time, when the feature extraction encoder is used to separate and extract the time features of the heartbeat related to latent atrial fibrillation, the features of the mechanical activities based on persistent atrial fibrillation and the time features of the heartbeat related to persistent atrial fibrillation can also be separated and extracted, so as to assist relevant professionals in subsequent processing without repeatedly detecting the measured object, saving costs while improving processing efficiency.

[0018] According to an embodiment of the present invention, further, since the information detection model is constructed according to the spatio-temporal encoder, the feature extraction encoder and the detection and evaluation sub-model, and a large number of model learning and training are performed on the constructed information detection model in advance, so that the information detection model can accurately obtain the target auxiliary prediction heartbeat information of the measured object, thereby obtaining an information detection model with high robustness, high reliability, high efficiency and high output accuracy. Brief Description of the Drawings

[0019] Through the following description of the embodiments of the present invention with reference to the drawings, the above content and other objects, features and advantages of the present invention will become clearer. In the drawings:

[0020] Figure 1 An application scenario diagram of the method for assisting in predicting latent atrial fibrillation according to an embodiment of the present invention is shown;

[0021] Figure 2 Shows a flowchart of a method for assisting in predicting latent atrial fibrillation according to an embodiment of the present invention;

[0022] Figure 3a Shows a schematic diagram of a comparison between an electrocardiogram of latent atrial fibrillation and an electrocardiogram of normal sinus rhythm according to an embodiment of the present invention;

[0023] Figure 3b Shows a schematic diagram of a comparison between a target electromagnetic echo signal of latent atrial fibrillation and a target electromagnetic echo signal of normal sinus rhythm according to an embodiment of the present invention;

[0024] Figure 4 Shows a schematic diagram of feature visualization of a feature extraction encoder of the present invention during the training process according to an embodiment of the present invention;

[0025] Figure 5a Shows a schematic diagram of the result output by an untrained spatio-temporal encoder according to an embodiment of the present invention;

[0026] Figure 5b Shows a schematic diagram of the result output by a spatio-temporal encoder trained by the prior art according to an embodiment of the present invention;

[0027] Figure 5c Shows a schematic diagram of the result output by a spatio-temporal encoder trained by the method of the present invention according to an embodiment of the present invention;

[0028] Figure 6a Shows a schematic diagram of the result output by an untrained detection and evaluation sub-model according to an embodiment of the present invention;

[0029] Figure 6b Shows a schematic diagram of the result output by a detection and evaluation sub-model trained by the prior art according to an embodiment of the present invention;

[0030] Figure 6c Shows a schematic diagram of the result output by a detection and evaluation sub-model trained by the method of the present invention according to an embodiment of the present invention;

[0031] Figure 7a Shows a schematic diagram of the excellence degree of various prediction indexes obtained after assisted detection using the method of the present invention according to an embodiment of the present invention;

[0032] Figure 7b Shows a schematic diagram of the target assisted prediction heartbeat information obtained after assisted detection using the method of the present invention when the monitoring time is increased according to an embodiment of the present invention;

[0033] Figure 8Shows a schematic diagram of the comparison between the various indicators of the target auxiliary predicted heartbeat information obtained by using the electrocardiogram of the prior art and the method of the present invention for auxiliary detection respectively according to an embodiment of the present invention;

[0034] Figure 9 Shows a schematic diagram of the process for auxiliary prediction of latent atrial fibrillation according to an embodiment of the present invention;

[0035] Figure 10 Shows a structural block diagram of an apparatus for auxiliary prediction of latent atrial fibrillation according to an embodiment of the present invention;

[0036] Figure 11 Shows a block diagram of an electronic device for a method of auxiliary prediction of latent atrial fibrillation according to an embodiment of the present invention. Detailed implementation manners

[0037] Hereinafter, embodiments of the present invention will be described with reference to the accompanying drawings. However, it should be understood that these descriptions are merely exemplary and are not intended to limit the scope of the present invention. In the following detailed description, for the sake of explanation, many specific details are set forth in order to provide a thorough understanding of the embodiments of the present invention. However, it is obvious that one or more embodiments can be implemented without these specific details. In addition, in the following description, descriptions of well-known structures and technologies are omitted to avoid unnecessarily obscuring the concepts of the present invention.

[0038] The terms used herein are merely for describing specific embodiments and are not intended to limit the present invention. The terms "including", "comprising", etc. used herein indicate the presence of the described features, steps, operations, and / or components, but do not exclude the presence or addition of one or more other features, steps, operations, or components.

[0039] All terms used herein (including technical and scientific terms) have the meanings commonly understood by those skilled in the art, unless otherwise defined. It should be noted that the terms used herein should be interpreted as having a meaning consistent with the context of this specification and should not be interpreted in an idealized or overly rigid manner.

[0040] In the case of using expressions such as "at least one of A, B, and C", generally, it should be interpreted according to the meaning commonly understood by those skilled in the art (for example, "a system having at least one of A, B, and C" should include, but not be limited to, a system having only A, only B, only C, having A and B, having A and C, having B and C, and / or having A, B, and C, etc.).

[0041] In the technical solution of the present invention, the user information involved (including but not limited to user personal information, user image information, user device information, such as location information, etc.) and data (including but not limited to data for analysis, stored data, displayed data, etc.) are all information and data authorized by the user or fully authorized by all parties. Moreover, the processing of relevant data, such as collection, storage, use, processing, transmission, provision, disclosure, and application, all complies with relevant laws, regulations, and standards, adopts necessary confidentiality measures, does not violate public order and good customs, and provides corresponding operation entrances for users to choose to authorize or refuse.

[0042] Atrial fibrillation is a common type of arrhythmia and also a relatively common cardiovascular disease. Moreover, atrial fibrillation increases the risk of stroke or other serious problems. An electrocardiogram can sense the electrical activity of the heart through body-attached electrodes to detect the rhythm abnormalities caused by the onset of atrial fibrillation. However, due to the paroxysmal and concealed nature of atrial fibrillation, the electrocardiogram detection effect for latent atrial fibrillation in atrial fibrillation is poor, resulting in relevant personnel being unable to accurately diagnose the target object based on the obtained electrocardiogram, thus delaying subsequent treatment. And by increasing the frequency and time of electrocardiogram examinations to obtain accurate electrocardiogram monitoring model results to help relevant personnel make judgments, but the multi-frequency detection leads to poor efficiency and cannot be applied on a large scale.

[0043] On this basis, according to the characteristics of latent atrial fibrillation, the prior art captures the tiny signal changes that appear in the electrocardiogram signal to assist relevant personnel in diagnosing the abnormal heartbeat information of the captured target object. However, the above prior art still cannot accurately obtain relevant auxiliary information, resulting in the inability to judge the heartbeat condition of the target object and delaying subsequent treatment, nor can it be widely applied in actual operations. During the research and development process, the R & D personnel found that due to the paroxysmal and concealed characteristics of latent atrial fibrillation in the relevant technology, it is impossible to directly obtain an accurate auxiliary image containing latent atrial fibrillation information, thus unable to timely assist relevant personnel in making predictive judgments. Moreover, when obtaining latent atrial fibrillation information, it is necessary to use contact devices to detect the target multiple times, resulting in technical problems of low processing efficiency and poor flexibility.

[0044] In view of this, an embodiment of the present invention provides a method for assisting in predicting latent atrial fibrillation, including: obtaining an initial electromagnetic echo signal, where the initial electromagnetic echo signal represents a plurality of signals that are spatially aliased with each other and reflected by multiple body regions of a subject within a predetermined time period; performing signal separation processing on the initial electromagnetic echo signal to obtain a target electromagnetic echo signal corresponding to the motion state of the heart region of the subject within the predetermined time period; using the spatio-temporal encoder of the information detection model to perform first feature extraction processing on the target electromagnetic echo signal to generate a plurality of mechanical motion spatio-temporal features related to normal and abnormal heartbeats of the heart region; using the feature extraction encoder of the information detection model to perform second feature extraction processing on the plurality of mechanical motion spatio-temporal features to generate target mechanical motion time features related to abnormal heartbeats of the heart region; using the detection and evaluation sub-model of the information detection model to perform feature transformation and mapping processing on the target mechanical motion time features to obtain the target assisted prediction heartbeat information of the subject.

[0045] Figure 1 Fig. shows an application scenario diagram of a method for assisting in predicting latent atrial fibrillation according to an embodiment of the present invention.

[0046] As Figure 1 shown, the application scenario according to this embodiment may include a first radar device 101, a target object 102, and a receiver 103. The first radar device 101 is used to send an electromagnetic wave signal to the target object 102.

[0047] The user can use the first radar device 101 to interact with the target object 102 and the receiver 103 to receive or send signals, etc.

[0048] The receiver 103 can be a receiver that receives various electromagnetic echo signals, such as receiving and processing the electromagnetic echo signal sent by the first radar device 101 (for example only). The receiver 103 can analyze and process data such as the received electromagnetic echo signal, and feedback the processing result (such as a web page, information, or data obtained or generated according to the user request) to the terminal device.

[0049] It should be noted that the method for assisting in predicting latent atrial fibrillation provided by the embodiments of the present invention can generally be executed by the receiver 103. Correspondingly, the device for assisting in predicting latent atrial fibrillation provided by the embodiments of the present invention can generally be arranged in the receiver 103. The method for assisting in predicting latent atrial fibrillation provided by the embodiments of the present invention can also be executed by a receiver or a cluster of receivers different from the receiver 103 and capable of communicating with the first radar device 101 and / or the receiver 103. Correspondingly, the device for assisting in predicting latent atrial fibrillation provided by the embodiments of the present invention can also be arranged in a receiver or a cluster of receivers different from the receiver 103 and capable of communicating with the first radar device 101 and / or the receiver 103.

[0050] It should be understood that Figure 1 the numbers of the first radar device, the target object, and the receiver in

[0051] are merely illustrative. According to the implementation requirements, there can be any number of radar devices, observed objects, and receivers. Figure 1 The following will be based on Figures 2 to 9 the described scenario, and will describe in detail the method for assisting in predicting latent atrial fibrillation according to the embodiments of the invention through

[0052] Figure 2 FIG. shows a flowchart of the method for assisting in predicting latent atrial fibrillation according to an embodiment of the present invention.

[0053] As Figure 2 shown, the method for assisting in predicting latent atrial fibrillation in this embodiment includes operations S210 to S250.

[0054] In operation S210, an initial electromagnetic echo signal is acquired.

[0055] According to an embodiment of the present invention, the initial electromagnetic echo signal characterizes a plurality of signals that are spatially aliased with each other and reflected by multiple body regions of the object under test within a predetermined time period.

[0056] According to an embodiment of the present invention, an electromagnetic signal can be transmitted to the object under test by using a millimeter-wave radar. Since the millimeter-wave radar can be located directly above the object under test, multiple body regions of the object under test can receive the electromagnetic signal and generate electromagnetic echo signals. When the multiple electromagnetic echo signals are transmitted in the channel, they are spatially aliased with each other to obtain the initial electromagnetic echo signal, where the millimeter-wave radar can have multiple transmitting antennas and multiple receiving antennas.

[0057] In operation S220, signal separation processing is performed on the initial electromagnetic echo signal to obtain a target electromagnetic echo signal corresponding to the motion state of the cardiac region of the object under test within a predetermined time period.

[0058] According to an embodiment of the present invention, since the initial electromagnetic echo signal includes electromagnetic echo signals from non-heart regions, it is necessary to separate multiple spatially aliased electromagnetic echo signals in order to extract the initial electromagnetic echo signal corresponding only to the heart region, and then process the initial electromagnetic echo signal corresponding to the heart region to obtain a target electromagnetic echo signal corresponding to the motion state of the heart region.

[0059] In operation S230, the spatio-temporal encoder of the information detection model is used to perform a first feature extraction process on the target electromagnetic echo signal to generate multiple mechanical motion spatio-temporal features related to normal and abnormal heartbeats in the heart region. According to an embodiment of the present invention, the information detection model can be characterized as a latent atrial fibrillation detection model, and the spatio-temporal encoder of the information detection model can be an encoder built based on the R(2+1)D architecture. Among them, the R(2+1)D architecture includes a 2D spatial convolutional network and a 1D temporal convolutional network. By using the spatio-temporal encoder built based on the R(2+1)D architecture to perform feature extraction on the target electromagnetic echo signal, while obtaining multiple mechanical motion spatio-temporal features, the computational complexity can be reduced and the number of non-linear layers can be increased, so as to facilitate the model to learn more complex spatio-temporal feature representations. Among them, the generated mechanical motion spatio-temporal features can be 128-dimensional feature vectors.

[0060] According to an embodiment of the present invention, abnormal heartbeats can include the beating information of latent atrial fibrillation and the beating information of persistent atrial fibrillation.

[0061] In operation S240, the feature extraction encoder of the information detection model is used to perform a second feature extraction process on the multiple mechanical motion spatio-temporal features to generate target mechanical motion time features related to abnormal heartbeats in the heart region.

[0062] According to an embodiment of the present invention, the feature extraction encoder of the information detection model can be built based on the encoder of the neural network (Transformer) architecture based on the self-attention mechanism. The multiple mechanical motion spatio-temporal features are subjected to feature extraction processing related to the time change of heartbeat information through the feature extraction encoder of the information detection model to generate target mechanical motion time features related to abnormal heartbeats in the heart region. Among them, the target mechanical motion time features can be a feature vector matrix of the number of mechanical motion spatio-temporal features × the dimension of the mechanical motion spatio-temporal features. For example, there are a total of 5 mechanical motion spatio-temporal features, and each mechanical motion spatio-temporal feature is a 128-dimensional feature vector. Then the target mechanical motion time features output by the feature extraction encoder of the information detection model are a 5×128 feature vector matrix.

[0063] According to an embodiment of the present invention, the relevant characteristics of the temporal variation of the beating information of latent atrial fibrillation can be characterized as weak peristalsis of the mechanical movement of the heart within a short period of time, that is, the signal variation amplitude is very small within a short period of time. The relevant characteristics of the temporal variation of the beating information of persistent atrial fibrillation can be characterized as weak peristalsis of the mechanical movement of the heart within a relatively long period of time, that is, the signal variation amplitude is very small within a long period of time.

[0064] In operation S250, the detection and evaluation sub-model of the information detection model performs a feature transformation mapping process on the target mechanical movement time feature to obtain the target auxiliary prediction heartbeat information of the object under test.

[0065] According to an embodiment of the present invention, the target auxiliary prediction information may include whether the heartbeat of the object under test belongs to a first predetermined category. For example, in the case of belonging to the first predetermined category, the obtained target auxiliary prediction heartbeat information may include predicting that the object under test may carry latent atrial fibrillation. In the case of not belonging to the first predetermined category, the obtained target auxiliary prediction heartbeat information may include predicting that the object under test may carry persistent atrial fibrillation or the object under test does not carry atrial fibrillation. Further, in the case where the target auxiliary prediction heartbeat information is predicting that the object under test may carry latent atrial fibrillation, the target auxiliary prediction heartbeat information may further include predicting whether the latent atrial fibrillation carried by the object under test belongs to any one of multiple degree levels. For example, in the case of belonging to the first degree level, the obtained target auxiliary prediction heartbeat information may include predicting that the latent atrial fibrillation that the object under test may carry has a relatively mild degree. In the case of belonging to the second degree level, the obtained target auxiliary prediction heartbeat information may include predicting that the latent atrial fibrillation that the object under test may carry has a general degree. In the case of belonging to the third degree level, the obtained target auxiliary prediction heartbeat information may include predicting that the latent atrial fibrillation that the object under test may carry has a relatively severe degree. Here, it should be noted that the target auxiliary prediction heartbeat information obtained here is only used as reference information for medical staff to make a diagnosis and does not serve as a direct diagnosis result.

[0066] For example, an electromagnetic signal is emitted to a body region of a subject to be measured, and a receiving antenna receives a plurality of electromagnetic echo signals that are spatially aliased with each other during transmission in a channel, that is, initial electromagnetic echo signals. The initial electromagnetic echo signals are subjected to a separation process to extract target electromagnetic echo signals that correspond only to the heartbeat motion state of the heart region of the subject to be measured. Then, the target electromagnetic echo signals are subjected to processes such as feature extraction and feature transformation mapping using a spatio-temporal encoder, a feature extraction encoder, and a detection and evaluation sub-model in an information detection model to obtain whether the heartbeat of the subject to be measured belongs to a first predetermined category and to predict the degree level of latent atrial fibrillation carried by the subject to be measured, that is, target auxiliary prediction heartbeat information, so as to facilitate scientific personnel or professionals in related fields to distinguish the heartbeat of the subject to be measured based on the target auxiliary prediction heartbeat information, and thus make a judgment suitable for the subject to be measured.

[0067] According to an embodiment of the present invention, by acquiring initial electromagnetic echo signals, performing a signal separation process on the initial electromagnetic echo signals to obtain target electromagnetic echo signals corresponding to the motion state of the heart region of the subject to be measured within a predetermined time period, using the spatio-temporal encoder of the information detection model to perform a first feature extraction process on the target electromagnetic echo signals to generate a plurality of mechanical motion spatio-temporal features related to normal and abnormal heartbeats in the heart region, using the feature extraction encoder of the information detection model to perform a second feature extraction process on the plurality of mechanical motion spatio-temporal features to generate target mechanical motion time features related to abnormal heartbeats in the heart region, and using the detection and evaluation sub-model of the information detection model to perform a feature transformation mapping process on the target mechanical motion time features to obtain target auxiliary prediction heartbeat information of the subject to be measured, it is realized that based on the characteristic that latent atrial fibrillation can cause changes in the heart structure and there is still abnormal mechanical activity even after the restoration of sinus rhythm, all mechanical activities (abnormal activities and normal activities) related to the heart region are extracted from both the spatial and temporal dimensions using the pre-trained spatio-temporal encoder of the information detection model (latent atrial fibrillation detection model), and then the feature extraction encoder is used to further separate and extract all features including both spatial and temporal dimensions, so as to extract the time features of the heartbeat related to abnormal heartbeats, that is, related to latent atrial fibrillation. Based on the time features of the heartbeat related to latent atrial fibrillation, the subject to be measured is predicted to obtain target auxiliary detection heartbeat information with high accuracy and high reliability. At the same time, when using the feature extraction encoder to separate and extract the time features of the heartbeat related to latent atrial fibrillation, the features of the mechanical activity based on persistent atrial fibrillation can also be separated and extracted, and the time features of the heartbeat related to persistent atrial fibrillation can also be separated and extracted, so as to assist relevant professionals in subsequent processing, without repeatedly detecting the subject to be measured multiple times, saving costs while improving processing efficiency.

[0068] According to an embodiment of the present invention, further, since an information detection model is built based on a spatio-temporal encoder, a feature extraction encoder, and a detection and evaluation sub-model, and through a large amount of model learning and training of the built information detection model in advance, the information detection model can accurately obtain the target auxiliary detection heartbeat information of the object to be measured, thereby obtaining an information detection model with high robustness, high reliability, high efficiency, and high output accuracy.

[0069] According to an embodiment of the present invention, furthermore, the present invention uses a non-contact radar device to perform auxiliary detection on the object to be measured, which can improve the detection experience of the object to be measured and the application scenarios and scope of application for the auxiliary detection of cryptogenic atrial fibrillation of the object to be measured.

[0070] It should be noted that the obtained target auxiliary detection heartbeat information of the present invention is only used as an intermediate result, and a diagnostic result or health status cannot be directly obtained from the target auxiliary detection heartbeat information obtained by the method according to the present invention.

[0071] According to an embodiment of the present invention, the method for performing signal separation processing on the initial electromagnetic echo signal to obtain the target electromagnetic echo signal corresponding to the motion state of the cardiac region of the object to be measured within a predetermined time period includes the following operations.

[0072] According to an embodiment of the present invention, beamforming processing is performed on the initial electromagnetic echo signal to obtain a plurality of intermediate electromagnetic echo signals.

[0073] According to an embodiment of the present invention, the intermediate electromagnetic echo signal represents a plurality of four-dimensional signals that are spatially separated from each other and reflected by a plurality of regions within a predetermined time period.

[0074] According to an embodiment of the present invention, the initial electromagnetic echo signal can be a four-dimensional signal, and the initial electromagnetic echo signal can be represented according to formula (1), and formula (1) is as follows.

[0075] (1);

[0076] Wherein, b(t) can represent the initial electromagnetic echo signal, h can represent the energy attenuation of the electromagnetic echo signal in the propagation path, j can represent the imaginary unit, d tx can represent the propagation distance of the electromagnetic signal from the transmitting antenna to the object to be measured, d rx can represent the distance of the electromagnetic echo signal reflected from the object to be measured to the receiving antenna, d e (t)can represent the motion change information of the objects in the environment, and the motion change information includes heartbeat, breathing, and other motion information, can represent the wavelength of the electromagnetic signal.

[0077] According to an embodiment of the present invention, based on the spatial resolution provided by the antenna array of the radar device and the large bandwidth of the radar, by performing beamforming processing on the initial electromagnetic echo signal, the electromagnetic echo signals reflected from different body regions to the receiving antenna, that is, the initial electromagnetic echo signals, can be separated to obtain multiple intermediate electromagnetic echo signals, so as to facilitate the extraction of the intermediate electromagnetic echo signal corresponding to the heart region.

[0078] According to an embodiment of the present invention, a three-dimensional electromagnetic echo signal corresponding to the heart region of the object to be measured is extracted from multiple intermediate electromagnetic echo signals.

[0079] According to an embodiment of the present invention, by determining the distance information of each intermediate electromagnetic echo signal and combining the distance information between the object to be measured and the radar receiving antenna array, the three-dimensional electromagnetic echo signal corresponding to the heart region can be distinguished and determined.

[0080] According to an embodiment of the present invention, the intermediate electromagnetic echo signal after beamforming processing includes information in four dimensions, that is, distance dimension information, azimuth angle dimension information, elevation angle dimension information, and Doppler frequency shift dimension information. Through the distance information between the object to be measured and the radar receiving antenna array, the three-dimensional electromagnetic echo signal corresponding to the heart region can be determined, and the three-dimensional electromagnetic echo signal can be characterized as a signal including information in three dimensions, that is, azimuth angle dimension information, elevation angle dimension information, and Doppler frequency shift dimension information.

[0081] According to an embodiment of the present invention, the three-dimensional electromagnetic echo signal is filtered by using a second-order difference time-domain filtering algorithm to obtain a target electromagnetic echo signal.

[0082] According to an embodiment of the present invention, the three-dimensional electromagnetic echo signal corresponding to the heart region may include heartbeat motion characteristics, breathing motion characteristics, and other motion characteristics. Based on the different accelerations between different motion characteristics, the second-order difference time-domain filtering algorithm can eliminate the motion characteristics other than the heartbeat motion characteristics with a larger acceleration, so as to obtain a target electromagnetic echo signal that only contains the heartbeat motion state corresponding to the heart region of the object to be measured.

[0083] According to an embodiment of the present invention, the second-order difference signal used in the second-order difference time-domain filtering algorithm can be represented according to formula (2), and formula (2) is shown as follows.

[0084] (2);

[0085] where s ’’(l, t) can be characterized as the second-order difference signal of the t-th frame at position l, △t can be characterized as the time interval between two adjacent frames of three-dimensional electromagnetic echo signals, s(l, t) can be characterized as the three-dimensional electromagnetic echo signal of the t-th frame after beamforming processing at position l, s(l, t + 1) can be characterized as the three-dimensional electromagnetic echo signal of the (t + 1)-th frame after beamforming processing at position l, s(l, t - 1) can be characterized as the three-dimensional electromagnetic echo signal of the (t - 1)-th frame after beamforming processing at position l, s(l, t + 2) can be characterized as the three-dimensional electromagnetic echo signal of the (t + 2)-th frame after beamforming processing at position l, s(l, t - 2) can be characterized as the three-dimensional electromagnetic echo signal of the (t - 2)-th frame after beamforming processing at position l, s(l, t + 3) can be characterized as the three-dimensional electromagnetic echo signal of the (t + 3)-th frame after beamforming processing at position l, s(l, t - 3) can be characterized as the three-dimensional electromagnetic echo signal of the (t - 3)-th frame after beamforming processing at position l.

[0086] According to an embodiment of the present invention, by performing beamforming processing on the initial electromagnetic echo signal, a plurality of intermediate electromagnetic echo signals are obtained. From the plurality of intermediate electromagnetic echo signals, a three-dimensional electromagnetic echo signal corresponding to the heart region of the object to be measured is extracted. The three-dimensional electromagnetic echo signal is filtered by using a second-order difference time-domain filtering algorithm to obtain a target electromagnetic echo signal, so as to accurately separate and extract the three-dimensional electromagnetic echo signal corresponding only to the heart region from the spatially aliased initial electromagnetic echo signal, and eliminate features other than the heartbeat feature from the three-dimensional electromagnetic echo signal, thereby obtaining a heartbeat signal corresponding only to the heart region, that is, the target electromagnetic echo signal, so as to facilitate the analysis and processing of the target electromagnetic echo signal, and thereby obtain the target auxiliary predicted heartbeat information of the object to be measured.

[0087] According to an embodiment of the present invention, the method for extracting the three-dimensional electromagnetic echo signal corresponding to the heart region of the object to be measured from the plurality of intermediate electromagnetic echo signals includes the following operations.

[0088] According to an embodiment of the present invention, the plurality of intermediate electromagnetic echo signals are superimposed to obtain an intermediate electromagnetic echo superimposed signal.

[0089] According to an embodiment of the present invention, the constant false alarm rate algorithm is used to process the intermediate electromagnetic echo superimposed signal to obtain a distance electromagnetic echo signal corresponding to the chest region of the object to be measured.

[0090] According to an embodiment of the present invention, based on the fact that the chest cavity region has a large receiving area and good reflection characteristics, so that the intensity of the reflected electromagnetic echo signal is large. According to the distance information between the object to be measured and the radar receiving antenna array, the distance cell information of each intermediate electromagnetic echo signal in the intermediate electromagnetic echo superposition signal is determined by using the CFAR algorithm (Constant False Alarm Rate algorithm), so as to determine a plurality of distance electromagnetic echo signals corresponding to the chest cavity region of the object to be measured. Among them, the distance cell information of the distance electromagnetic echo signal can be consistent with the distance information between the object to be measured and the radar receiving antenna array.

[0091] According to an embodiment of the present invention, the distance electromagnetic echo signal is a three-dimensional signal, that is, a signal including information in three dimensions: azimuth angle dimension information, elevation angle dimension information, and Doppler frequency shift dimension information.

[0092] According to an embodiment of the present invention, the density clustering algorithm is used to perform clustering processing on the distance electromagnetic echo signal to obtain a three-dimensional electromagnetic echo signal corresponding to the heart region of the object to be measured.

[0093] According to an embodiment of the present invention, since there can be multiple distance cell information that is consistent with the distance information between the object to be measured and the radar receiving antenna array, according to the angle information between the object to be measured and the radar receiving antenna array, based on the characteristic that the amplitude of the chest cavity movement is consistent, the DBSCAN algorithm (Density-Based Spatial Clustering of Applications with Noise) is used to perform clustering processing on a plurality of distance electromagnetic echo signals to obtain a three-dimensional electromagnetic echo signal corresponding to the heart region of the object to be measured.

[0094] According to an embodiment of the present invention, by performing superposition processing on a plurality of intermediate electromagnetic echo signals, an intermediate electromagnetic echo superposition signal is obtained, and the CFAR algorithm is used to process the intermediate electromagnetic echo superposition signal to obtain a distance electromagnetic echo signal corresponding to the chest cavity region of the object to be measured. Among them, the distance electromagnetic echo signal is a three-dimensional signal; the density clustering algorithm is used to perform clustering processing on the distance electromagnetic echo signal to obtain a three-dimensional electromagnetic echo signal corresponding to the heart region of the object to be measured, realizing the determination of an accurate three-dimensional electromagnetic echo signal corresponding to the heart region based on dimension feature information such as distance and angle, combined with the reflection characteristics and motion characteristics of the chest cavity region.

[0095] According to an embodiment of the present invention, a method for generating a plurality of mechanical motion spatio-temporal features related to normal and abnormal heartbeats of the heart region by performing first feature extraction processing on the target electromagnetic echo signal using the spatio-temporal encoder of the information detection model includes the following operations.

[0096] According to an embodiment of the present invention, based on a predetermined coding period, a target electromagnetic echo signal is split into target electromagnetic echo sub-signals of multiple periods.

[0097] For example, if the initially acquired initial electromagnetic echo signal is 10 s, then the processed target electromagnetic echo signal is also 10 s, and the predetermined coding period is 2 s. Based on the predetermined coding period, the 10-s target electromagnetic echo signal is split into target electromagnetic echo sub-signals of 0-2 s, 2-4 s, 4-6 s, 6-8 s, and 8-10 s, so as to facilitate the spatio-temporal encoder of the information detection model to perform feature extraction processing on the target electromagnetic echo sub-signals.

[0098] According to an embodiment of the present invention, the target electromagnetic echo sub-signals of multiple periods are respectively input into the spatio-temporal encoder of the information detection model for first feature extraction processing, and a plurality of mechanical motion spatio-temporal features related to normal heartbeats and abnormal heartbeats in the heart region are generated.

[0099] According to an embodiment of the present invention, the plurality of mechanical motion spatio-temporal features correspond to the target electromagnetic echo sub-signals of multiple periods.

[0100] According to an embodiment of the present invention, the mechanical motion spatio-temporal feature can be characterized as an accurate motion feature of the heart of the object to be measured in the time dimension and the space dimension, wherein the mechanical motion spatio-temporal feature can be a 128-dimensional feature vector.

[0101] According to an embodiment of the present invention, by splitting the target electromagnetic echo signal into target electromagnetic echo sub-signals of multiple periods based on a predetermined coding period, it is convenient for the spatio-temporal encoder of the information detection model to perform feature extraction processing on the target electromagnetic echo sub-signals, and the speed and efficiency of feature extraction of the spatio-temporal encoder are improved. Then, the target electromagnetic echo sub-signals of multiple periods are respectively input into the spatio-temporal encoder of the information detection model for first feature extraction processing, and a plurality of mechanical motion spatio-temporal features related to normal heartbeats and abnormal heartbeats in the heart region are generated, realizing the conversion of the electromagnetic echo signal into an accurate motion feature of the heart in the time dimension and the space dimension through the information detection model, so as to facilitate analysis and processing based on the real and accurate motion features.

[0102] According to an embodiment of the present invention, the detection and evaluation sub-model includes a first double-layer fully connected network and a second double-layer fully connected network, and both the first double-layer fully connected network and the second double-layer fully connected network include two linear layers.

[0103] According to an embodiment of the present invention, a method for obtaining target auxiliary predicted heartbeat information of a measured object by performing feature transformation mapping processing on target mechanical motion time characteristics using a detection and evaluation sub-model of an information detection model includes the following operations.

[0104] According to an embodiment of the present invention, the target mechanical motion time characteristics are respectively input into the first linear layer of the first double-layer fully connected network and the first linear layer of the second double-layer fully connected network for dimensionality reduction processing, to obtain a first dimensionality-reduced mechanical motion state characteristic and a second dimensionality-reduced mechanical motion time characteristic.

[0105] According to an embodiment of the present invention, both the first dimensionality-reduced mechanical motion state characteristic and the second dimensionality-reduced mechanical motion time characteristic after dimensionality reduction processing are 10-dimensional feature vectors. Among them, the first dimensionality-reduced mechanical motion state characteristic may be a characteristic related to the mechanical motion of the heart with abnormal heartbeats, and the second dimensionality-reduced mechanical motion time characteristic may be a characteristic related to the change trend of mechanical motion during the period when the heart has abnormal heartbeats.

[0106] According to an embodiment of the present invention, the first dimensionality-reduced mechanical motion state characteristic is input into the second linear layer of the first double-layer fully connected network for feature transformation mapping processing, to obtain the auxiliary atrial beating information of the measured object.

[0107] According to an embodiment of the present invention, the second linear layer of the first double-layer fully connected network performs feature transformation mapping processing on the input first dimensionality-reduced mechanical motion state characteristic, to generate the target heart beating state characteristic of the measured object. Among them, the target heart beating state characteristic is a 2-dimensional feature vector, and the target heart beating state characteristic can represent whether the target electromagnetic echo signal of the measured object contains the target mechanical motion time characteristic related to abnormal heartbeats. Thus, based on the target heart beating state characteristic, the auxiliary atrial beating information of the measured object is obtained. For example, when the target electromagnetic echo signal of the measured object contains the target mechanical motion time characteristic related to abnormal heartbeats, the auxiliary atrial beating information of the measured object may belong to a first predetermined category, that is, the measured object may carry auxiliary information of latent atrial fibrillation. On this basis, it should be noted that the auxiliary atrial beating information is only an intermediate result, and a diagnostic result or health condition cannot be directly obtained from the target auxiliary detected heartbeat information obtained by the method according to the present invention.

[0108] According to an embodiment of the present invention, the second dimensionality-reduced mechanical motion time characteristic is input into the second linear layer of the second double-layer fully connected network for feature transformation mapping processing, to obtain the auxiliary atrial fibrillation information of the measured object.

[0109] According to an embodiment of the present invention, the second linear layer of the second double-layer fully connected network performs a feature transformation mapping process on the input second dimension-reduced mechanical motion time feature to generate a target heart beat change trend feature of the object to be measured. The target heart beat change trend feature is a 1D feature vector, and the target heart beat change trend feature can represent the change trend of the signal related to abnormal heart beats in the target electromagnetic echo signal of the object to be measured. Thus, based on the target heart beat change trend feature, auxiliary atrial fibrillation information of the object to be measured is obtained. For example, when the target heart beat change trend feature of the object to be measured is that the mechanical motion of the heart is a weak peristalsis within a short time, the auxiliary atrial fibrillation information of the object to be measured may belong to the first degree level, that is, the degree of latent atrial fibrillation that the object to be measured may carry is relatively light; when the target heart beat change trend feature of the object to be measured is that the mechanical motion of the heart is a weak peristalsis within a long time, the auxiliary atrial fibrillation information of the object to be measured may belong to the third degree level, that is, the degree of latent atrial fibrillation that the object to be measured may carry is relatively serious. On this basis, it should be noted that the auxiliary atrial fibrillation information is only an intermediate result and cannot directly obtain a diagnosis result or health status from the target auxiliary prediction heart beat information obtained by the method according to the present invention.

[0110] According to an embodiment of the present invention, by respectively inputting the target mechanical motion time feature into the first linear layer of the first double-layer fully connected network and the first linear layer of the second double-layer fully connected network for dimension reduction processing, a first dimension-reduced mechanical motion state feature and a second dimension-reduced mechanical motion time feature are obtained. The first dimension-reduced mechanical motion state feature is input into the second linear layer of the first double-layer fully connected network for a feature transformation mapping process to obtain auxiliary atrial beat information of the object to be measured. At the same time, the second dimension-reduced mechanical motion time feature is input into the second linear layer of the second double-layer fully connected network for a feature transformation mapping process to obtain auxiliary atrial fibrillation information of the object to be measured, achieving the effect of dimension reduction on the target mechanical motion time feature through the first linear layer in the two double-layer fully connected networks and extracting different features to facilitate the output of different auxiliary information. While performing secondary dimension reduction through the second linear layer, multiple target auxiliary information with high accuracy related to the heart is output in parallel, improving the model processing efficiency.

[0111] According to an embodiment of the present invention, the information detection model is trained by the following method, and the training method includes the following operations.

[0112] According to an embodiment of the present invention, an initial model to be trained and a sample training data set are obtained.

[0113] According to an embodiment of the present invention, the initial model to be trained includes an initial spatio-temporal encoder, a feature extraction encoder to be trained, and a detection and evaluation sub-model to be trained. The sample training data set includes electromagnetic echo signal samples, latent atrial fibrillation disease period information samples, atrial beating information samples, and atrial fibrillation information samples.

[0114] According to an embodiment of the present invention, the latent atrial fibrillation disease period information sample can be characterized as a period sample of the disease time when the sample object has latent atrial fibrillation. The atrial beating information sample can be characterized as an information sample of whether the sample object has latent atrial fibrillation. The atrial fibrillation information sample can be characterized as an information sample of the degree level of latent atrial fibrillation of the sample object.

[0115] According to an embodiment of the present invention, based on the predetermined sample coding period information, the electromagnetic echo signal sample is split into a plurality of electromagnetic echo signal sub-samples.

[0116] According to an embodiment of the present invention, the plurality of electromagnetic echo signal sub-samples are input into the initial spatio-temporal encoder of the initial model to be trained for the first feature extraction process, generating a plurality of training mechanical motion spatio-temporal features related to normal and abnormal heartbeats in the heart region.

[0117] According to an embodiment of the present invention, the training mechanical motion spatio-temporal feature can be a 128-dimensional feature vector.

[0118] According to an embodiment of the present invention, the plurality of mechanical motion spatio-temporal training features are input into the feature extraction encoder to be trained for the second feature extraction process, generating training mechanical motion time features related to abnormal heartbeats in the heart region.

[0119] According to an embodiment of the present invention, the training mechanical motion time feature can be a feature vector of the number of training mechanical motion spatio-temporal features × the dimension of the training mechanical motion spatio-temporal feature.

[0120] According to an embodiment of the present invention, the training mechanical motion time feature is input into the detection and evaluation sub-model to be trained for the feature conversion mapping process, obtaining training detection heartbeat information.

[0121] According to an embodiment of the present invention, the training detection heartbeat information can include training atrial beating information and training atrial fibrillation information.

[0122] According to an embodiment of the present invention, based on the detection and evaluation loss function, according to the training detection heartbeat information, latent atrial fibrillation disease period information sample, atrial beating information sample, and atrial fibrillation information sample, a detection and evaluation loss value is obtained.

[0123] According to an embodiment of the present invention, according to the detection and evaluation loss value, the model parameters of the initial model are adjusted to obtain a trained information detection model.

[0124] According to an embodiment of the present invention, based on the detected and evaluated loss value, the weight parameters of the initial spatio-temporal encoder, the feature extraction encoder to be trained, and the detection and evaluation sub-model to be trained can be adjusted respectively, so as to obtain an information detection model with a detection and evaluation loss value less than a predetermined evaluation loss threshold, that is, a trained information detection model.

[0125] According to an embodiment of the present invention, in addition to setting a predetermined evaluation loss threshold to help determine whether the information detection model meets the target standard, the number of training times can also be set to help determine whether the information detection model meets the target standard, so as to complete the training.

[0126] According to an embodiment of the present invention, by using the samples in the sample training dataset to train the initial spatio-temporal encoder, the feature extraction encoder to be trained, and the detection and evaluation sub-model to be trained in the initial model to be trained, and then combining the information samples in the sample training dataset, based on the calculated detection and evaluation loss value, an information detection model can be obtained that can output accurate features and auxiliary result information according to the electromagnetic echo signal, improving the robustness and maturity of the information detection model during the learning and training process, so as to help relevant professionals improve the processing efficiency and the efficiency of auxiliary detection of the object to be measured.

[0127] According to an embodiment of the present invention, the initial spatio-temporal encoder is pre-trained, and the pre-training method includes the following operations.

[0128] According to an embodiment of the present invention, a basic spatio-temporal encoder to be trained is obtained.

[0129] According to an embodiment of the present invention, the basic spatio-temporal encoder to be trained can be a basic encoder built based on the R(2+1)D architecture. By using the output features of the basic spatio-temporal encoder, features including both the time dimension and the space dimension can be output.

[0130] According to an embodiment of the present invention, a plurality of electromagnetic echo signal sub-samples are input into the basic spatio-temporal encoder to be trained for the first feature extraction process, generating a plurality of basic training mechanical motion spatio-temporal features related to normal heartbeats and abnormal heartbeats in the heart region.

[0131] According to an embodiment of the present invention, the basic training mechanical motion spatio-temporal feature can be a 128-dimensional feature vector.

[0132] According to an embodiment of the present invention, based on a preset contrast weight function for the recessive atrial fibrillation period, the recessive atrial fibrillation period contrast weight is obtained according to the recessive atrial fibrillation disease period information sample.

[0133] According to an embodiment of the present invention, the recessive atrial fibrillation period contrast weight can be calculated according to formula (3), and formula (3) is shown as follows.

[0134] (3);

[0135] where w i，a can be characterized as the weight for comparing recessive atrial fibrillation periods, and y a can be characterized as the a-th information sample of the recessive atrial fibrillation affected period, and y i can be characterized as the i-th information sample of the recessive atrial fibrillation affected period, and eps can be characterized as a non-zero parameter.

[0136] According to an embodiment of the present invention, based on a preset spatio-temporal loss function, regression contrast learning is performed on the weight for comparing recessive atrial fibrillation periods and multiple basic training mechanical motion spatio-temporal features to obtain a spatio-temporal loss value.

[0137] According to an embodiment of the present invention, the spatio-temporal loss value can be calculated according to formula (4), and formula (4) is as follows.

[0138] (4);

[0139] where L regcon can be characterized as the spatio-temporal loss value, I can be characterized as there are a total of I sample indices, that is, there are a total of I information samples of the recessive atrial fibrillation affected period, and A(i) can be characterized as the set of information samples of the recessive atrial fibrillation affected period except the i-th information sample of the recessive atrial fibrillation affected period, and z i can be characterized as the basic training mechanical motion spatio-temporal feature after feature extraction by the basic spatio-temporal encoder corresponding to the i-th information sample of the recessive atrial fibrillation affected period, and z a can be characterized as the basic training mechanical motion spatio-temporal feature after feature extraction by the basic spatio-temporal encoder corresponding to the a-th information sample of the recessive atrial fibrillation affected period, and z b can be characterized as the basic training mechanical motion spatio-temporal feature after feature extraction by the basic spatio-temporal encoder corresponding to the b-th information sample of the recessive atrial fibrillation affected period, can be characterized as the temperature coefficient used in the contrast learning, and can be set to 0.07 during specific implementation.

[0140] According to an embodiment of the present invention, based on the regression contrast learning technique, under the guidance of multiple information samples of the recessive atrial fibrillation affected period, by adjusting the basic spatio-temporal encoder to be trained according to the spatio-temporal loss value, the feature difference between the multiple basic training mechanical motion spatio-temporal features output by the basic spatio-temporal encoder and the multiple information samples of the recessive atrial fibrillation affected period is narrowed or driven away.

[0141] According to an embodiment of the present invention, the weight for comparing recessive atrial fibrillation periods of the basic spatio-temporal encoder to be trained is adjusted according to the spatio-temporal loss value to obtain a trained initial spatio-temporal encoder.

[0142] According to an embodiment of the present invention, during the training of the basic spatio-temporal encoder to be trained, the basic spatio-temporal encoder can be tuned once to obtain an initial spatio-temporal encoder. Then, during the learning and training process of the information detection model to be trained containing the initial spatio-temporal encoder, the initial spatio-temporal encoder and other structural models are tuned once again to obtain a trained information detection model.

[0143] According to an embodiment of the present invention, by obtaining the basic spatio-temporal encoder to be trained, using the sub-samples of electromagnetic echo signals and the information samples of the recessive atrial fibrillation disease period to perform the learning and training of the encoder on the basic spatio-temporal encoder to be trained, the basic spatio-temporal encoder can output the basic training mechanical motion spatio-temporal features. Then, regression contrast learning is used to calculate the loss of the basic spatio-temporal encoder that can output the basic training mechanical motion spatio-temporal features, so as to adjust the weight parameters of the basic spatio-temporal encoder, thereby enhancing the contrast loss between the features of the recessive atrial fibrillation disease period information samples and the basic training mechanical motion spatio-temporal features, weakening the contrast loss between the features of the recessive atrial fibrillation disease period information samples with large differences and the basic training mechanical motion spatio-temporal features, and improving the feature extraction accuracy of the initial spatio-temporal encoder and the robustness of the initial spatio-temporal encoder.

[0144] According to an embodiment of the present invention, the training of detecting heartbeat information includes the training of atrial beating information and the training of atrial fibrillation information.

[0145] According to an embodiment of the present invention, based on the detection and evaluation loss function, according to the training of detecting heartbeat information, the information samples of the recessive atrial fibrillation disease period, the atrial beating information samples, and the atrial fibrillation information samples, obtaining the detection and evaluation loss value includes multiple iteration rounds, where the t-th iteration round includes the following operations, where t≥1 and t is a positive integer.

[0146] According to an embodiment of the present invention, in the t-th iteration round, according to the training atrial beating information and the atrial beating information samples of the t-th iteration round, the cross-entropy loss value corresponding to the t-th iteration round is obtained.

[0147] According to an embodiment of the present invention, according to the multiple training atrial fibrillation information and the atrial fibrillation information samples of the t-th iteration round, the mean square error loss value corresponding to the t-th iteration round is obtained.

[0148] According to an embodiment of the present invention, the training atrial fibrillation information can be the time carrying recessive atrial fibrillation obtained through training by the first double-layer fully connected network, and the atrial fibrillation information sample can be the time when the sample object actually carries recessive atrial fibrillation.

[0149] According to an embodiment of the present invention, the atrial beat loss change rate corresponding to the t-th iteration round is determined based on the average cross-entropy loss corresponding to the t iteration rounds and the average cross-entropy loss corresponding to the t-1 iteration rounds.

[0150] According to an embodiment of the present invention, the atrial beat loss change rate corresponding to the t-th iteration round can be calculated according to formula (5), and formula (5) is shown as follows.

[0151] (5);

[0152] Wherein, can be characterized as the atrial beat loss change rate corresponding to the t-th iteration round, L1(t) can be characterized as the average cross-entropy loss corresponding to the t iteration rounds, and L1(t - 1) can be characterized as the average cross-entropy loss corresponding to the t-1 iteration rounds.

[0153] According to an embodiment of the present invention, the atrial fibrillation loss change rate corresponding to the t-th iteration round is determined based on the average mean square error loss corresponding to the t iteration rounds and the average mean square error loss corresponding to the t-1 iteration rounds.

[0154] According to an embodiment of the present invention, the atrial fibrillation loss change rate corresponding to the t-th iteration round can be calculated according to formula (6), and formula (6) is shown as follows.

[0155] (6);

[0156] Wherein, can be characterized as the atrial fibrillation loss change rate corresponding to the t-th iteration round, L2(t) can be characterized as the average mean square error loss corresponding to the t iteration rounds, and L2(t - 1) can be characterized as the average mean square error loss corresponding to the t-1 iteration rounds.

[0157] According to an embodiment of the present invention, the training loss change rate corresponding to the t-th iteration round is obtained based on the atrial beat loss change rate corresponding to the t-th iteration round and the atrial fibrillation loss change rate corresponding to the t-th iteration round.

[0158] According to an embodiment of the present invention, by adding the atrial beat loss change rate corresponding to the t-th iteration round and the atrial fibrillation loss change rate corresponding to the t-th iteration round, the training loss change rate corresponding to the t-th iteration round can be obtained.

[0159] According to an embodiment of the present invention, based on a preset beat loss change rate function and a preset flutter loss change rate function, an atrial beat loss weighting factor and an atrial flutter loss weighting factor for the t-th iteration round are obtained according to the training loss change rate corresponding to the t-th iteration round, the atrial beat loss change rate corresponding to the t-th iteration round, and the atrial flutter loss change rate corresponding to the t-th iteration round.

[0160] According to an embodiment of the present invention, the atrial beat loss weighting factor for the t-th iteration round can be calculated according to formula (7), and formula (7) is as follows.

[0161] (7);

[0162] Wherein, w1(t) can represent the atrial beat loss weighting factor for the t-th iteration round, N can represent that there are a total of N loss change rates, that is, the t-th iteration round includes an atrial flutter loss change rate and an atrial beat loss change rate, and n can represent the n-th loss change rate, that is, the atrial flutter loss change rate or the atrial beat loss change rate for the t-th iteration round.

[0163] According to an embodiment of the present invention, the atrial flutter loss weighting factor for the t-th iteration round can be calculated according to formula (8), and formula (8) is as follows.

[0164] (8);

[0165] Wherein, w2(t) can represent the atrial flutter loss weighting factor for the t-th iteration round.

[0166] According to an embodiment of the present invention, based on the detection and evaluation loss function, a detection and evaluation loss value for the t-th iteration round is obtained according to the cross-entropy loss value corresponding to the t-th iteration round, the mean squared error loss value corresponding to the t-th iteration round, the atrial beat loss weighting factor for the t-th iteration round, and the atrial flutter loss weighting factor for the t-th iteration round.

[0167] According to an embodiment of the present invention, the detection and evaluation loss value for the t-th iteration round can be calculated according to formula (9), and formula (9) is as follows.

[0168] (9);

[0169] Wherein, L total (t) can represent the detection and evaluation loss value for the t-th iteration round, w n (t) can represent the loss weighting factor for the t-th iteration round, including the atrial flutter loss weighting factor and the atrial beat loss weighting factor for the t-th iteration round, L n(t) can be characterized as the loss value corresponding to the t-th iteration round, including the cross-entropy loss value corresponding to the t-th iteration round and the mean square error loss value corresponding to the t-th iteration round.

[0170] According to an embodiment of the present invention, in the case where the detection and evaluation loss value in the t-th iteration round is greater than a predetermined evaluation loss threshold, the weights of the atrial beating loss weighting factor and the atrial fibrillation loss weighting factor are adjusted, so as to obtain a trained information detection model. For example, in the case where the loss output by any double-layer fully connected network changes rapidly, it indicates that the double-layer fully connected network has a large optimization space, and thus a greater weight is given; conversely, if the loss no longer changes significantly, it indicates that the double-layer fully connected network has tended to be optimal. At this time, the model optimization of the information detection model will focus more on adjusting the parameters of other double-layer fully connected networks or the initial spatio-temporal encoder, so as to obtain an information detection model that can output high-accuracy information.

[0171] According to an embodiment of the present invention, by calculating the respective loss values and loss weighting factors between the first double-layer fully connected network and the second double-layer fully connected network, the total loss value of the detection and evaluation sub-model to be trained is obtained according to the respective loss values and loss weighting factors, so as to adjust the weight parameters in the detection and evaluation sub-model to be trained and the weight parameters of the initial spatio-temporal encoder, so that the information detection model can accurately obtain the target auxiliary prediction heartbeat information of the object to be measured, thereby obtaining an information detection model with high robustness, high reliability, high efficiency and high output accuracy.

[0172] Figure 3a A schematic diagram showing a comparison between the electrocardiogram of latent atrial fibrillation and the electrocardiogram of normal sinus rhythm according to an embodiment of the present invention is shown. Figure 3b A schematic diagram showing a comparison between the target electromagnetic echo signal of latent atrial fibrillation and the target electromagnetic echo signal of normal sinus rhythm according to an embodiment of the present invention is shown.

[0173] As Figures 3a to 3b shown, Figures 3a to 3b A comparison between the electrocardiogram of latent atrial fibrillation and the electrocardiogram of normal sinus rhythm and a comparison between the target electromagnetic echo signal of latent atrial fibrillation and the target electromagnetic echo signal of normal sinus rhythm are shown. As can be seen from Figure 3a it, in the electrocardiogram, the electrocardiogram of the object with normal sinus rhythm and the electrocardiogram of the object with latent atrial fibrillation are almost indistinguishable, the electrocardiogram rhythms of both are normal, and the electrical signals corresponding to atrial beating exist. From Figure 3bIt can be seen that there are obvious differences between the normal electromagnetic echo signals of objects with normal sinus rhythm and the target electromagnetic echo signals of objects with latent atrial fibrillation in the signal diagram of electromagnetic echo signals. The signal change amplitude of the target electromagnetic echo signals of objects with latent atrial fibrillation is very small in a short period of time, and the atrium is in a state of weak peristalsis in a short period of time, while the signal fluctuations of normal electromagnetic echo signals are obvious and the atrial contraction function is normal.

[0174] Figure 4 The figure shows a schematic diagram of the feature visualization of the feature extraction encoder of the present invention during the training process according to an embodiment of the present invention.

[0175] As Figure 4 shown, Figure 4 The figure shows the feature visualization of the feature extraction encoder during the training process. The abscissa can be characterized as the distribution rate of the visualized aggregation area of the trained model, and the ordinate can be characterized as the distribution rate of the visualized aggregation area of the trained model. Blue represents normal heart rate, black represents latent atrial fibrillation, and red represents persistent atrial fibrillation. After one round of training of the feature extraction encoder, the feature extraction encoder cannot clearly distinguish and output normal heart rate, latent atrial fibrillation, and persistent atrial fibrillation. After multiple rounds of training of the feature extraction encoder, the feature extraction encoder can clearly distinguish and output normal heart rate, latent atrial fibrillation, and persistent atrial fibrillation.

[0176] Figure 5a The figure shows a schematic diagram of the result output by the untrained spatio-temporal encoder according to an embodiment of the present invention. Figure 5b The figure shows a schematic diagram of the result output by the spatio-temporal encoder trained by the prior art according to an embodiment of the present invention. Figure 5c The figure shows a schematic diagram of the result output by the spatio-temporal encoder trained by the method of the present invention according to an embodiment of the present invention.

[0177] As Figures 5a to 5c shown, Figures 5a to 5cThe predicted results of the outputs of three spatio-temporal encoders are shown. The abscissa can be characterized as the true information, the ordinate can be characterized as the target auxiliary detection heartbeat information output by the information detection model, NAF can be characterized as non-latent atrial fibrillation, SAF can be characterized as latent atrial fibrillation. The upper left square in each 2×2 graph can be characterized as the number of objects where both the true information and the target auxiliary detection heartbeat information are non-latent atrial fibrillation. The upper right square can be characterized as the number of objects where the true information is latent atrial fibrillation while the target auxiliary detection heartbeat information is non-latent atrial fibrillation. The lower left square can be characterized as the number of objects where the true information is non-latent atrial fibrillation while the target auxiliary detection heartbeat information is latent atrial fibrillation. The lower right square can be characterized as the number of objects where both the true information and the target auxiliary detection heartbeat information are latent atrial fibrillation. It can be seen from the figure that the results output by the spatio-temporal encoder trained by the method of the present invention have the highest accuracy when used to obtain the target auxiliary detection heartbeat information and can distinguish normal heartbeats from heartbeats that may carry latent atrial fibrillation. The index information of the predicted results output by the spatio-temporal encoder in three training cases is shown in Table 1.

[0178] Table 1

[0179]

[0180] Among them, it can also be seen from the table that various indicators of the results output by the spatio-temporal encoder trained by the method of the present invention are good.

[0181] Figure 6a The schematic diagram of the results output by the untrained detection and evaluation sub-model according to the embodiment of the present invention is shown. Figure 6b The schematic diagram of the results output by the detection and evaluation sub-model trained by the prior art according to the embodiment of the present invention is shown. Figure 6c The schematic diagram of the results output by the detection and evaluation sub-model trained by the method of the present invention according to the embodiment of the present invention is shown.

[0182] As Figures 6a to 6c shown, Figures 6a to 6cThree predicted results of the output of the detection and evaluation sub-model are shown. The abscissa can be characterized as the true information, the ordinate can be characterized as the target auxiliary detection heartbeat information output by the information detection model, NAF can be characterized as non-latent atrial fibrillation, SAF can be characterized as latent atrial fibrillation. The upper left square in each 2×2 graph can represent the number of objects where both the true information and the target auxiliary detection heartbeat information are non-latent atrial fibrillation. The upper right square can represent the number of objects where the true information is latent atrial fibrillation while the target auxiliary detection heartbeat information is non-latent atrial fibrillation. The lower left square can represent the number of objects where the true information is non-latent atrial fibrillation while the target auxiliary detection heartbeat information is latent atrial fibrillation. The lower right square can represent the number of objects where both the true information and the target auxiliary detection heartbeat information are latent atrial fibrillation. It can be seen from the figure that the result output by the detection and evaluation sub-model trained by the method of the present invention has the highest accuracy and can distinguish normal heartbeats from heartbeats that may carry latent atrial fibrillation. The index information of the predicted results output by the detection and evaluation sub-model under three training conditions is shown in Table 2.

[0183] Table 2

[0184]

[0185] Among them, it can also be seen from the table that various indexes of the result output by the detection and evaluation sub-model trained by the method of the present invention are good.

[0186] Figure 7a It shows a schematic diagram of the excellent degree of each prediction index obtained after auxiliary detection using the method of the present invention according to an embodiment of the present invention. Figure 7b It shows a schematic diagram of the target auxiliary prediction heartbeat information obtained after auxiliary detection using the method of the present invention under the condition of increasing the monitoring time according to an embodiment of the present invention.

[0187] As Figures 7a to 7b shown, Figures 7a to 7b It shows a schematic diagram after auxiliary detection of multiple measured objects. Figure 7b The abscissa of which can be characterized as time, and the ordinate can be characterized as the result output by the method of the present invention. It can be seen from the figure that the performance of the information detection model used for auxiliary detection among the six measured objects is stable and excellent. Even if the output target auxiliary prediction heartbeat information is not accurate enough, the effect of auxiliary prediction can be improved through a longer time. Specifically, except for the measured object 5, the performance of the information detection model is stable and excellent, with an average accuracy rate of 99% and an average F1 score reaching 0.76. After performing auxiliary detection on the measured object 5 for a longer time, the success degree of the target auxiliary prediction heartbeat information output by the information detection model for the measured object 5 is also improving.

[0188] Figure 8A schematic diagram showing the comparison between the various indicators of the target auxiliary detection heartbeat information obtained by using the electrocardiogram of the prior art and the method of the present invention for auxiliary prediction respectively according to an embodiment of the present invention.

[0189] As Figure 8 shown, Figure 8 The various indicators of the target auxiliary detection heartbeat information obtained by using the electrocardiogram of the prior art and the method of the present invention for auxiliary detection respectively are shown. It can be seen from the figure that the target auxiliary prediction heartbeat information output by the method of the present invention is superior to the auxiliary information output based on the electrocardiogram in the four evaluation indicators of accuracy, sensitivity, specificity and F1 score. Among them, the method of the present invention has the largest improvement in the F1 score, with a 108% improvement compared to the electrocardiogram-based method.

[0190] Figure 9 A schematic diagram showing the process for auxiliary prediction of latent atrial fibrillation according to an embodiment of the present invention.

[0191] As Figure 9 shown, Figure 9 The process for auxiliary prediction of latent atrial fibrillation of the present invention is shown. In S901, an electromagnetic signal is transmitted to the object to be measured by a millimeter-wave radar. After the object to be measured receives the electromagnetic signal, an initial electromagnetic echo signal is generated. After signal separation processing of the initial electromagnetic echo signal, a target electromagnetic echo signal corresponding to the motion state of the heart region is obtained. In S902, the target electromagnetic echo signal is input into the spatio-temporal encoder of the information detection model after regression contrast learning training, and a plurality of mechanical motion spatio-temporal features related to normal and abnormal heartbeats in the heart region are output. Then, the plurality of mechanical motion spatio-temporal features are input into the feature extraction encoder for feature extraction processing to obtain the target mechanical motion time feature related to abnormal heartbeat. In S903, the target mechanical motion time feature related to abnormal heartbeat is input into the first double-layer fully connected network to obtain the auxiliary atrial beating information of the object to be measured. In S904, the target mechanical motion time feature related to abnormal heartbeat is input into the second double-layer fully connected network to obtain the auxiliary atrial fibrillation information of the object to be measured.

[0192] Figure 10 A structural block diagram showing the device for auxiliary prediction of latent atrial fibrillation according to an embodiment of the present invention.

[0193] As Figure 10 shown, the device for evaluating the ejection function value based on the semi-supervised depth segmentation model in this embodiment includes: an acquisition module 1010, a separation module 1020, a first extraction module 1030, a second extraction module 1040 and a obtaining module 1050.

[0194] An acquisition module 1010 is configured to acquire an initial electromagnetic echo signal, where the initial electromagnetic echo signal represents a plurality of signals that are spatially aliased with each other and reflected by multiple body regions of the object under test within a predetermined time period. The acquisition module 1010 can be used to perform the operation S210 described above, which will not be elaborated here.

[0195] A separation module 1020 is configured to perform signal separation processing on the initial electromagnetic echo signal to obtain a target electromagnetic echo signal corresponding to the motion state of the heart region of the object under test within a predetermined time period. The separation module 1020 can be used to perform the operation S220 described above, which will not be elaborated here.

[0196] A first extraction module 1030 is configured to perform first feature extraction processing on the target electromagnetic echo signal by using the spatio-temporal encoder of the information detection model to generate a plurality of mechanical motion spatio-temporal features related to normal and abnormal heartbeats of the heart region. The first extraction module 1030 can be used to perform the operation S230 described above, which will not be elaborated here.

[0197] A second extraction module 1040 is configured to perform second feature extraction processing on the plurality of mechanical motion spatio-temporal features by using the feature extraction encoder of the information detection model to generate a target mechanical motion time feature related to the abnormal heartbeat of the heart region. The second extraction module 1040 can be used to perform the operation S240 described above, which will not be elaborated here.

[0198] An obtaining module 1050 is configured to perform feature conversion mapping processing on the target mechanical motion time feature by using the detection and evaluation sub-model of the information detection model to obtain the target auxiliary prediction heartbeat information of the object under test. The obtaining module 1050 can be used to perform the operation S250 described above, which will not be elaborated here.

[0199] According to an embodiment of the present invention, the separation module 1020 includes: a first beamforming sub-module, a first extraction sub-module, and a first filtering sub-module.

[0200] The first beamforming sub-module is configured to perform beamforming processing on the initial electromagnetic echo signal to obtain a plurality of intermediate electromagnetic echo signals, where the intermediate electromagnetic echo signal represents a plurality of four-dimensional signals that are spatially separated from each other and reflected by multiple regions within a predetermined time period.

[0201] The first extraction sub-module is configured to extract a three-dimensional electromagnetic echo signal corresponding to the heart region of the object under test from the plurality of intermediate electromagnetic echo signals.

[0202] The first filtering sub-module is configured to perform filtering processing on the three-dimensional electromagnetic echo signal by using a second-order difference time-domain filtering algorithm to obtain the target electromagnetic echo signal.

[0203] According to an embodiment of the present invention, the first extraction sub-module includes: a first superposition unit, a first processing unit, and a second processing unit.

[0204] The first superposition unit is configured to perform superposition processing on a plurality of intermediate electromagnetic echo signals to obtain an intermediate electromagnetic echo superposition signal.

[0205] The first processing unit is configured to process the intermediate electromagnetic echo superposition signal by using a constant false alarm rate algorithm to obtain a distance electromagnetic echo signal corresponding to the thoracic region of the object to be measured, wherein the distance electromagnetic echo signal is a three-dimensional signal.

[0206] The second processing unit is configured to perform clustering processing on the distance electromagnetic echo signal by using a density clustering algorithm to obtain a three-dimensional electromagnetic echo signal corresponding to the cardiac region of the object to be measured.

[0207] According to an embodiment of the present invention, the first extraction module 1030 includes: a first splitting sub-module and a second extraction sub-module.

[0208] The first splitting sub-module is configured to split the target electromagnetic echo signal into target electromagnetic echo sub-signals of multiple time periods based on a predetermined coding time period.

[0209] The second extraction sub-module is configured to respectively input the target electromagnetic echo sub-signals of multiple time periods into the spatio-temporal encoder of the information detection model for first feature extraction processing to generate a plurality of mechanical motion spatio-temporal features related to normal heartbeats and abnormal heartbeats in the cardiac region, wherein the plurality of mechanical motion spatio-temporal features correspond to the target electromagnetic echo sub-signals of multiple time periods.

[0210] According to an embodiment of the present invention, the obtaining module 1050 includes: a first dimensionality reduction sub-module, a first conversion sub-module, and a second conversion sub-module.

[0211] The first dimensionality reduction sub-module is configured to respectively input the target mechanical motion time features into the first linear layer of the first double-layer fully connected network and the first linear layer of the second double-layer fully connected network for dimensionality reduction processing to obtain a first dimensionality reduction mechanical motion state feature and a second dimensionality reduction mechanical motion time feature.

[0212] The first conversion sub-module is configured to input the first dimensionality reduction mechanical motion state feature into the second linear layer of the first double-layer fully connected network for feature conversion mapping processing to obtain auxiliary atrial beating information of the object to be measured.

[0213] The second conversion sub-module is configured to input the second dimensionality reduction mechanical motion time feature into the second linear layer of the second double-layer fully connected network for feature conversion mapping processing to obtain auxiliary atrial fibrillation information of the object to be measured.

[0214] According to an embodiment of the present invention, the device for assisting in predicting latent atrial fibrillation further includes: a training module.

[0215] The training module is used to train an information detection model.

[0216] According to an embodiment of the present invention, the training module includes: a first acquisition sub-module, a second splitting sub-module, a third extraction sub-module, a fourth extraction sub-module, a first mapping sub-module, a first obtaining sub-module, and a first adjustment sub-module.

[0217] The first acquisition sub-module is used to acquire an initial model to be trained and a sample training data set. Among them, the initial model to be trained includes an initial spatio-temporal encoder, a feature extraction encoder to be trained, and a detection and evaluation sub-model to be trained. The sample training data set includes electromagnetic echo signal samples, latent atrial fibrillation disease period information samples, atrial beating information samples, and atrial fibrillation information samples.

[0218] The second splitting sub-module is used to split the electromagnetic echo signal samples into multiple electromagnetic echo signal sub-samples based on predetermined sample coding period information.

[0219] The third extraction sub-module is used to input the multiple electromagnetic echo signal sub-samples into the initial spatio-temporal encoder of the initial model to be trained for the first feature extraction process, and generate multiple training mechanical motion spatio-temporal features related to normal and abnormal heartbeats in the heart region.

[0220] The fourth extraction sub-module is used to input the multiple mechanical motion spatio-temporal training features into the feature extraction encoder to be trained for the second feature extraction process, and generate training mechanical motion time features related to abnormal heartbeats in the heart region.

[0221] The first mapping sub-module is used to input the training mechanical motion time features into the detection and evaluation sub-model to be trained for feature transformation mapping processing, and obtain training detection heartbeat information.

[0222] The first obtaining sub-module is used to obtain a detection and evaluation loss value based on a detection and evaluation loss function, according to the training detection heartbeat information, latent atrial fibrillation disease period information samples, atrial beating information samples, and atrial fibrillation information samples.

[0223] The first adjustment sub-module is used to adjust the model parameters of the initial model according to the detection and evaluation loss value, and obtain a trained information detection model.

[0224] According to an embodiment of the present invention, the training module further includes: a second acquisition sub-module, a fifth extraction sub-module, a second obtaining sub-module, a first regression and comparison sub-module, and a second adjustment sub-module.

[0225] The second acquisition sub-module is used to acquire a basic spatio-temporal encoder to be trained.

[0226] The fifth extraction sub-module is configured to input multiple sub-samples of electromagnetic echo signals into a to-be-trained basic spatio-temporal encoder for first feature extraction processing, and generate multiple basic training mechanical motion spatio-temporal features related to normal and abnormal heartbeats in the heart region.

[0227] The second obtaining sub-module is configured to obtain a hidden atrial fibrillation period comparison weight based on a preset hidden atrial fibrillation period comparison weight function according to the hidden atrial fibrillation disease period information sample.

[0228] The first regression comparison sub-module is configured to perform regression comparison learning processing on the hidden atrial fibrillation period comparison weight and multiple basic training mechanical motion spatio-temporal features based on a preset spatio-temporal loss function to obtain a spatio-temporal loss value.

[0229] The second adjustment sub-module is configured to adjust the hidden atrial fibrillation period comparison weight of the to-be-trained basic spatio-temporal encoder according to the spatio-temporal loss value to obtain a trained initial spatio-temporal encoder.

[0230] According to an embodiment of the present invention, the first obtaining sub-module includes: a first obtaining unit, a second obtaining unit, a first determining unit, a second determining unit, a third obtaining unit, a fourth obtaining unit, and a fifth obtaining unit.

[0231] The first obtaining unit is configured to obtain a cross-entropy loss value corresponding to the t-th iteration round according to the training atrial beat information and the atrial beat information sample in the t-th iteration round, where t≥1 and t is a positive integer.

[0232] The second obtaining unit is configured to obtain a mean square error loss value corresponding to the t-th iteration round according to the multiple training atrial fibrillation information and the atrial fibrillation information sample in the t-th iteration round.

[0233] The first determining unit is configured to determine an atrial beat loss change rate corresponding to the t-th iteration round according to the average cross-entropy loss corresponding to the t-th iteration round and the average cross-entropy loss corresponding to the (t - 1)-th iteration round.

[0234] The second determining unit is configured to determine an atrial fibrillation loss change rate corresponding to the t-th iteration round according to the average mean square error loss corresponding to the t-th iteration round and the average mean square error loss corresponding to the (t - 1)-th iteration round.

[0235] The third obtaining unit is configured to obtain a training loss change rate corresponding to the t-th iteration round according to the atrial beat loss change rate corresponding to the t-th iteration round and the atrial fibrillation loss change rate corresponding to the t-th iteration round.

[0236] A fourth obtaining unit, configured to obtain, based on a preset beat loss change rate function and a preset fibrillation loss change rate function, an atrial beat loss weighting factor and an atrial fibrillation loss weighting factor for the t-th iteration round according to the training loss change rate corresponding to the t-th iteration round, the atrial beat loss change rate corresponding to the t-th iteration round, and the atrial fibrillation loss change rate corresponding to the t-th iteration round.

[0237] A fifth obtaining unit, configured to obtain, based on a detection and evaluation function, a detection and evaluation loss value for the t-th iteration round according to the cross-entropy loss value corresponding to the t-th iteration round, the mean square error loss value corresponding to the t-th iteration round, the atrial beat loss weighting factor for the t-th iteration round, and the atrial fibrillation loss weighting factor for the t-th iteration round.

[0238] According to an embodiment of the present invention, any plurality of the obtaining module 1010, the separating module 1020, the first extraction module 1030, the second extraction module 1040, and the obtaining module 1050 may be combined and implemented in one module, or any one of them may be split into multiple modules. Alternatively, at least part of the functions of one or more of these modules may be combined with at least part of the functions of other modules and implemented in one module. According to an embodiment of the present invention, at least one of the obtaining module 1010, the separating module 1020, the first extraction module 1030, the second extraction module 1040, and the obtaining module 1050 may be at least partially implemented as a hardware circuit, such as a field programmable gate array (FPGA), a programmable logic array (PLA), a system on chip, a system on substrate, a system on package, an application specific integrated circuit (ASIC), or any other reasonable way of integrating or packaging circuits, etc., implemented by hardware or firmware, or implemented in any one of the three implementation manners of software, hardware, and firmware, or in an appropriate combination of any several of them. Alternatively, at least one of the obtaining module 1010, the separating module 1020, the first extraction module 1030, the second extraction module 1040, and the obtaining module 1050 may be at least partially implemented as a computer program module, which can execute the corresponding functions when the computer program module is run.

[0239] Figure 11 A block diagram of an electronic device for a method of assisting in predicting latent atrial fibrillation according to an embodiment of the present invention is shown.

[0240] As Figure 11As shown, the electronic device according to an embodiment of the present invention includes a processor 1101, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 1102 or a program loaded from a storage section 1108 into a random access memory (RAM) 1103. The processor 1101 may include, for example, a general-purpose microprocessor (such as a CPU), an instruction set processor, and / or a related chipset, and / or a dedicated microprocessor (such as an application-specific integrated circuit (ASIC)), etc. The processor 1101 may also include on-board memory for caching purposes. The processor 1101 may include a single processing unit or multiple processing units for performing different actions of the method flow according to an embodiment of the present invention.

[0241] In the RAM 1103, various programs and data required for the operation of the electronic device are stored. The processor 1101, the ROM 1102, and the RAM 1103 are connected to each other via a bus 1104. The processor 1101 performs various operations of the method flow according to an embodiment of the present invention by executing a program in the ROM 1102 and / or the RAM 1103. It should be noted that the program may also be stored in one or more memories other than the ROM 1102 and the RAM 1103. The processor 1101 may also perform various operations of the method flow according to an embodiment of the present invention by executing a program stored in the one or more memories.

[0242] According to an embodiment of the present invention, the electronic device may further include an input / output (I / O) interface 1105, and the input / output (I / O) interface 1105 is also connected to the bus 1104. The electronic device may further include one or more of the following components connected to the I / O interface 1105: an input section 1106 including a keyboard, a mouse, etc.; an output section 1107 including, for example, a cathode ray tube (CRT), a liquid crystal display (LCD), etc., and a speaker, etc.; a storage section 1108 including a hard disk, etc.; and a communication section 1109 including a network interface card such as a LAN card, a modem, etc. The communication section 1109 performs communication processing via a network such as the Internet. A driver 1110 is also connected to the I / O interface 1105 as needed. A removable medium 1111, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the driver 1110 as needed so that a computer program read from it can be installed into the storage section 1108 as needed.

[0243] The present invention also provides a computer-readable storage medium, which may be included in the device / apparatus / system described in the above embodiments; or may exist alone without being assembled into the device / apparatus / system. The above computer-readable storage medium carries one or more programs, and when the one or more programs are executed, the method according to the embodiments of the present invention is implemented.

[0244] According to an embodiment of the present invention, the computer-readable storage medium may be a non-volatile computer-readable storage medium, and may include, for example, but not limited to: portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the above. In the present invention, the computer-readable storage medium may be any tangible medium that contains or stores a program, and the program may be used by or in conjunction with an instruction execution system, apparatus, or device. For example, according to an embodiment of the present invention, the computer-readable storage medium may include the above-described ROM 1102 and / or RAM 1103 and / or one or more memories other than ROM 1102 and RAM 1103.

[0245] An embodiment of the present invention also includes a computer program product, which includes a computer program that contains program code for executing the method shown in the flowchart. When the computer program product runs in a computer system, the program code is used to cause the computer system to implement the method for assisting in predicting latent atrial fibrillation provided by the embodiments of the present invention.

[0246] When the computer program is executed by the processor 1101, the above functions defined in the system / apparatus of the embodiments of the present invention are executed. According to an embodiment of the present invention, the above-described systems, apparatuses, modules, units, etc. may be implemented by computer program modules.

[0247] In one embodiment, the computer program may rely on tangible storage media such as optical storage devices and magnetic storage devices. In another embodiment, the computer program may also be transmitted and distributed in the form of a signal on a network medium, and be downloaded and installed through the communication part 1109, and / or be installed from the removable medium 1111. The program code included in the computer program may be transmitted by any suitable network medium, including but not limited to: wireless, wired, etc., or any suitable combination of the above.

[0248] In such an embodiment, the computer program can be downloaded and installed from a network through the communication section 1109, and / or installed from the removable medium 1111. When the computer program is executed by the processor 1101, the above-described functions defined in the system of the embodiment of the present invention are performed. According to an embodiment of the present invention, the above-described system, device, apparatus, module, unit, etc. can be implemented by computer program modules.

[0249] Those skilled in the art can understand that the features described in the various embodiments of the present invention can be combined or combined in various ways, even if such combinations or combinations are not explicitly described in the present invention. In particular, without departing from the spirit and teachings of the present invention, the features described in the various embodiments of the present invention can be combined and / or combined in various ways. All such combinations and / or combinations fall within the scope of the present invention.

Claims

1. A method for assisting in predicting latent atrial fibrillation, characterized in that, Including: Obtaining an initial electromagnetic echo signal, where the initial electromagnetic echo signal represents multiple signals that are spatially aliased with each other and reflected by multiple body regions of the object under test within a predetermined time period; Performing signal separation processing on the initial electromagnetic echo signal to obtain a target electromagnetic echo signal corresponding to the motion state of the heart region of the object under test within the predetermined time period; Performing first feature extraction processing on the target electromagnetic echo signal using the spatio-temporal encoder of the information detection model to generate multiple mechanical motion spatio-temporal features related to normal heartbeats and abnormal heartbeats of the heart region; Performing second feature extraction processing on the multiple mechanical motion spatio-temporal features using the feature extraction encoder of the information detection model to generate target mechanical motion time features related to abnormal heartbeats of the heart region; Performing feature transformation mapping processing on the target mechanical motion time features using the detection and evaluation sub-model of the information detection model to obtain target auxiliary prediction heartbeat information of the object under test, where the detection and evaluation sub-model includes a first double-layer fully connected network and a second double-layer fully connected network, both the first double-layer fully connected network and the second double-layer fully connected network include two linear layers, inputting the target mechanical motion time features into the first linear layer of the first double-layer fully connected network and the second double-layer fully connected network respectively for dimensionality reduction processing to obtain a first dimensionality-reduced mechanical motion state feature and a second dimensionality-reduced mechanical motion time feature; inputting the first dimensionality-reduced mechanical motion state feature into the second linear layer of the first double-layer fully connected network and inputting the second dimensionality-reduced mechanical motion time feature into the second linear layer of the second double-layer fully connected network for feature transformation mapping processing to obtain auxiliary atrial beating information and auxiliary atrial fibrillation information.

2. The method according to claim 1, wherein The performing signal separation processing on the initial electromagnetic echo signal to obtain a target electromagnetic echo signal corresponding to the motion state of the heart region of the object under test within the predetermined time period includes: Performing beamforming processing on the initial electromagnetic echo signal to obtain multiple intermediate electromagnetic echo signals, where the intermediate electromagnetic echo signals represent multiple four-dimensional signals that are spatially separated from each other and reflected by multiple regions within a predetermined time period; Extracting a three-dimensional electromagnetic echo signal corresponding to the heart region of the object under test from the multiple intermediate electromagnetic echo signals; Performing filtering processing on the three-dimensional electromagnetic echo signal using a second-order difference time-domain filtering algorithm to obtain the target electromagnetic echo signal.

3. The method according to claim 2, characterized in that, The extracting a three-dimensional electromagnetic echo signal corresponding to the heart region of the object under test from the multiple intermediate electromagnetic echo signals includes: Performing superposition processing on the multiple intermediate electromagnetic echo signals to obtain an intermediate electromagnetic echo superposition signal; Performing processing on the intermediate electromagnetic echo superposition signal using a constant false alarm rate algorithm to obtain a distance electromagnetic echo signal corresponding to the thoracic region of the object under test, where the distance electromagnetic echo signal is a three-dimensional signal; The distance electromagnetic echo signal is clustered using a density clustering algorithm to obtain the three-dimensional electromagnetic echo signal corresponding to the cardiac region of the object under test.

4. The method according to claim 1, wherein The target electromagnetic echo signal is subjected to a first feature extraction process using the spatio-temporal encoder of the information detection model to generate a plurality of mechanical motion spatio-temporal features related to normal and abnormal heartbeats in the cardiac region, including: Based on a predetermined coding period, the target electromagnetic echo signal is split into target electromagnetic echo sub-signals of multiple periods; The target electromagnetic echo sub-signals of the multiple periods are respectively input into the spatio-temporal encoder of the information detection model for a first feature extraction process to generate a plurality of mechanical motion spatio-temporal features related to normal and abnormal heartbeats in the cardiac region, wherein the plurality of mechanical motion spatio-temporal features correspond to the target electromagnetic echo sub-signals of the multiple periods.

5. The method according to claim 1, characterized in that The information detection model is trained using the following methods, including: An initial model to be trained and a sample training data set are obtained, wherein the initial model to be trained includes an initial spatio-temporal encoder, a feature extraction encoder to be trained, and a detection and evaluation sub-model to be trained, and the sample training data set includes electromagnetic echo signal samples, information samples of recessive atrial fibrillation disease periods, atrial beating information samples, and atrial fibrillation information samples; Based on the predetermined sample coding period information, the electromagnetic echo signal samples are split into a plurality of electromagnetic echo signal sub-samples; The plurality of electromagnetic echo signal sub-samples are input into the initial spatio-temporal encoder of the initial model to be trained for a first feature extraction process to generate a plurality of training mechanical motion spatio-temporal features related to normal and abnormal heartbeats in the cardiac region; The plurality of training mechanical motion spatio-temporal features are input into the feature extraction encoder to be trained for a second feature extraction process to generate training mechanical motion time features related to abnormal heartbeats in the cardiac region; The training mechanical motion time features are input into the detection and evaluation sub-model to be trained for a feature conversion mapping process to obtain training detection heartbeat information; Based on a detection and evaluation loss function, according to the training detection heartbeat information, the information samples of recessive atrial fibrillation disease periods, the atrial beating information samples, and the atrial fibrillation information samples, a detection and evaluation loss value is obtained; According to the detection and evaluation loss value, the model parameters of the initial model are adjusted to obtain a trained information detection model.

6. The method according to claim 5, characterized in that, The initial spatio-temporal encoder is pre-trained, and the pre-training method includes the following operations: A basic spatio-temporal encoder to be trained is obtained; The plurality of electromagnetic echo signal sub-samples are input into the basic spatio-temporal encoder to be trained for a first feature extraction process to generate a plurality of basic training mechanical motion spatio-temporal features related to normal and abnormal heartbeats in the cardiac region; Based on a preset recessive atrial fibrillation period comparison weight function, according to the information samples of recessive atrial fibrillation disease periods, a recessive atrial fibrillation period comparison weight is obtained; Based on a preset spatio-temporal loss function, perform regression contrast learning processing on the concealed atrial fibrillation period comparison weight and the multiple basic training mechanical motion spatio-temporal features to obtain a spatio-temporal loss value; According to the spatio-temporal loss value, adjust the concealed atrial fibrillation period comparison weight of the to-be-trained basic spatio-temporal encoder to obtain the trained initial spatio-temporal encoder.

7. The method according to claim 5, characterized in that, The training detection heartbeat information includes training atrial beating information and training atrial fibrillation information; The operation of obtaining a detection evaluation loss value based on the detection evaluation loss function according to the training detection heartbeat information, the concealed atrial fibrillation disease period information sample, the atrial beating information sample, and the atrial fibrillation information sample includes multiple iterative rounds, and the operation of the t-th iterative round includes: In the t-th iterative round, according to the training atrial beating information and the atrial beating information sample of the t-th iterative round, obtain a cross-entropy loss value corresponding to the t-th iterative round, where t≥1 and t is a positive integer; According to the multiple training atrial fibrillation information and the atrial fibrillation information sample of the t-th iterative round, obtain a mean square error loss value corresponding to the t-th iterative round; According to the average cross-entropy loss corresponding to the t iterative rounds and the average cross-entropy loss corresponding to the t - 1 iterative rounds, determine the atrial beating loss change rate corresponding to the t-th iterative round; According to the average mean square error loss corresponding to the t iterative rounds and the average mean square error loss corresponding to the t - 1 iterative rounds, determine the atrial fibrillation loss change rate corresponding to the t-th iterative round; According to the atrial beating loss change rate corresponding to the t-th iterative round and the atrial fibrillation loss change rate corresponding to the t-th iterative round, obtain the training loss change rate corresponding to the t-th iterative round; Based on a preset beating loss change rate function and a preset fibrillation loss change rate function, according to the training loss change rate corresponding to the t-th iterative round, the atrial beating loss change rate corresponding to the t-th iterative round, and the atrial fibrillation loss change rate corresponding to the t-th iterative round, obtain the atrial beating loss weighting factor and the atrial fibrillation loss weighting factor of the t-th iterative round; Based on the detection evaluation loss function, according to the cross-entropy loss value corresponding to the t-th iterative round, the mean square error loss value corresponding to the t-th iterative round, the atrial beating loss weighting factor of the t-th iterative round, and the atrial fibrillation loss weighting factor of the t-th iterative round, obtain the detection evaluation loss value of the t-th iterative round.

8. A device for assisting in predicting latent atrial fibrillation, characterized in that, Including: An acquisition module, configured to acquire an initial electromagnetic echo signal, where the initial electromagnetic echo signal represents multiple signals that are spatially aliased with each other and are reflected by multiple body regions of the object to be measured within a preset time period; A separation module, configured to perform signal separation processing on the initial electromagnetic echo signal to obtain a target electromagnetic echo signal corresponding to the motion state of the heart region of the object to be measured within the preset time period; The first extraction module is configured to perform first feature extraction processing on the target electromagnetic echo signal by using the spatio-temporal encoder of the information detection model, and generate a plurality of mechanical motion spatio-temporal features related to normal heartbeats and abnormal heartbeats in the heart region; The second extraction module is configured to perform second feature extraction processing on the plurality of mechanical motion spatio-temporal features by using the feature extraction encoder of the information detection model, and generate target mechanical motion time features related to abnormal heartbeats in the heart region; The obtaining module is configured to perform feature conversion mapping processing on the target mechanical motion time features by using the detection and evaluation sub-model of the information detection model, and obtain target auxiliary prediction heartbeat information of the object under test. The detection and evaluation sub-model includes a first double-layer fully connected network and a second double-layer fully connected network. Both the first double-layer fully connected network and the second double-layer fully connected network include two linear layers. The target mechanical motion time features are respectively input into the first linear layers of the first double-layer fully connected network and the second double-layer fully connected network for dimensionality reduction processing, to obtain a first dimensionality-reduced mechanical motion state feature and a second dimensionality-reduced mechanical motion time feature; the first dimensionality-reduced mechanical motion state feature is input into the second linear layer of the first double-layer fully connected network, and the second dimensionality-reduced mechanical motion time feature is input into the second linear layer of the second double-layer fully connected network for feature conversion mapping processing, to obtain auxiliary atrial beating information and auxiliary atrial fibrillation information.

9. An electronic device, characterized in that, Comprising: One or more processors; A storage device for storing one or more programs, wherein, when the one or more programs are executed by the one or more processors, the one or more processors are caused to execute the method according to any one of claims 1 to 7.

Images