Human body sleeping posture prediction equipment

Abstract

The embodiment of the utility model provides human body sleeping posture prediction equipment, which comprises a radar data acquisition module, a data transmission module, a data preprocessing module and a sleeping posture prediction module, and is characterized in that the radar data acquisition module is used for acquiring human body sleeping data of a target human body; the data transmission module is used for transmitting the collected human body sleep data to the data preprocessing module; the data preprocessing module is used for extracting sleep signal features of a target human body from the human body sleep data and transmitting the sleep signal features to the sleep posture prediction module through the data transmission module; and the sleeping posture prediction module is used for inputting the sleeping signal features into a pre-trained human body sleeping posture prediction model to obtain a predicted sleeping posture corresponding to the target human body. According to the technical scheme, the body movement of the patient in the sleep period can be accurately tracked under multiple different positions and environment conditions, and therefore the sleep postures of the target human body can be accurately classified.

Description

Technical Field

[0001] This manual relates to the field of medical device technology, and in particular to a human sleeping posture prediction device. Background Technology

[0002] With the development of IoT technology, more and more home sleep monitoring devices have emerged. These devices are used to identify the body position of people with sleep disorders and monitor their movements.

[0003] In related technical solutions, wearable and non-wearable devices are used to track the body movements of patients with sleep disorders during sleep. However, in these solutions, wearable devices are limited by their wearing position, and the cameras of non-wearable devices are easily affected by different lighting conditions, making it difficult to accurately track the patient's body movements during sleep.

[0004] Therefore, accurately tracking patients' body movements during sleep under various locations and environmental conditions has become a pressing technical challenge.

[0005] The information in the background section is merely information known only to the inventor and does not imply that such information had entered the public domain before the date of this application, nor does it imply that it can be considered prior art in this disclosure. Utility Model Content

[0006] This manual provides a human sleep posture prediction device that can accurately track a patient's body movements during sleep under multiple different locations and environmental conditions, thereby accurately classifying the sleep posture of the target human body.

[0007] In a first aspect, this specification provides a human sleeping posture prediction device, which includes a radar data acquisition module, a data transmission module, a data preprocessing module, and a sleeping posture prediction module, wherein:

[0008] The radar data acquisition module includes a radar module, which is used to collect human sleep data of the target human body;

[0009] The data transmission module is communicatively connected to the radar data acquisition module and the data preprocessing module, and is used to transmit the human sleep data of the target human body acquired by the radar module to the data preprocessing module.

[0010] A data preprocessing module, communicatively connected to the sleep position prediction module, is used to extract sleep signal features from the sleep data of the target human body, and transmit the sleep signal features to the sleep position prediction module via the data transmission module; and

[0011] The sleeping posture prediction module is communicatively connected to the data transmission module and is used to input the sleep signal features into a pre-trained human sleeping posture prediction model to obtain the predicted sleeping posture corresponding to the target human body.

[0012] In some example embodiments, based on the above scheme, the human sleeping posture prediction device further includes:

[0013] The display module is communicatively connected to the data transmission module and is used to provide data display and radar parameter setting functions through a human-computer interaction interface. The data display includes displaying the human sleep data of the target human body collected by the radar module and the predicted sleeping posture of the target human body.

[0014] In some example embodiments, based on the above scheme, the data preprocessing module includes:

[0015] The range selection module is used to crop the human sleep data of the target human body based on a predetermined range to obtain the human sleep data of the target human body within the predetermined range.

[0016] In some example embodiments, based on the above scheme, the data preprocessing module further includes:

[0017] The noise reduction module is used to filter out static noise signals in the human sleep data of the target human body within the predetermined range, so as to obtain the noise-reduced human sleep data corresponding to the target human body.

[0018] In some example embodiments, based on the above scheme, the data preprocessing module further includes:

[0019] The feature extraction module is used to extract frequency domain features and time domain features from the noise-reduced human sleep data to obtain the sleep signal features of the target human body.

[0020] In some example embodiments, based on the above scheme, the human sleeping posture prediction model includes a feature convolutional layer and a causal dilated convolutional layer, and the sleeping posture prediction module includes:

[0021] The feature convolution module is used to input the sleep signal features into the feature convolution layer to obtain the sleeping posture signal features corresponding to the target human body; and to input the sleeping posture signal features into the causal dilation convolution layer to obtain multi-scale sleeping posture features.

[0022] The prediction module is used to predict the sleeping posture of the target human body based on the multi-scale sleeping posture features.

[0023] In some example embodiments, based on the above scheme, the causal dilated convolutional layer includes a dilated convolutional layer and a causal convolutional layer. The dilated convolutional layer is used to obtain sleeping posture features at multiple time scales from the sleeping posture signal features, and the causal convolutional layer is used to introduce causality in the convolution operation of the dilated convolutional layer, so that the sleeping posture features obtained by the dilated convolutional layer depend only on historical input data.

[0024] In some example embodiments, based on the above scheme, the human sleeping posture prediction device further includes:

[0025] The model training module is used to determine the cross-entropy loss between the predicted sleeping position and the labeled sleeping position; based on the cross-entropy loss, it determines the gradient of the corresponding model parameters of the human sleeping position prediction model, and adjusts the model parameters of the human sleeping position prediction model through gradient backpropagation.

[0026] In some example embodiments, based on the above scheme, the model training module is further used for:

[0027] Determine the learning rate corresponding to the sample dataset;

[0028] Based on the learning rate corresponding to the sample dataset, the model parameters of the human sleeping posture prediction model are updated.

[0029] In some example embodiments, based on the above scheme, the human sleeping posture prediction model is a deep neural network model, and the radar module is a millimeter-wave radar.

[0030] As can be seen from the above technical solutions, the human sleep posture prediction device provided in the embodiments of this specification, on the one hand, collects human sleep data of the target human body through a radar data acquisition module, and extracts the sleep signal features of the target human body from the human sleep data through a data preprocessing module. This not only enables high-precision and non-invasive acquisition of the sleep signal features of the target human body, but also accurately tracks the body movement of the target human body during sleep under multiple different locations and environmental conditions. On the other hand, the sleep signal features are input into the human sleep posture prediction model through a sleep posture prediction module to obtain the predicted sleep posture of the target human body, thereby accurately classifying the sleep posture of the target human body. This has a potentially broad application prospect for improving sleep quality monitoring and health management.

[0031] Other functions of the human sleeping position prediction device provided in this specification will be partially listed in the following description. The figures and examples described below will be readily apparent to those skilled in the art. The inventive aspects of the human sleeping position prediction device provided in this specification can be fully understood through practice or use of the methods, apparatus, and combinations described in the detailed examples below. Attached Figure Description

[0032] To more clearly illustrate the technical solutions in the embodiments of this specification, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this specification. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0033] Figure 1 A schematic diagram of the implementation environment provided in the embodiments of this specification is shown;

[0034] Figure 2 A hardware structure diagram of a human sleeping posture prediction device 200 provided according to an embodiment of this specification is shown;

[0035] Figure 3 A schematic diagram of the structure of a human sleeping posture prediction device provided according to some embodiments of this specification is shown;

[0036] Figure 4 A schematic diagram of the model structure of a human sleeping posture prediction model provided according to some embodiments of this specification is shown;

[0037] Figure 5 A schematic diagram of the structure of a sleeping posture prediction module provided according to some other embodiments of this specification is shown;

[0038] Figure 6 A schematic diagram of the structure of a data preprocessing module provided according to some embodiments of this specification is shown;

[0039] Figure 7 A schematic diagram of the structure of a human sleeping posture prediction device according to some other embodiments of this specification is shown; and

[0040] Figure 8 A schematic diagram of the structure of a human sleeping posture prediction system provided according to some embodiments of this specification is shown. Detailed Implementation

[0041] The following description provides specific application scenarios and requirements for this specification, intended to enable those skilled in the art to make and use the contents of this specification. Various partial modifications to the disclosed embodiments will be apparent to those skilled in the art, and the general principles defined herein can be applied to other embodiments and applications without departing from the spirit and scope of this specification. Therefore, this specification is not limited to the embodiments shown, but rather to the widest scope consistent with the claims.

[0042] The terminology used herein is for the purpose of describing particular exemplary embodiments only and is not restrictive. For example, unless the context clearly indicates otherwise, the singular forms “a,” “an,” and “the” used herein may also include the plural forms. When used in this specification, the terms “comprising,” “including,” and / or “containing” mean that the associated integers, steps, operations, elements, and / or components are present, but do not exclude the presence of one or more other features, integers, steps, operations, elements, components, and / or groups, or that other features, integers, steps, operations, elements, components, and / or groups may be added to the system / method.

[0043] Considering the following description, these and other features of this specification, as well as the operation and function of the related components of the structure, and the economy of assembly and manufacture of the parts, can be significantly improved. All of these form part of this specification with reference to the accompanying drawings. However, it should be clearly understood that the drawings are for illustrative and descriptive purposes only and are not intended to limit the scope of this specification. It should also be understood that the drawings are not drawn to scale.

[0044] The flowcharts used in this specification illustrate operations implemented according to some embodiments of this specification. It should be clearly understood that the operations in the flowcharts may not be implemented in a sequential order. Instead, the operations may be implemented in reverse order or simultaneously. Furthermore, one or more additional operations may be added to the flowcharts. One or more operations may be removed from the flowcharts.

[0045] In related technical solutions, radar-based sleep monitoring systems can be used to detect vital signs such as heart rate and respiratory rate, and can also be used to identify breathing disorders and body movements at different sleep stages. For example, continuous wave Doppler radar and deep convolutional neural networks are used to accurately detect human movement by utilizing the Doppler effect of reflected signals. However, how to detect human sleep movements using radar has not yet been studied.

[0046] Based on the above, this specification provides a human sleep posture prediction device. On the one hand, it collects human sleep data of a target human body through a radar data acquisition module, and extracts sleep signal features of the target human body from the human sleep data through a data preprocessing module. This not only enables high-precision and non-invasive acquisition of the sleep signal features of the target human body, but also accurately tracks the body movements of the target human body during sleep under multiple different locations and environmental conditions. On the other hand, it inputs the sleep signal features into a human sleep posture prediction model through a sleep posture prediction module to obtain the predicted sleep posture of the target human body, thereby accurately classifying the sleep posture of the target human body. This has the potential for broad application prospects in improving sleep quality monitoring and health management.

[0047] The technical solutions of the embodiments of this specification will now be described in detail with reference to the accompanying drawings.

[0048] Figure 1 This is a schematic diagram illustrating application scenarios according to some embodiments of this specification. The monitoring device 100 includes radar, such as millimeter-wave radar, which transmits detection signals to detect human sleep signals of a target human body 140, such as position, movement, and posture characteristics. For example, the radar of the monitoring device 100 transmits detection signals towards a target human body in a sleeping state and receives echo signals reflected back from the target human body, the echo signals containing human sleep signals. When operating, the monitoring device 100 can transmit detection signals to different azimuth angles within its field of view and receive echo signals reflected back from the target human body at the corresponding azimuth angle, thereby obtaining the human sleep signal of the human body's reflecting portion within that azimuth angle.

[0049] like Figure 1 As shown, the monitoring device 100 may include a transmitting device 110, a receiving device 120, and a processing device 130. When operating, the transmitting device 110 can emit detection signals outward according to a preset timing sequence. The transmitting device 110 can be a radar, such as a millimeter-wave radar. The receiving device 120 can receive the echo signal reflected from the target human body, perform photoelectric conversion, and output an echo electrical signal to acquire the target human body's sleep signal. The processing device 130 can control the transmitting device 110 and the receiving device 120, and process the detection data to obtain the target human body's motion information, such as the distance and orientation of the reflecting part.

[0050] After describing the implementation environment of the embodiments in this specification, the application scenarios of the embodiments will be introduced below in conjunction with the above implementation environment. The technical solutions provided by the embodiments in this specification can be applied in various sleep monitoring scenarios, such as sleep monitoring scenarios in hospitals or home environments.

[0051] Taking the technical solution provided in the embodiments of this specification as an example in the context of sleep monitoring in a home environment, the monitoring device 100 acquires human sleep data of the target human body collected by radar; extracts sleep signal features of the target human body from the human sleep data of the target human body; and inputs the sleep signal features into a pre-trained human sleeping posture prediction model to obtain the predicted sleeping posture of the target human body, such as supine, prone, or lateral.

[0052] It should be noted that the above description is based on the application of the technical solution provided in the embodiments of this specification to a sleep monitoring scenario in a home environment. The technical solution provided in the embodiments of this specification can also be applied to other suitable sleep monitoring scenarios, such as sleep monitoring scenarios in hospitals or nursing homes. The implementation process is the same as the above description and belongs to the same inventive concept, so it will not be repeated here.

[0053] It should be noted that the steps in the example embodiments of this specification may be partially performed by the client, such as the monitoring device 100, partially performed by the server, or entirely performed by the server or entirely by the client. This specification does not impose any special limitations on this.

[0054] based on Figure 1 The implementation environment shown below will be combined with... Figures 2-8 This specification provides a detailed description of the human sleeping posture prediction device provided in the embodiments. It should be noted that the above-described implementation environment is only shown to facilitate understanding of the spirit and principles of this specification, and the embodiments of this specification are not limited in any way. Rather, the embodiments of this specification can be applied to any applicable scenario.

[0055] Figure 2 This is a schematic diagram of a human sleeping posture prediction device 200 according to some embodiments of this specification. The aforementioned human sleeping posture prediction device 200 can execute the human sleeping posture prediction method and model training method described in this specification. The aforementioned human sleeping posture prediction method and model training method are described in other parts of this specification. The aforementioned human sleeping posture prediction device 200 may include a transmitting device and a receiving device. When the transmitting device is working, it can emit detection signals outward according to a preset timing sequence. The transmitting device 200 can be a radar, such as a millimeter-wave radar. The receiving device 400 can receive the echo signal reflected from the target human body, perform photoelectric conversion, and output an echo electrical signal for acquiring the human sleep signal of the target human body. The aforementioned human sleeping posture prediction device can be... Figure 1 The monitoring device 100 can also be a terminal device used by multiple developers to develop programs on an integrated development platform.

[0056] The human sleeping posture prediction device described in this specification may include one or more of the following components: processor 210, memory 220, input device 230, output device 240, and bus 250. The processor 210, memory 220, input device 230, and output device 240 may be connected to each other via bus 250.

[0057] Processor 210 may include one or more processing cores. Processor 210 connects to various parts within the entire human sleeping posture prediction device using various interfaces and lines. It executes the human sleeping posture prediction method and model training method described in this specification by running or executing instructions, programs, code sets, or instruction sets stored in memory 220, and by calling data stored in memory 220. Optionally, processor 210 may be implemented using at least one hardware form of digital signal processing (DSP), field-programmable gate array (FPGA), or programmable logic array (PLA). Processor 210 may integrate one or more of a central processing unit (CPU), a graphics processing unit (GPU), and a modem. The CPU primarily handles the operating system, user interface, and applications; the GPU is responsible for rendering and drawing the displayed content; and the modem handles wireless communication. It is understood that the modem may also not be integrated into processor 210 and may be implemented separately through a communication chip.

[0058] The memory 220 may include random access memory (RAM) or read-only memory (ROM). Optionally, the memory 220 may include non-transitory computer-readable storage medium. The memory 220 may be used to store instructions, programs, code, code sets, or instruction sets. The memory 220 may include a program storage area and a data storage area, wherein the program storage area may store instructions for implementing an operating system, instructions for implementing at least one function (e.g., touch function, sound playback function, image playback function, etc.), instructions for implementing the various method embodiments described below, etc. The operating system may be an Android system, including systems deeply developed based on the Android system, an iOS system, including systems deeply developed based on the iOS system, or other systems.

[0059] In order for the operating system to distinguish the specific application scenarios of third-party applications, it is necessary to establish data communication between the third-party applications and the operating system. This would allow the operating system to obtain the current scenario information of the third-party applications at any time, and then perform targeted system resource adaptation based on the current scenario.

[0060] The input device 230 is used to receive input instructions or data, and includes, but is not limited to, a keyboard, mouse, camera, microphone, or touch device. The output device 240 is used to output instructions or data, and includes, but is not limited to, a display device and a speaker. In one example, the input device 230 and the output device 240 can be combined, and both the input device 230 and the output device 240 can be a touch display screen.

[0061] In addition, those skilled in the art will understand that the structure of the human sleeping posture prediction device shown in the above figures does not constitute a limitation on the human sleeping posture prediction device. The human sleeping posture prediction device may include more or fewer components than shown, or combine certain components, or have different component arrangements. For example, the human sleeping posture prediction device may also include radio frequency circuits, input units, sensors, audio circuits, Wireless Fidelity (WiFi) modules, power supplies, Bluetooth modules, etc., which will not be described in detail here.

[0062] Figure 3 A schematic diagram of a human sleeping posture prediction device according to some embodiments of this specification is shown. The human sleeping posture prediction device according to embodiments of this specification will now be described in detail with reference to the accompanying drawings.

[0063] like Figure 3 As shown, the human sleeping posture prediction device 300 includes a radar data acquisition module 310, a data transmission module 320, a data preprocessing module 330, and a sleeping posture prediction module 340. The human sleeping posture prediction device 300 includes a housing, the radar data acquisition module 310 is disposed on the housing, and the data transmission module 320, the data preprocessing module 330, and the sleeping posture prediction module 340 can be integrated inside the housing.

[0064] In an example embodiment, the radar data acquisition module 310 includes a radar module for acquiring human sleep data of a target human body. The radar module can be a millimeter-wave radar. The radar module includes a radar housing, a transmitting module, a scanning module, and a receiving module. The transmitting module is disposed inside the radar housing and fixedly connected to it; the receiving module is disposed inside the radar housing and fixedly connected to it, and the receiving module is correspondingly configured with the transmitting module. It should be noted that the transmitting module and the receiving module can be directly or indirectly fixedly connected to the radar housing. The transmitting module includes multiple laser emitting units, and the receiving module includes multiple detection units, which are correspondingly configured with the multiple laser emitting units. The transmitting module emits detection signals, which are reflected by external objects to form echo signals, which are then converged to the receiving module.

[0065] The data transmission module 320 is communicatively connected to the radar data acquisition module 310 and the data preprocessing module 330, and is used to transmit the human sleep data of the target human body acquired by the radar module to the data preprocessing module. The data transmission module 320 can be a data bus.

[0066] The data preprocessing module 330, communicatively connected to the sleep posture prediction module 340 via the data transmission module 320, is used to extract sleep signal features from the sleep data of the target human body and transmit these sleep signal features to the sleep posture prediction module 340 via the data transmission module 320. The data preprocessing module 330 can preprocess the human sleep data via software or hardware. For example, it can be implemented using a preprocessing circuit or as a software module stored in the memory.

[0067] The sleep position prediction module 340, communicatively connected to the data transmission module 320, is used to input sleep signal features into a pre-trained human sleep position prediction model to obtain the predicted sleep position corresponding to the target human body. The sleep position prediction module 340 can be implemented using a processor. For example, the sleep position prediction module 340 can input sleep signal features into the pre-trained human sleep position prediction model by running or executing instructions, programs, code sets, or instruction sets stored in memory, and by calling data stored in memory, to obtain the predicted sleep position corresponding to the target human body.

[0068] In the example embodiment, the human sleeping posture prediction model is a machine learning model that classifies human sleeping postures based on the sleep signal features of the target human body. The machine learning model can be a neural network model, a decision tree model, or a support vector machine model, such as a Transformer model or a convolutional neural network model.

[0069] The human sleeping posture prediction device 300's sleeping posture prediction module 340 inputs sleep signal features into the human sleeping posture prediction model. This model determines the corresponding sleeping posture signal features of the target human body based on these features, and identifies the target human body's sleeping posture, such as supine, prone, or lateral. Taking a convolutional neural network model as an example, the human sleeping posture prediction model includes convolutional layers, pooling layers, fully connected layers, and an output layer. The convolutional layers detect local features of the target human body, such as features of different body parts. The pooling layers reduce data dimensionality and computational complexity while retaining important features. The fully connected layers map features to a high-dimensional space for easier classification. The output layer converts the model's output into a probability distribution of human sleeping posture categories.

[0070] according to Figure 3The technical solution in the example embodiment, on the one hand, collects human sleep data of the target human body through a radar data acquisition module, and extracts the sleep signal features of the target human body from the human sleep data through a data preprocessing module. This not only enables high-precision and non-invasive acquisition of the sleep signal features of the target human body, but also accurately tracks the body movement of the target human body during sleep under multiple different locations and environmental conditions. On the other hand, the sleep signal features are input into the human sleep posture prediction model through a sleep posture prediction module to obtain the predicted sleep posture of the target human body, thereby accurately classifying the sleep posture of the target human body. This has a potentially broad application prospect for improving sleep quality monitoring and health management.

[0071] Furthermore, in some example embodiments, the human sleeping posture prediction device 300 further includes: a model training module, used to determine the cross-entropy loss between the predicted sleeping posture and the labeled sleeping posture; determine the gradient of the model parameters of the corresponding human sleeping posture prediction model based on the cross-entropy loss, and adjust the model parameters of the human sleeping posture prediction model through gradient backpropagation. The model training module can be a processor such as a GPU or CPU used for model training.

[0072] For example, the model training module of the human sleeping posture prediction device 300 calculates the gradient of the loss function with respect to each model parameter based on the cross-entropy loss value of the predicted sleeping posture obtained from the real sample labels and the human sleeping posture prediction model, and backpropagates the gradient to update the model parameters with a predetermined learning rate, thereby adjusting the weights and biases of the human sleeping posture network.

[0073] Furthermore, the model training module of the human sleeping posture prediction device 300 is also used to: determine the learning rate corresponding to the sample dataset; and update the model parameters of the human sleeping posture prediction model based on the learning rate corresponding to the sample dataset.

[0074] In the example embodiment, the electronic device 200 determines the learning rate corresponding to the sample dataset; based on the learning rate corresponding to the sample dataset, the model parameters of the human sleeping posture prediction model are updated. When the value of the loss function does not decrease within a certain number of iterations, the learning rate is reduced. For example, the initial learning rate is set to 0.001, and the learning rate reduction scheme is to halve it; the learning rate can be changed according to different needs.

[0075] For example, in the training method of a deep learning network for a human sleeping posture prediction model, the ratio of the training set, validation set, and test set can be 7:2:1, or it can be changed to other ratios depending on the actual dataset. The initial learning rate is preferably 0.001, but can be adjusted to different values ​​such as 0.01, 0.0001, and 0.00001, depending on the actual situation. The learning rate decay strategy can be to halve the learning rate when the loss function no longer decreases, or different choices can be made depending on the actual situation, such as using the criterion that the accuracy of the test set no longer increases, and exponentially decreasing the learning rate with the number of training iterations.

[0076] Figure 5 A schematic diagram of the structure of a sleeping posture prediction module provided according to some other embodiments of this specification is shown.

[0077] Reference Figure 5 As shown, the human sleeping posture prediction model includes a feature convolutional layer and a causal dilation convolutional layer. The sleeping posture prediction module 340 includes: a feature convolutional module 342, which is used to input the sleep signal features into the feature convolutional layer to obtain the sleeping posture signal features corresponding to the target human body; input the sleeping posture signal features into the causal dilation convolutional layer to obtain multi-scale sleeping posture features; and a prediction module 344, which is used to predict the predicted sleeping posture corresponding to the target human body based on the multi-scale sleeping posture features.

[0078] The sleep posture prediction module 340 of the human sleep posture prediction device 300 inputs sleep signal features into a feature convolutional layer to obtain sleep posture signal features corresponding to the target human body. The feature convolutional layer includes a residual network. The sleep posture signal features are then input into a causal dilation convolutional layer to obtain multi-scale sleep posture features, which are used to predict the target human body's sleep posture. The causal dilation convolutional layer includes a dilation convolutional layer and a causal convolutional layer. The dilation convolutional layer is used to obtain multi-scale temporal sleep posture features from the sleep posture signal features, while the causal convolutional layer introduces causality into the convolutional operation of the dilation convolutional layer, ensuring that the sleep posture features obtained by the dilation convolutional layer depend only on historical input data.

[0079] Taking the human sleeping posture prediction model as an example, see [link to relevant documentation]. Figure 4As shown, the human sleeping posture network model includes an input layer 410, a feature upscaling network 420, a feature convolutional layer 430, a residual structure 440, a fully connected layer 450, and a causal dilated convolutional layer 460. The input layer 410 for radar data has a dimension of C*H*W, where C is the number of feature channels in a single sample in the training sample set. The feature upscaling network 420 increases the number of feature channels to N (e.g., N is 64, 128, 256, etc.), correspondingly reducing the length and width of the feature map to 1 / K times that of the previous layer (e.g., K is 4, 6, 8, etc.). The feature extraction layer uses multiple feature convolutional layers 430 with residual structures 440 to extract sleeping posture signal features. The residual structure 440 is used to overcome the gradient vanishing problem and accelerate training.

[0080] Furthermore, the feature convolutional layer 430 is connected to the fully connected layer 450. The fully connected layer 450 integrates the features extracted by the feature convolutional layer 430 to obtain the sleeping posture signal feature vector, which is convenient for subsequent classification tasks. The output layer 470 outputs 512-dimensional sleeping posture signal features. The introduction of the causal dilated convolutional layer 460 can capture temporal information and multi-scale temporal dependencies, as well as a richer hierarchical structure. At the same time, it can also process time windows in parallel, improving the running efficiency. In order to avoid the problem of "future information leakage", the causal dilated convolutional layer is used to ensure that when outputting at a certain time step, only the input data of that time step and the previous time step are used. The output layer 470 can contain the same number of neurons as the number of categories of predicted sleeping postures. Using the softmax activation function, the output of the neurons is converted into a probability distribution, representing the probability of belonging to each category. For example, the number of predicted sleeping postures output is 4, that is, the predicted human sleeping posture can be supine, prone, left lateral, or right lateral.

[0081] According to the technical solutions in the above example embodiments, the human sleeping posture network model uses convolutional kernels of various sizes, and the network architecture includes components such as dilated convolution, causal convolution, residual blocks and pooling layers, which can better capture the spatiotemporal characteristics of sleep data and improve the model performance of the human sleeping posture network model.

[0082] Figure 6 A schematic diagram of the structure of a data preprocessing module provided according to some embodiments of this specification is shown.

[0083] See Figure 6 As shown, the data preprocessing module 320 includes: a range selection module 322, a noise reduction module 324, and a feature extraction module 326. Wherein:

[0084] The range selection module 326 is used to crop the human sleep data of a target human body based on a predetermined range, thereby obtaining the human sleep data of the target human body within the predetermined range. In the example embodiment, in order to reduce the redundancy of radar data, a selection range, i.e., a predetermined range, is preset for cropping the human sleep data collected by the radar. The electronic device 200 crops the human sleep data of the target human body based on the predetermined range, thereby obtaining the human sleep data of the target human body within the predetermined range.

[0085] For example, radar data within a predetermined distance (range d) from a radar device can be cropped. The optimal value of this distance d can be calculated using an optimization method based on variance variation. Variance reflects the degree of signal fluctuation; typically, the variance of the target echo signal is greater than that of the background noise. Therefore, the variation in variance can serve as a basis for selecting a suitable distance d. For instance, a radar continuously scans a human target, collecting echo signals from different range units. The signal strength of each range unit is statistically analyzed, and the variance of the echo signal for each range unit is calculated. The location where the variance increases significantly is taken as the human target's position. For example, the distance where the variance begins to increase significantly can be selected as the predetermined distance d, which typically marks the starting position of the target human target.

[0086] By selecting a specific range for radar data, the data can be made to include sufficient information about human sleeping postures while reducing the size of the dataset and allowing for focused analysis of information within the region of interest. In other words, selecting a specific range for radar data helps improve the efficiency and accuracy of data processing, enabling a better understanding and utilization of the radar data.

[0087] The noise reduction module 324 is used to filter out static clutter signals in the human sleep data of the target human body within a predetermined range, thereby obtaining noise-reduced human sleep data corresponding to the target human body. To reduce noise in radar data, the electronic device 200 performs static clutter filtering on the radar data, for example, using a moving target indication (MTI) method. For example, the electronic device 200 filters out static clutter signals in the human sleep data of the target human body within the aforementioned predetermined range, thereby obtaining noise-reduced human sleep data corresponding to the target human body.

[0088] By performing static clutter filtering on human sleep data within a predetermined range, the quality and accuracy of radar data can be improved.

[0089] It should be noted that although the moving target display method has been used as an example for illustration, those skilled in the art should understand that other appropriate methods, such as zero-velocity channel zeroing algorithm or vector mean cancellation algorithm, can also be used to filter out static clutter from radar data, which is also within the scope of the embodiments in this specification.

[0090] The feature extraction module 326 is used to perform frequency domain feature extraction and time domain feature extraction on the denoised human sleep data to obtain the sleep signal features of the target human. In an example embodiment, the feature extraction module 326 extracts the sleep signal features of the target human from the human sleep data through feature extraction. Feature extraction identifies specific patterns from the processed human sleep data, such as respiratory rate, body contour, and movement patterns. Sleep signal features can include time domain features, frequency domain features, and morphological features. For example, the feature extraction module 326 performs frequency domain feature extraction and time domain feature extraction on the human sleep data to obtain the sleep signal features of the target human.

[0091] For example, the feature extraction module 326 extracts time-domain features from the time-domain data of the target human's sleep data to obtain the time-domain features of the sleep signal. These time-domain features include maxima, minima, kurtosis, skewness, spectral maximum frequency, entropy, and differential mean. Furthermore, the electronic device 200 uses Fourier transform to analyze the acquired radar signal's human sleep data, converting the time-domain data into frequency-domain data and extracting frequency-domain features from the human sleep data. For instance, it extracts phase, spectral peaks, and other frequency-domain features from the Fourier-transformed radar frequency-domain signal.

[0092] Figure 7 A schematic diagram of the structure of a human sleeping posture prediction device provided according to some other embodiments of this specification is shown.

[0093] See Figure 7 As shown, the human sleeping posture prediction device also includes a display module 610, which is communicatively connected to the data transmission module 320. The display module 610 provides data display and radar parameter settings through a human-computer interaction interface. The data display includes showing the human sleep data of the target human body collected by the radar module and the predicted sleeping posture of the target human body. The display module 610 includes a display screen, on which a human-computer interaction interface is provided.

[0094] The display module 610 connects the user to the human sleeping posture prediction device through a human-computer interaction interface, providing users with real-time data display and radar parameter settings. It also acts as a data transmission channel, transmitting collected data to the sleeping posture prediction module for monitoring and analysis of human sleeping postures. The display module 610 is crucial for the operability and real-time performance of the human sleeping posture prediction system, enabling users to effectively interact with the system and obtain useful information.

[0095] Figure 8 A schematic diagram of the structure of a human sleeping posture prediction device provided according to some embodiments of this specification is shown.

[0096] Reference Figure 8As shown, the human sleeping posture prediction device or system includes: a millimeter-wave radar data acquisition module 810, a data transmission module 820, a data preprocessing module 830, a model training module 840, a sleeping posture estimation module 850, and a display module 860. The millimeter-wave radar data acquisition module 810 acquires millimeter-wave radar data of the human body during sleep, saves the acquired data via the data transmission module 820, and transmits it to the data preprocessing module and the display module 860.

[0097] The data transmission module 820 is communicatively connected to the radar data acquisition module 810 and the data preprocessing module 830, and is used to transmit the human sleep data of the target human body acquired by the radar module to the data preprocessing module 830.

[0098] The data preprocessing module 830 is communicatively connected to the data transmission module 820. It acquires radar wave data collected by the millimeter-wave radar data acquisition module 810 from the data transmission module 820, performs data preprocessing according to the aforementioned preprocessing method, and extracts radar data features. The radar signal features are saved via the data transmission module 820 and transmitted to the model training module 840 or the sleeping posture trajectory module 850. The data preprocessing module 830 includes a range selection module 8310, a noise reduction module 8320, a frequency domain feature extraction module 8330, a time domain feature extraction module 8340, a feature analysis module 8350, and a normalization module 8360.

[0099] The model training module 840 is communicatively connected to the data transmission module 820. It acquires radar signal features processed and extracted by the data preprocessing module 830 from the data transmission module 820, and trains the weights and biases of the human sleeping posture prediction model according to the aforementioned model training method. The trained network parameters are saved via the data transmission module 820 and transmitted to the sleeping posture estimation module 850.

[0100] The sleeping posture estimation module 850 is communicatively connected to the data transmission module 820. It obtains the pre-trained network weights trained by the deep model training module 840 from the data transmission module 820 to initialize the human sleeping posture prediction model. It also obtains radar signal features extracted by the data preprocessing module 830 from the data transmission module 820 and inputs them into the model network of the pre-trained human sleeping posture model to obtain the predicted sleeping posture (e.g., supine, prone, or lateral) corresponding to that data segment. The sleeping posture estimation result is then transmitted to the display module 860 via the data transmission module 820.

[0101] The display module 860 is used for human-computer interaction with the user, displaying data collected by millimeter-wave radar and sleep posture analysis results. The following are the specific functions and roles of the display module 860: (1) Human-computer interaction page: The display module includes a graphical user interface (GUI) for human-computer interaction. This GUI provides a way for users to interact with the human sleep posture prediction system. Users can operate and observe data on the GUI. (2) Data acquisition interface: The display module 860 works closely with the acquisition module 810 to achieve dynamic real-time data acquisition. It communicates with computer equipment through the data acquisition interface to control the acquisition module 810 to start, stop, or adjust the data acquisition process. (3) Data display: The display module 860 receives data from the radar acquisition module 810 and can display this radar data in real time. The radar data includes radar signals, sleep posture data, etc. Users can observe the changes and trends of the data on the interface. (4) Radar parameter setting: Users can set radar parameters in real time through the GUI. The parameters include adjusting the radar's operating frequency, receiving gain, number of transmitting and receiving antennas, etc., to ensure the quality and accuracy of data acquisition. (5) Sleep posture analysis: The display module displays the raw data and receives the sleep posture estimation results and displays them on the interface, so that users can understand the current human sleeping posture in real time.

[0102] The display module 860 connects the user to the human sleeping posture prediction system through a human-computer interaction interface, providing users with real-time data display and radar parameter settings. It also acts as a data transmission channel, transferring collected data to the sleeping posture analysis module to monitor and analyze human sleeping postures. The display module 860 is crucial for the operability and real-time performance of the human sleeping posture prediction system, enabling users to effectively interact with the system and obtain useful information.

[0103] It should be noted that the above modules can be deployed in the following ways: on local terminal computing devices; on mobile devices or servers; or in a distributed manner, such as deploying the acquisition module and display module locally, and the training module in the cloud.

[0104] This specification, in another aspect, provides a non-transitory storage medium storing at least one set of executable instructions for performing sleep posture prediction. When the executable instructions are executed by a processor, they instruct the processor to perform the steps of human sleep posture prediction described in this specification. In some possible embodiments, various aspects of this specification can also be implemented as a program product comprising program code. When the program product is run on a human sleep posture prediction device 200, the program code causes the human sleep posture prediction device 200 to perform the steps of human sleep posture prediction described in this specification. The program product for implementing the above methods may employ a portable compact disc read-only memory (CD-ROM) containing program code and may run on the human sleep posture prediction device 200. However, the program product of this specification is not limited thereto. In this specification, a readable storage medium may be any tangible medium containing or storing a program that may be used by or in conjunction with an instruction execution system. The program product may employ any combination of one or more readable media. A readable medium may be a readable signal medium or a readable storage medium. A readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of readable storage media include: electrical connections having one or more wires, portable disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. The computer-readable storage medium may include data signals propagated in baseband or as part of a carrier wave, carrying readable program code. Such propagated data signals may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A readable storage medium may also be any readable medium other than a readable storage medium that can send, propagate, or transmit programs for use by or in connection with an instruction execution system, apparatus, or device. Program code contained on a readable storage medium may be transmitted using any suitable medium, including but not limited to wireless, wired, optical fiber, RF, etc., or any suitable combination thereof. Program code for performing the operations described herein can be written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Java and C++, and conventional procedural programming languages ​​such as C or similar languages. The program code can be executed entirely on the human sleeping posture prediction device 200, partially on the human sleeping posture prediction device 200, as a standalone software package, partially on the human sleeping posture prediction device 200 and partially on a remote computing device, or entirely on a remote computing device.

[0105] The foregoing has described specific embodiments of this specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims may be performed in a different order than that shown in the embodiments and may still achieve the desired result. Furthermore, the processes depicted in the drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are possible or may be advantageous.

[0106] In summary, after reading this detailed disclosure, those skilled in the art will understand that the foregoing detailed disclosure is presented by way of example only and is not restrictive. Although not explicitly stated herein, those skilled in the art will understand that this specification requires various reasonable changes, improvements, and modifications to the embodiments. These changes, improvements, and modifications are intended to be made by this specification and are within the spirit and scope of the exemplary embodiments described herein.

[0107] Furthermore, certain terms in this specification have been used to describe embodiments of this specification. For example, "an embodiment," "an embodiment," and / or "some embodiments" mean that a particular feature, structure, or characteristic described in connection with that embodiment may be included in at least one embodiment of this specification. Therefore, it is to be emphasized and understood that two or more references to "an embodiment" or "an embodiment" or "alternative embodiment" in various parts of this specification do not necessarily refer to the same embodiment. Moreover, specific features, structures, or characteristics may be suitably combined in one or more embodiments of this specification.

[0108] It should be understood that in the foregoing description of the embodiments in this specification, various features are combined in a single embodiment, drawing, or description for the purpose of simplifying the description and aiding in the understanding of a feature. However, this does not mean that the combination of these features is necessary, and those skilled in the art may readily identify some of the devices as separate embodiments when reading this specification. That is, the embodiments in this specification can also be understood as an integration of multiple secondary embodiments. It is also valid when each secondary embodiment contains fewer than all the features of a single foregoing disclosed embodiment.

[0109] Each patent, patent application, publication of the patent application, and other materials such as articles, books, specifications, publications, documents, articles, etc., cited herein may be incorporated by reference. All contents used for all purposes, except for any history of prosecution documents relating to it, that may be inconsistent with or conflict with this document, or any such history of prosecution documents that may have a limiting effect on the widest extent of the claims, are now or hereafter associated with this document. For example, in the event of any inconsistency or conflict between the description, definition, and / or use of terms associated with any of the included materials and the terms, description, definition, and / or used in connection with this document, the terms used herein shall prevail.

[0110] Finally, it should be understood that the embodiments disclosed herein are illustrative of the principles of the embodiments described in this specification. Other modified embodiments are also within the scope of this specification. Therefore, the embodiments disclosed in this specification are merely examples and not limitations. Those skilled in the art can implement the applications described in this specification using alternative configurations based on the embodiments in this specification. Therefore, the embodiments in this specification are not limited to the embodiments precisely described in the applications.

Claims

1. A human sleeping posture prediction device, comprising a radar data acquisition module, a data transmission module, a data preprocessing module, and a sleeping posture prediction module, wherein: The radar data acquisition module includes a radar module, which is used to collect human sleep data of the target human body; The data transmission module is communicatively connected to the radar data acquisition module and the data preprocessing module, and is used to transmit the human sleep data of the target human body acquired by the radar module to the data preprocessing module. The data preprocessing module is communicatively connected to the sleep position prediction module and is used to extract the sleep signal features of the target human body from the human sleep data of the target human body, and transmit the sleep signal features to the sleep position prediction module via the data transmission module. as well as The sleeping posture prediction module is communicatively connected to the data transmission module and is used to input the sleep signal features into a pre-trained human sleeping posture prediction model to obtain the predicted sleeping posture corresponding to the target human body.

2. The human sleeping posture prediction device according to claim 1, wherein, The human sleeping posture prediction device also includes: The display module is communicatively connected to the data transmission module and is used to provide data display and radar parameter setting functions through a human-computer interaction interface. The data display includes displaying the human sleep data of the target human body collected by the radar module and the predicted sleeping posture of the target human body.

3. The human sleeping posture prediction device according to claim 1, wherein, The data preprocessing module includes: The range selection module is used to crop the human sleep data of the target human body based on a predetermined range to obtain the human sleep data of the target human body within the predetermined range.

4. The human sleeping posture prediction device according to claim 3, wherein, The data preprocessing module further includes: The noise reduction module is used to filter out static noise signals in the human sleep data of the target human body within the predetermined range, so as to obtain the noise-reduced human sleep data corresponding to the target human body.

5. The human sleeping posture prediction device according to claim 4, wherein, The data preprocessing module further includes: The feature extraction module is used to extract frequency domain features and time domain features from the noise-reduced human sleep data to obtain the sleep signal features of the target human body.

6. The human sleeping posture prediction device according to claim 1, wherein, The human sleeping posture prediction model includes a feature convolutional layer and a causal dilated convolutional layer, and the sleeping posture prediction module includes: The feature convolution module is used to input the sleep signal features into the feature convolution layer to obtain the sleeping posture signal features corresponding to the target human body; and to input the sleeping posture signal features into the causal dilation convolution layer to obtain multi-scale sleeping posture features. The prediction module is used to predict the sleeping posture of the target human body based on the multi-scale sleeping posture features.

7. The human sleeping posture prediction device according to claim 6, wherein, The causal dilated convolutional layer includes a dilated convolutional layer and a causal convolutional layer. The dilated convolutional layer is used to obtain sleeping posture features at multiple time scales from the sleeping posture signal features. The causal convolutional layer is used to introduce causality into the convolution operation of the dilated convolutional layer, so that the sleeping posture features obtained by the dilated convolutional layer depend only on historical input data.

8. The human sleeping posture prediction device according to claim 1, wherein, The human sleeping posture prediction device also includes: The model training module is used to determine the cross-entropy loss between the predicted sleeping position and the labeled sleeping position; based on the cross-entropy loss, it determines the gradient of the corresponding model parameters of the human sleeping position prediction model, and adjusts the model parameters of the human sleeping position prediction model through gradient backpropagation.

9. The human sleeping posture prediction device according to claim 8, wherein, The model training module is also used for: Determine the learning rate for the sample dataset; Based on the learning rate corresponding to the sample dataset, the model parameters of the human sleeping posture prediction model are updated.

10. The human sleeping posture prediction device according to claim 1, wherein, The human sleeping posture prediction model is a deep neural network model, and the radar module is a millimeter-wave radar.

Images