Blood oxygen saturation detection method and device, noise reduction equipment, storage medium and product

Abstract

This application discloses a method, apparatus, noise reduction device, storage medium, and product for blood oxygen saturation detection, relating to the field of artificial intelligence technology. The method includes: receiving an initial pulse wave signal acquired by a photoplethysmography (PPG) detection device; determining whether noise signals exist in the initial pulse wave signal; if noise signals are present in the initial pulse wave signal, inputting the initial pulse wave signal into a noise reduction model to obtain a target pulse wave signal, wherein the noise reduction model is trained using historical pulse wave signals as input data and denoised historical pulse wave signals as labels; and sending the target pulse wave signal to the PPG detection device, wherein the PPG detection device performs blood oxygen concentration detection based on the target pulse wave signal. This application improves the accuracy of blood oxygen saturation detection.

Description

Technical Field

[0001] This application relates to the field of artificial intelligence technology, and in particular to a method, apparatus, noise reduction device, storage medium and product for blood oxygen saturation detection. Background Technology

[0002] Photoplethysmogram (PPG) is a non-invasive method for measuring blood oxygen saturation in the human body. It uses infrared or fluorescence sensors to detect minute changes in the skin or nail bed to obtain blood oxygen saturation information. However, in practical applications, pulse wave signals are easily interfered with, such as by ambient light and electromyography (EMG) signals, affecting the accuracy of blood oxygen saturation detection.

[0003] The above content is only used to help understand the technical solution of this application and does not represent an admission that the above content is prior art. Summary of the Invention

[0004] The main purpose of this application is to provide a method, device, noise reduction equipment, storage medium and product for detecting blood oxygen saturation, aiming to solve the technical problem that the pulse wave signal is easily interfered with, affecting the accurate detection of blood oxygen saturation.

[0005] To achieve the above objectives, this application provides a method for detecting blood oxygen saturation. The method is applied to a noise reduction device, which is communicatively connected to a photoplethysmography (PPG) detection device. The blood oxygen saturation detection method includes:

[0006] Receive the initial pulse wave signal collected by the photoplethysmography detection device, and determine whether there is a noise signal in the initial pulse wave signal;

[0007] If the initial pulse wave signal contains the noise signal, the initial pulse wave signal is input into the denoising model to obtain the target pulse wave signal. The denoising model is trained using historical pulse wave signals as input data and denoised historical pulse wave signals as labels.

[0008] The target pulse wave signal is sent to the photoplethysmography (PPG) detection device, wherein the PPG detection device performs blood oxygen concentration detection based on the target pulse wave signal.

[0009] In one embodiment, the step of inputting the initial pulse wave signal into the denoising model to obtain the target pulse wave signal includes:

[0010] The initial pulse wave signal is filtered to obtain a filtered pulse wave signal of the target frequency band, wherein the target frequency band is the frequency band corresponding to the human heart rate range;

[0011] The filtered pulse wave signal is input into the noise reduction model to obtain the target pulse wave signal.

[0012] In one embodiment, the noise reduction model includes an encoder, a recurrent neural network, an attention mechanism module, a decoder, and a fully connected layer. The step of inputting the filtered pulse wave signal into the noise reduction model to obtain the target pulse wave signal includes:

[0013] The filtered pulse wave signal is input into the encoder for processing to obtain a first processing result;

[0014] The first processing result is input into the recurrent neural network module for processing to obtain a second processing result. When processing the first processing result through the recurrent neural network module, the recurrent neural network module is used to process the previous frame processing result, which is the result obtained by the encoder processing the previous frame signal of the initial pulse wave signal.

[0015] The second processing result is input into the decoder for further processing to obtain the third processing result;

[0016] The first processing result and the third processing result are input into the attention mechanism module for processing to obtain the fourth processing result;

[0017] The third processing result and the fourth processing result are connected, and the result obtained by the connection is input into the fully connected layer for processing to obtain the target pulse wave signal.

[0018] In one embodiment, the attention mechanism module includes a weight calculation block and a weighted processing block. The step of inputting the first processing result and the third processing result into the attention mechanism module for processing to obtain a fourth processing result includes:

[0019] The first processing result and the third processing result are input into the weight calculation block for processing to obtain the attention weight matrix;

[0020] The attention weight matrix and the first processing result are input into the weighted processing block for processing to obtain the fourth processing result.

[0021] In one embodiment, the step of determining whether there is a noise signal in the initial pulse wave signal includes:

[0022] Feature extraction is performed on the initial pulse wave signal to obtain signal waveform features, and the target interference type of the initial pulse wave signal is determined based on the signal waveform features;

[0023] If the target interference type is a non-physiological interference type, then it is determined that there is a noise signal in the initial pulse wave signal;

[0024] If the target interference type is a physiological interference type, then it is determined that the noise signal does not exist in the initial pulse wave signal.

[0025] In one embodiment, the signal waveform features include signal variance and baseline offset; the step of determining the target interference type of the initial pulse wave signal based on the signal waveform features includes:

[0026] If the signal variance is greater than a preset variance threshold and the baseline offset is greater than a preset offset, then the target interference type of the initial pulse wave signal is determined to be physiological interference.

[0027] If the signal variance is greater than the preset variance threshold and the baseline offset is less than or equal to the preset offset, then the target interference type is determined to be non-physiological interference.

[0028] To achieve the above objectives, this application also provides a blood oxygen saturation detection device, which is applied to a noise reduction device and is communicatively connected to a photoplethysmography detection device. The blood oxygen saturation detection device includes:

[0029] The judgment module is used to receive the initial pulse wave signal collected by the photoplethysmography detection device and determine whether there is a noise signal in the initial pulse wave signal.

[0030] A noise reduction module is used to input the initial pulse wave signal into a noise reduction model to obtain a target pulse wave signal if the initial pulse wave signal contains the noise signal. The noise reduction model is trained using historical pulse wave signals as input data and the denoised historical pulse wave signals as labels.

[0031] The transmitting module is used to transmit the target pulse wave signal to the photoplethysmography detection device, wherein the photoplethysmography detection device performs blood oxygen concentration detection based on the target pulse wave signal.

[0032] To achieve the above objectives, this application also provides a noise reduction device, which includes: a memory, a processor, and a blood oxygen saturation detection program stored in the memory and executable on the processor. When the blood oxygen saturation detection program is executed by the processor, it implements the steps of the blood oxygen saturation detection method as described above.

[0033] In addition, to achieve the above objectives, this application also proposes a computer-readable storage medium storing a blood oxygen saturation detection program, which, when executed by a processor, implements the steps of the blood oxygen saturation detection method as described above.

[0034] In this application, by receiving the initial pulse wave signal collected by the photoplethysmography (PPG) detection device, it is determined whether there is a noise signal in the initial pulse wave signal; if the noise signal is present in the initial pulse wave signal, the initial pulse wave signal is input into a denoising model to obtain a target pulse wave signal, wherein the denoising model is trained using historical pulse wave signals as input data and denoised historical pulse wave signals as labels; the target pulse wave signal is sent to the PPG detection device, wherein the PPG detection device performs blood oxygen concentration detection based on the target pulse wave signal.

[0035] This application introduces a noise reduction model to identify and remove noise signals from the initial pulse wave signal, thereby extracting a purer target pulse wave signal and reducing the interference of noise on the pulse wave signal. Therefore, blood oxygen concentration detection based on the processed target pulse wave signal can more accurately reflect the blood oxygen saturation of the human body, thereby improving the accuracy of blood oxygen detection. Attached Figure Description

[0036] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0037] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0038] Figure 1 This is a flowchart illustrating the blood oxygen saturation detection method of this application in Embodiment 1.

[0039] Figure 2 This is a flowchart illustrating Embodiment 2 of the blood oxygen saturation detection method of this application;

[0040] Figure 3 This is a structural diagram of a deep neural network according to an embodiment of this application;

[0041] Figure 4 This is a schematic diagram of the module structure of the blood oxygen saturation detection device according to an embodiment of this application;

[0042] Figure 5This is a schematic diagram of the hardware operating environment involved in the blood oxygen saturation detection method in the embodiments of this application.

[0043] The realization of the purpose, functional features and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0044] It should be understood that the specific embodiments described herein are merely illustrative of the technical solutions of this application and are not intended to limit this application.

[0045] To better understand the technical solution of this application, a detailed description will be provided below in conjunction with the accompanying drawings and specific implementation methods.

[0046] The main solution of this application embodiment is as follows: receiving the initial pulse wave signal collected by the photoplethysmography (PPG) detection device, determining whether there is a noise signal in the initial pulse wave signal; if the noise signal is present in the initial pulse wave signal, inputting the initial pulse wave signal into a denoising model to obtain a target pulse wave signal, wherein the denoising model is trained using historical pulse wave signals as input data and denoised historical pulse wave signals as labels; sending the target pulse wave signal to the PPG detection device, wherein the PPG detection device performs blood oxygen concentration detection based on the target pulse wave signal.

[0047] In this embodiment, for ease of description, the following description will focus on the noise reduction device as the implementing entity.

[0048] Pulse oximetry is a non-invasive method for measuring blood oxygen saturation. It uses infrared or fluorescence sensors to detect minute changes in the skin or nail bed to obtain blood oxygen saturation information. However, in practical applications, pulse wave signals are easily interfered with by factors such as ambient light and electromyography (EMG), affecting the accuracy of blood oxygen saturation detection.

[0049] This application introduces a noise reduction model to identify and remove noise signals from the initial pulse wave signal, thereby extracting a purer target pulse wave signal and reducing the interference of noise on the pulse wave signal. Therefore, blood oxygen concentration detection based on the processed target pulse wave signal can more accurately reflect the blood oxygen saturation of the human body, thereby improving the accuracy of blood oxygen detection.

[0050] It should be noted that the executing entity in this embodiment can be a computing service device with data processing, network communication, and program execution functions, such as a tablet computer, personal computer, or mobile phone, or an electronic device or noise reduction device capable of performing the above functions. The following description uses a noise reduction device as an example to illustrate this embodiment and the subsequent embodiments. It should be noted that in this embodiment, the noise reduction device is communicatively connected to the photoplethysmography detection device, and a noise reduction model is deployed in the noise reduction device.

[0051] Based on this, the embodiments of this application provide a method for detecting blood oxygen saturation, referring to... Figure 1 , Figure 1 This is a flowchart illustrating the first embodiment of the blood oxygen saturation detection method of this application.

[0052] In this embodiment, the noise reduction device is communicatively connected to the photoplethysmography detection device, and the blood oxygen saturation detection method includes steps S10 to S30:

[0053] Step S10: Receive the initial pulse wave signal collected by the photoplethysmography detection device, and determine whether there is a noise signal in the initial pulse wave signal;

[0054] It receives the initial pulse wave signal acquired by a photoplethysmography (PPG) detection device. PPG technology monitors the pulse wave by measuring changes in blood volume under the skin. The PPG device illuminates the skin with a light-emitting diode (LED) and uses a photoelectric sensor to receive the reflected or transmitted light signal. The change in the light signal is proportional to the change in blood volume, thereby generating a pulse wave signal.

[0055] It is understandable that the received initial pulse wave signal may contain various noises, such as ambient light interference, device noise, motion artifacts, etc. Signal processing techniques (such as threshold detection and spectrum analysis) are used to preliminarily determine whether there is significant noise in the signal. The specific judgment process will not be elaborated here.

[0056] Step S20: If the initial pulse wave signal contains the noise signal, then the initial pulse wave signal is input into the denoising model to obtain the target pulse wave signal, wherein the denoising model is trained using historical pulse wave signals as input data and denoised historical pulse wave signals as labels.

[0057] If noise is present in the initial pulse wave signal, a pre-trained denoising model is used to denoise the signal. This model is trained based on historical pulse wave signals (including noise) and their corresponding denoised signals, enabling it to identify and remove noise, thus obtaining a cleaner target pulse wave signal. The denoising model employs deep learning algorithms, such as Convolutional Neural Networks (CNNs) and Recurrent Neural Networks (RNNs). During the model training phase, a large amount of noisy historical pulse wave signals are input, along with corresponding denoised labeled data. Optimization algorithms (such as gradient descent) are used to adjust the model parameters, enabling the model to accurately predict the denoised signal.

[0058] By inputting a noisy initial pulse wave signal into a trained denoising model and outputting a denoised target pulse wave signal, the signal-to-noise ratio of the pulse wave signal can be significantly improved, errors can be reduced, and the denoising model can adapt to noise from different sources and types, thus improving generalization ability.

[0059] Step S30: The target pulse wave signal is sent to the photoplethysmography detection device, wherein the photoplethysmography detection device performs blood oxygen concentration detection based on the target pulse wave signal.

[0060] The denoised target pulse wave signal is sent back to the PPG detection device for further blood oxygen saturation detection. Blood oxygen saturation is one of the important indicators for assessing human health, and it is usually calculated by analyzing specific characteristics (such as pulse amplitude and frequency) in the pulse wave signal.

[0061] Understandably, compared to the initial signal, the target pulse wave signal has a higher signal-to-noise ratio and can more accurately reflect changes in blood volume. PPG devices can calculate blood oxygen saturation (SpO2) values ​​by analyzing the absorption spectral characteristics of the signal, displaying the results to the user or transmitting them to other medical devices for further analysis, thus providing more reliable data support for clinical diagnosis and treatment.

[0062] In one feasible implementation, step S10: determining whether there is a noise signal in the initial pulse wave signal, includes:

[0063] Step S101: Extract features from the initial pulse wave signal to obtain signal waveform features, and determine the target interference type of the initial pulse wave signal based on the signal waveform features;

[0064] Feature extraction is performed on the initial pulse wave signal. These features typically include waveform characteristics such as frequency, amplitude, period, and morphology. These features describe the basic properties of the pulse wave signal and serve as input data to a pre-defined interference identification model. The interference identification model, trained based on historical data and machine learning algorithms, is capable of identifying and classifying different types of interference. It should be noted that the feature extraction process is not limited here; methods such as Fourier transform, wavelet transform, and filtering can be used to extract key waveform features from the initial pulse wave signal.

[0065] The initial pulse wave signal is subjected to feature extraction to obtain signal waveform features. The specific method is not limited here. For example, the signal waveform features can be input into a preset interference identification model to obtain the target interference type. The interference identification model is trained using the signal features of historical pulse wave signals as input data and the interference type of the historical pulse wave signals as training labels. The signal waveform features are then input into the preset interference identification model to obtain the target interference type.

[0066] By extracting features and identifying interference, the type of interference in pulse wave signals can be automatically and accurately determined, avoiding the subjectivity and inefficiency of manual judgment, improving the efficiency and accuracy of signal processing, and providing reliable input data for subsequent steps.

[0067] Step S102: If the target interference type is a non-physiological interference type, then it is determined that there is a noise signal in the initial pulse wave signal;

[0068] Based on the output of the interference identification model, it is determined whether the target interference type is non-physiological. Non-physiological interference typically includes motion artifacts, equipment noise, electromagnetic interference, etc. These interferences can significantly affect the quality of the pulse wave signal, thereby affecting subsequent blood oxygen concentration detection. Therefore, when non-physiological interference is detected, it is determined that there is a noise signal in the initial pulse wave signal.

[0069] Step S103: If the target interference type is a physiological interference type, then it is determined that the noise signal does not exist in the initial pulse wave signal.

[0070] If the interference types output by the interference identification model do not include non-physiological interference types, but rather physiological interference (such as respiratory fluctuations, changes in heart rhythm, etc.), then it is determined that there is no noise signal in the initial pulse wave signal that requires special processing. Physiological interference reflects that normal physiological activities of the human body usually do not significantly affect blood oxygen concentration detection.

[0071] Understandably, by reducing unnecessary noise reduction steps, the efficiency of signal processing is improved while preserving the physiological information in the pulse wave signal, which helps to improve the efficiency of blood oxygen concentration detection.

[0072] In one feasible implementation, the signal waveform features include signal variance and baseline offset; step S101, the step of determining the target interference type of the initial pulse wave signal based on the signal waveform features, includes:

[0073] Step S1011: If the signal variance is greater than a preset variance threshold and the baseline offset is greater than a preset offset, then the target interference type of the initial pulse wave signal is determined to be physiological interference.

[0074] In this embodiment, the influence of physiological interference on the initial pulse wave signal is determined based on a joint judgment of signal variance and baseline offset. Specifically, the variance of the initial pulse wave signal is calculated and compared with a preset variance threshold. If the signal variance is greater than the threshold, it indicates that there is a large fluctuation in the signal. The baseline offset of the signal is calculated and compared with a preset offset threshold. If the baseline offset is also greater than the threshold, it indicates that the signal baseline has shifted significantly, and physiological interference is considered to be present. It can be understood that by jointly judging signal variance and baseline offset, physiological interference can be identified more accurately, the possibility of misjudgment can be reduced, the efficiency and accuracy of signal processing can be improved, and more reliable input data can be provided for subsequent steps.

[0075] Step S1012: If the signal variance is greater than the preset variance threshold and the baseline offset is less than or equal to the preset offset, then the target interference type is determined to be non-physiological interference.

[0076] If the signal variance is greater than the preset variance threshold but the baseline offset is less than or equal to the preset offset, it means that there is a large fluctuation in the signal, but the baseline has not shifted significantly. The target interference type of the initial pulse wave signal is determined to be non-physiological interference.

[0077] Understandably, by jointly judging the signal variance and baseline offset, it is possible to more accurately identify the types of physiological and non-physiological interference in the initial pulse wave signal, providing reliable input data and decision-making basis for subsequent steps.

[0078] This embodiment introduces a noise reduction model to identify and remove noise signals from the initial pulse wave signal, thereby extracting a purer target pulse wave signal and reducing the interference of noise on the pulse wave signal. Therefore, blood oxygen concentration detection based on the processed target pulse wave signal can more accurately reflect the blood oxygen saturation of the human body, thereby improving the accuracy of blood oxygen detection.

[0079] Based on the first embodiment of this application, in the second embodiment of this application, the content that is the same as or similar to that in the first embodiment described above can be referred to the above description, and will not be repeated hereafter. Based on this, please refer to... Figure 2 Step S20: The step of inputting the filtered pulse wave signal into the denoising model to obtain the target pulse wave signal includes:

[0080] Step S201: Filter the initial pulse wave signal to obtain a filtered pulse wave signal of the target frequency band, wherein the target frequency band is the frequency band corresponding to the human heart rate range;

[0081] The purpose of filtering the initial pulse wave signal is to extract the signal components related to the target frequency band (the frequency band corresponding to the range of human heart rate). Filtering can remove noise and interference that are not in the target frequency band, thereby obtaining a clearer pulse wave signal.

[0082] Specifically, a frequency band corresponding to the range of human heart rate is preset. Generally, the normal heart rate range for adults is between 60 and 100 beats per minute, corresponding to a frequency range of 1 to 1.67 Hz (assuming the heart rate is regular and each heartbeat cycle is constant). It is understandable that in practical applications, the heart rate may vary due to individual differences, exercise status, and other factors, so the target frequency band may be adjusted appropriately according to the specific situation. A target filter is determined, and the filtered pulse signal wave is extracted through the filter.

[0083] Understandably, filtering can remove noise and interference from the initial pulse wave signal that is not in the target frequency band, thereby improving the signal-to-noise ratio. The filtered pulse wave signal is clearer, which is beneficial for subsequent noise reduction model processing and the accuracy of blood oxygen concentration detection.

[0084] Step S202: Input the filtered pulse wave signal into the noise reduction model to obtain the target pulse wave signal.

[0085] The filtered pulse wave signal is processed for noise reduction using a deep neural network incorporating an attention mechanism. The result is referred to as the first signal for differentiation. In this embodiment, the features of the filtered pulse wave signal extracted by the deep neural network are weighted based on attention weights using an attention mechanism. The higher the similarity between the features of the filtered pulse wave signal and the features of the denoised filtered pulse wave signal, the greater the attention weight. This allows the deep neural network to focus on important information in the filtered pulse wave signal through the weighted processing of the attention mechanism, thereby making the noise reduction of the filtered pulse wave signal more accurate.

[0086] In a specific implementation, the noise reduction device can perform frame-by-frame processing on the received pulse wave signal, and perform blood oxygen saturation detection processing on each frame of the received pulse wave signal in the order of frames. In this implementation, the currently processed frame of the received pulse wave signal is used as the initial pulse wave signal.

[0087] This embodiment introduces an attention mechanism into the deep neural network. The attention mechanism weights the filtered pulse wave signal features extracted by the deep neural network based on attention weights. The higher the similarity between the filtered pulse wave signal features and the denoised filtered pulse wave signal features, the greater the attention weight. This allows the deep neural network to focus on important information in the filtered pulse wave signal during processing, thereby making noise removal more accurate. This improves the accuracy of blood oxygen saturation detection, which is easily affected by interference in the pulse wave signal.

[0088] In one feasible embodiment, the deep neural network includes an encoder, an attention mechanism module, a decoder, and a fully connected layer, and step S202 includes:

[0089] Step S2021: Input the filtered pulse wave signal into the encoder for processing to obtain a first processing result;

[0090] In this embodiment, the pre-defined deep neural network may include an encoder, a recurrent neural network module, an attention mechanism module, a decoder, and a fully connected layer. The encoder layer extracts and downsamples the data features of the input initial pulse wave signal; the recurrent neural network module processes the output of the encoder layer, utilizing the results from processing the initial pulse wave signal of the previous frame to eliminate noise in the current frame and improve blood oxygen saturation detection; the attention mechanism module focuses on key signals in the initial pulse wave signal, i.e., signals in the target frequency band; the decoder upsamples the output of the recurrent neural network; and the fully connected layer processes the output of the decoder to output the noise-reduced signal. This deep neural network can be pre-trained using a training dataset, and the training method can employ conventional neural network training methods, which will not be elaborated here.

[0091] In one feasible implementation, the encoder and decoder can draw on the encoder-decoder structure in the U-net network, that is, the decoder is used to implement the bridging and upsampling of data features. The noise reduction device inputs the filtered pulse wave signal from the initial pulse wave signal into the encoder for processing to obtain the processing result corresponding to the target initial pulse wave signal (hereinafter referred to as the first processing result for distinction).

[0092] Step S2022: The first processing result is input into the recurrent neural network module for processing to obtain a second processing result. When the first processing result is processed by the recurrent neural network module, the result obtained by processing the previous frame processing result is used. The previous frame processing result is the result obtained by the encoder processing the previous frame signal of the initial pulse wave signal.

[0093] When the noise reduction device performs noise cancellation, it processes the received initial pulse wave signal according to the time sequence. In this embodiment, the previous frame signal of the current initial pulse wave signal is called the previous frame signal, and the processing result of the encoder on the previous frame signal is called the previous frame processing result.

[0094] In this embodiment, the first processing result is input into the recurrent neural network module for processing to obtain the second processing result. It can be understood that when the first processing result is processed by the recurrent neural network module, the result obtained by processing the previous frame's processing result is also processed by the recurrent neural network module.

[0095] Step S2023: Input the second processing result into the decoder for processing to obtain the third processing result;

[0096] The second processing result is input into the decoder for further processing, and the resulting result is called the third processing result to distinguish it.

[0097] Step S2024: Input the first processing result and the third processing result into the attention mechanism module for processing to obtain the fourth processing result;

[0098] The first and third processing results are input into the attention mechanism module for processing, and the resulting result is called the fourth processing result to distinguish them.

[0099] In a specific implementation, an attention weight matrix representing the degree of attention to the information in the first processing result can be calculated based on the first processing result and the third processing result, and the first processing result can be weighted by the attention weight matrix to obtain the fourth processing result.

[0100] The attention mechanism module performs weighted processing on the first processing result, thereby assigning weights to the features in the first processing result. This allows for the filtering and enhancement of important information in the first processing result, enabling the deep neural network to focus on key information during the decoding process, namely the signal in the target frequency band corresponding to the heart rate, thus improving the accuracy of noise cancellation.

[0101] Step S2025: Connect the third processing result and the fourth processing result, and input the connected result into the fully connected layer for processing to obtain the target pulse wave signal.

[0102] The third and fourth processing results are connected, and the result obtained from the connection is input into the fully connected layer for processing to obtain the target pulse wave signal.

[0103] like Figure 3 As shown, a schematic diagram of the deep neural network in this embodiment is drawn. In the diagram, R_rnn represents a recurrent neural network, and the encoder can include multiple coding layers (such as...). Figure 3 The three encoding layers shown can be represented by a decoder, which may include multiple decoding layers. Figure 3 The diagram shows three decoding layers. The decoding layers can be used for deconvolution. The number of attention mechanism modules is the same as the number of encoder layers (i.e., the number of decoder layers). Each layer of the encoder and decoder is connected to the attention mechanism module. The output data of the attention mechanism module is connected to the output data of the decoder as the input data of the next layer.

[0104] The coding layer can employ one-dimensional convolution (1D-conv, i.e.) Figure 3 The implementation shown is conv+downsampling+activation function. Downsampling can be achieved using a 2*2 pooling layer, and the activation function can be LeakyRelu.

[0105] Further, in a feasible embodiment, the attention mechanism module includes a weight calculation block and a weighted processing block. Step S2024: inputting the first processing result and the third processing result into the attention mechanism module for processing to obtain a fourth processing result includes:

[0106] Step S20241: Input the first processing result and the third processing result into the weight calculation block for processing to obtain the attention weight matrix;

[0107] In this embodiment, the first processing result and the third processing result are input into the weight calculation block for processing to obtain the attention weight matrix. The specific method for calculating the attention weight matrix is ​​not limited in this embodiment.

[0108] Step S20242: Input the attention weight matrix and the first processing result into the weighted processing block for processing to obtain the fourth processing result.

[0109] The attention weight matrix and the first processing result are input into the weighted processing block for further processing to obtain the fourth processing result.

[0110] Further, in a feasible embodiment, the weight calculation block includes three convolutional layers, one dot product layer, and two activation functions. The process of inputting the first processing result and the third processing result into the weight calculation block to obtain the attention weight matrix can be as follows: inputting the first processing result into the first convolutional layer for convolution operation to obtain the first convolution result; inputting the third processing result into the second convolutional layer for convolution operation to obtain the second convolution result; inputting the first convolution result and the second convolution result into the first dot product layer to calculate the first dot product result; inputting the first dot product result into the first activation function for processing to obtain the first activation result; inputting the first activation result into the third convolutional layer for convolution operation to obtain the third convolution result; and inputting the third convolution result into the second activation function for processing to obtain the attention weight matrix. It should be noted that the first convolution weight, the second convolution weight, and the third convolution weight can be obtained during the model training phase; the training process will not be elaborated here. In this embodiment, feature enhancement is performed through convolution operations, and nonlinear factors are added through activation functions to improve the expressive power of the deep neural network for the model.

[0111] Further, in a feasible embodiment, the weighted processing block includes a convolutional layer and a dot-product layer. The process of inputting the attention weight matrix and the first processing result into the weighted processing block to obtain the fourth processing result can be as follows: inputting the first processing result into the fourth convolutional layer to perform a convolution operation to obtain the fourth convolution result; inputting the fourth convolution result and the attention weight matrix into the second dot-product layer to calculate the fourth processing result.

[0112] During the training of a deep neural network with recurrent neural network layers, backpropagation can be used to calculate the gradient of each parameter, and the parameters can be updated based on the gradient.

[0113] In this embodiment, an attention mechanism module is introduced to perform weighted processing on the first processing result, thereby allocating weights to the features in the first processing result. This allows for the filtering and enhancement of important information in the first processing result, enabling the deep neural network to focus on key information during the decoding process, thus improving the accuracy of noise cancellation.

[0114] For example, to help understand the implementation flow of the blood oxygen saturation detection method obtained by combining this embodiment with the above embodiment one:

[0115] a. Data Preprocessing: To address noise and electromyographic interference in the raw pulse wave signal, preprocessing techniques such as filtering and noise reduction are employed to improve signal quality. Specific methods include Butterworth filtering, wavelet transform, and adaptive filtering.

[0116] b. Feature Extraction: Distinguishing features are extracted from the preprocessed pulse wave signal to reflect the essential characteristics of the interference waveform. Specific features include: time-domain features (such as mean, standard deviation, kurtosis), frequency-domain features (power spectral density, Fourier transform coefficients, etc.), and time-frequency features (Hilbert-Huang transform coefficients, etc.).

[0117] c. Machine learning algorithm: The extracted features are classified and identified using a neural network (NN) machine learning algorithm to obtain candidate interference types, and the parameters are optimized to improve the recognition accuracy (that is, feature extraction is performed on the initial pulse wave signal to obtain signal waveform features, and the target interference type of the initial pulse wave signal is determined based on the signal waveform features).

[0118] d. Deep Learning Algorithm: A Convolutional Neural Network (CNN) deep learning algorithm is used to further identify candidate interference types and obtain the final interference waveform type. The deep learning algorithm can adaptively learn feature representations to improve recognition accuracy and robustness (i.e., receiving the initial pulse wave signal collected by the photoplethysmography detection device, determining whether there is a noise signal in the initial pulse wave signal; if the noise signal exists in the initial pulse wave signal, then inputting the initial pulse wave signal into the denoising model to obtain the target pulse wave signal).

[0119] e. Adaptive Adjustment: Based on the recognition results, the preprocessing and feature extraction methods are dynamically adjusted to adapt to the interference waveform recognition needs under different environments and usage scenarios. For example, when the recognition accuracy is low, the filter order can be increased or a more complex feature extraction method can be adopted.

[0120] f. Real-time optimization: For application scenarios with high real-time requirements, it supports the use of optimized algorithms and hardware platforms, such as FPGA and ASIC, to improve the real-time performance of interference waveform recognition.

[0121] It should be noted that the above examples are only for understanding this application and do not constitute a limitation on the blood oxygen saturation detection method of this application. Any simple modifications based on this technical concept are within the protection scope of this application.

[0122] This application also provides a blood oxygen saturation detection device, which is applied to a noise reduction device. The noise reduction device is communicatively connected to a photoplethysmography (PPG) detection device. Please refer to [reference needed]. Figure 4 The blood oxygen saturation detection device includes:

[0123] The judgment module 10 is used to receive the initial pulse wave signal collected by the photoplethysmography detection device and determine whether there is a noise signal in the initial pulse wave signal.

[0124] The noise reduction module 20 is used to input the initial pulse wave signal into the noise reduction model to obtain the target pulse wave signal if the initial pulse wave signal contains the noise signal. The noise reduction model is trained using historical pulse wave signals as input data and the denoised historical pulse wave signals as labels.

[0125] The transmitting module 30 is used to transmit the target pulse wave signal to the photoplethysmography detection device, wherein the photoplethysmography detection device performs blood oxygen concentration detection based on the target pulse wave signal.

[0126] Optionally, the noise reduction module 20 is used for:

[0127] The initial pulse wave signal is filtered to obtain a filtered pulse wave signal of the target frequency band, wherein the target frequency band is the frequency band corresponding to the human heart rate range;

[0128] The filtered pulse wave signal is input into the noise reduction model to obtain the target pulse wave signal.

[0129] Optionally, the noise reduction model includes an encoder, a recurrent neural network, an attention mechanism module, a decoder, and a fully connected layer. The noise reduction module 20 is used for:

[0130] The filtered pulse wave signal is input into the encoder for processing to obtain a first processing result;

[0131] The first processing result is input into the recurrent neural network module for processing to obtain a second processing result. When processing the first processing result through the recurrent neural network module, the recurrent neural network module is used to process the previous frame processing result, which is the result obtained by the encoder processing the previous frame signal of the initial pulse wave signal.

[0132] The second processing result is input into the decoder for further processing to obtain the third processing result;

[0133] The first processing result and the third processing result are input into the attention mechanism module for processing to obtain the fourth processing result;

[0134] The third processing result and the fourth processing result are connected, and the result obtained by the connection is input into the fully connected layer for processing to obtain the target pulse wave signal.

[0135] Optionally, the attention mechanism module includes a weight calculation block and a weighted processing block, and the noise reduction module 20 is used for:

[0136] The first processing result and the third processing result are input into the weight calculation block for processing to obtain the attention weight matrix;

[0137] The attention weight matrix and the first processing result are input into the weighted processing block for processing to obtain the fourth processing result.

[0138] Optionally, the determination module 10 is used to:

[0139] Feature extraction is performed on the initial pulse wave signal to obtain signal waveform features, and the target interference type of the initial pulse wave signal is determined based on the signal waveform features;

[0140] If the target interference type is a non-physiological interference type, then it is determined that there is a noise signal in the initial pulse wave signal;

[0141] If the target interference type is a physiological interference type, then it is determined that the noise signal does not exist in the initial pulse wave signal.

[0142] Optionally, the signal waveform features include signal variance and baseline offset; the judgment module 10 is used to:

[0143] If the signal variance is greater than a preset variance threshold and the baseline offset is greater than a preset offset, then the target interference type of the initial pulse wave signal is determined to be physiological interference.

[0144] If the signal variance is greater than the preset variance threshold and the baseline offset is less than or equal to the preset offset, then the target interference type is determined to be non-physiological interference.

[0145] The blood oxygen saturation detection device provided in this application, employing the blood oxygen saturation detection method described in the above embodiments, can solve the technical problem that pulse wave signals are easily interfered with, affecting the accurate detection of blood oxygen saturation. Compared with the prior art, the beneficial effects of the blood oxygen saturation detection device provided in this application are the same as those of the blood oxygen saturation detection method provided in the above embodiments, and other technical features in the blood oxygen saturation detection device are the same as those disclosed in the methods of the above embodiments, and will not be repeated here.

[0146] This application provides a noise reduction device, which includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, which are executed by the at least one processor to enable the at least one processor to perform the blood oxygen saturation detection method in the above embodiment 1.

[0147] The following is for reference. Figure 5The diagram illustrates a structural schematic suitable for implementing the noise reduction device in the embodiments of this application. The noise reduction device in the embodiments of this application may include, but is not limited to, mobile terminals such as mobile phones, laptops, digital broadcast receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Descriptions), PMPs (Portable Media Players), in-vehicle terminals (e.g., in-vehicle navigation terminals), and fixed terminals such as digital TVs and desktop computers. Figure 5 The noise reduction device shown is merely an example and should not be construed as limiting the functionality and scope of the embodiments of this application.

[0148] like Figure 5 As shown, the noise reduction device may include a processing unit 1001 (e.g., a central processing unit, a graphics processing unit, etc.), which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 1002 or a program loaded from a storage device 1003 into a random access memory (RAM) 1004. The RAM 1004 also stores various programs and data required for the operation of the noise reduction device. The processing unit 1001, ROM 1002, and RAM 1004 are interconnected via a bus 1005. An input / output (I / O) interface 1006 is also connected to the bus. Typically, the following systems can be connected to the I / O interface 1006: input devices 1007 including, for example, a touchscreen, touchpad, keyboard, mouse, image sensor, accelerometer, gyroscope, etc.; output devices 1008 including, for example, a liquid crystal display (LCD), speaker, vibrator, etc.; storage devices 1003 including, for example, magnetic tape, hard disk, etc.; and communication devices 1009. The communication device 1009 allows the noise reduction device to communicate wirelessly or wiredly with other devices to exchange data. Although the figure shows noise reduction devices with various systems, it should be understood that it is not required to implement or have all of the systems shown. More or fewer systems may be implemented alternatively.

[0149] Specifically, according to the embodiments disclosed in this application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments disclosed in this application include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device, or installed from storage device 1003, or installed from ROM 1002. When the computer program is executed by processing device 1001, it performs the functions defined in the methods of the embodiments disclosed in this application.

[0150] The noise reduction device provided in this application, employing the blood oxygen saturation detection method described in the above embodiments, can solve the technical problem that pulse wave signals are easily interfered with, affecting the accurate detection of blood oxygen saturation. Compared with the prior art, the beneficial effects of the noise reduction device provided in this application are the same as those of the blood oxygen saturation detection method provided in the above embodiments, and other technical features of this noise reduction device are the same as those disclosed in the previous embodiment method, and will not be repeated here.

[0151] It should be understood that the various parts disclosed in this application can be implemented using hardware, software, firmware, or a combination thereof. In the description of the above embodiments, specific features, structures, materials, or characteristics can be combined in any suitable manner in one or more embodiments or examples.

[0152] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0153] This application provides a computer-readable storage medium having computer-readable program instructions (i.e., a computer program) stored thereon, the computer-readable program instructions being used to execute the blood oxygen saturation detection method in the above embodiments.

[0154] The computer-readable storage medium provided in this application may be, for example, a USB flash drive, but is not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems or devices, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to: electrical connections having one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this embodiment, the computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system or device. The program code contained on the computer-readable storage medium may be transmitted using any suitable medium, including but not limited to: wires, optical cables, RF (Radio Frequency), etc., or any suitable combination thereof.

[0155] The aforementioned computer-readable storage medium may be included in the noise reduction device; or it may exist independently and not assembled into the noise reduction device.

[0156] The aforementioned computer-readable storage medium carries one or more programs. When the one or more programs are executed by the noise reduction device, the noise reduction device: receives an initial pulse wave signal acquired by the photoplethysmography (PPG) detection device; determines whether there is a noise signal in the initial pulse wave signal; if the noise signal is present in the initial pulse wave signal, inputs the initial pulse wave signal into a noise reduction model to obtain a target pulse wave signal, wherein the noise reduction model is trained using historical pulse wave signals as input data and denoised historical pulse wave signals as labels; and sends the target pulse wave signal to the PPG detection device, wherein the PPG detection device performs blood oxygen concentration detection based on the target pulse wave signal.

[0157] Computer program code for performing the operations of this application can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, and C++, and conventional procedural programming languages ​​such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a Local Area Network (LAN) or a Wide Area Network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0158] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0159] The modules described in the embodiments of this application can be implemented in software or hardware. The names of the modules do not necessarily limit the functionality of the unit itself.

[0160] The readable storage medium provided in this application is a computer-readable storage medium that stores computer-readable program instructions (i.e., a computer program) for executing the above-described blood oxygen saturation detection method. This solves the technical problem that pulse wave signals are easily interfered with, affecting the accurate detection of blood oxygen saturation. Compared with the prior art, the beneficial effects of the computer-readable storage medium provided in this application are the same as those of the blood oxygen saturation detection method provided in the above embodiments, and will not be repeated here.

[0161] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the blood oxygen saturation detection method described above.

[0162] The computer program product provided in this application solves the technical problem that pulse wave signals are easily interfered with, affecting the accurate detection of blood oxygen saturation. Compared with the prior art, the beneficial effects of the computer program product provided in this application are the same as those of the blood oxygen saturation detection method provided in the above embodiments, and will not be repeated here.

[0163] The above description is only a part of the embodiments of this application and does not limit the patent scope of this application. All equivalent structural transformations made under the technical concept of this application and using the contents of the specification and drawings of this application, or direct / indirect applications in other related technical fields, are included in the patent protection scope of this application.

Claims

1. A method for detecting blood oxygen saturation, characterized in that, The blood oxygen saturation detection method is applied to a noise reduction device, which is communicatively connected to a photoplethysmography detection device. The blood oxygen saturation detection method includes: Receive the initial pulse wave signal collected by the photoplethysmography detection device, and determine whether there is a noise signal in the initial pulse wave signal; If the initial pulse wave signal contains the noise signal, the initial pulse wave signal is filtered to obtain a filtered pulse wave signal of the target frequency band. The filtered pulse wave signal is then input into a denoising model to obtain the target pulse wave signal. The denoising model is trained using historical pulse wave signals as input data and denoised historical pulse wave signals as labels. The target frequency band is the frequency band corresponding to the range of human heart rate. The target pulse wave signal is sent to the photoplethysmography detection device, wherein the photoplethysmography detection device performs blood oxygen concentration detection based on the target pulse wave signal; The denoising model includes an encoder, a recurrent neural network, an attention mechanism module, a decoder, and a fully connected layer. The step of inputting the filtered pulse wave signal into the denoising model to obtain the target pulse wave signal includes: The filtered pulse wave signal is input into the encoder for processing to obtain a first processing result; The first processing result is input into the recurrent neural network for processing to obtain a second processing result. When processing the first processing result through the recurrent neural network, the recurrent neural network is used to process the previous frame processing result, which is the result obtained by the encoder processing the previous frame signal of the initial pulse wave signal. The second processing result is input into the decoder for further processing to obtain the third processing result; The first processing result and the third processing result are input into the attention mechanism module for processing to obtain the fourth processing result; The third processing result and the fourth processing result are connected, and the result obtained by the connection is input into the fully connected layer for processing to obtain the target pulse wave signal.

2. The blood oxygen saturation detection method as described in claim 1, characterized in that, The attention mechanism module includes a weight calculation block and a weighted processing block. The step of inputting the first processing result and the third processing result into the attention mechanism module for processing to obtain a fourth processing result includes: The first processing result and the third processing result are input into the weight calculation block for processing to obtain the attention weight matrix; The attention weight matrix and the first processing result are input into the weighted processing block for processing to obtain the fourth processing result.

3. The blood oxygen saturation detection method according to any one of claims 1 to 2, characterized in that, The step of determining whether there is noise in the initial pulse wave signal includes: Feature extraction is performed on the initial pulse wave signal to obtain signal waveform features, and the target interference type of the initial pulse wave signal is determined based on the signal waveform features; If the target interference type is a non-physiological interference type, then it is determined that there is a noise signal in the initial pulse wave signal; If the target interference type is a physiological interference type, then it is determined that the noise signal does not exist in the initial pulse wave signal.

4. The blood oxygen saturation detection method as described in claim 3, characterized in that, The signal waveform features include signal variance and baseline offset; the step of determining the target interference type of the initial pulse wave signal based on the signal waveform features includes: If the signal variance is greater than a preset variance threshold and the baseline offset is greater than a preset offset, then the target interference type of the initial pulse wave signal is determined to be physiological interference. If the signal variance is greater than the preset variance threshold and the baseline offset is less than or equal to the preset offset, then the target interference type is determined to be non-physiological interference.

5. A blood oxygen saturation detection device, characterized in that, The blood oxygen saturation detection device is used in a noise reduction device, which is communicatively connected to a photoplethysmography detection device. The blood oxygen saturation detection device includes: The judgment module is used to receive the initial pulse wave signal collected by the photoplethysmography detection device and determine whether there is a noise signal in the initial pulse wave signal. The noise reduction module is used to filter the initial pulse wave signal to obtain a filtered pulse wave signal of the target frequency band if the initial pulse wave signal contains noise signal, and input the filtered pulse wave signal into the noise reduction model to obtain the target pulse wave signal. The noise reduction model is trained using historical pulse wave signals as input data and the denoised historical pulse wave signals as labels. The target frequency band is the frequency band corresponding to the range of human heart rate. A transmitting module is used to transmit the target pulse wave signal to the photoplethysmography detection device, wherein the photoplethysmography detection device performs blood oxygen concentration detection based on the target pulse wave signal; The noise reduction model includes an encoder, a recurrent neural network, an attention mechanism module, a decoder, and a fully connected layer. The noise reduction module is further configured to: input the filtered pulse wave signal into the encoder for processing to obtain a first processing result; input the first processing result into the recurrent neural network for processing to obtain a second processing result, wherein, when processing the first processing result through the recurrent neural network, the recurrent neural network is used to process the previous frame processing result, the previous frame processing result being the result obtained by the encoder processing the previous frame signal of the initial pulse wave signal; input the second processing result into the decoder for processing to obtain a third processing result; input the first processing result and the third processing result into the attention mechanism module for processing to obtain a fourth processing result; connect the third processing result and the fourth processing result, and input the connected result into the fully connected layer for processing to obtain the target pulse wave signal.

6. A noise reduction device, characterized in that, The noise reduction device includes: a memory, a processor, and a blood oxygen saturation detection program stored in the memory and executable on the processor. When the blood oxygen saturation detection program is executed by the processor, it implements the steps of the blood oxygen saturation detection method as described in any one of claims 1 to 4.

7. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a blood oxygen saturation detection program, which, when executed by a processor, implements the steps of the blood oxygen saturation detection method as described in any one of claims 1 to 4.

8. A computer program product, characterized in that, The computer program product includes a computer program that, when executed by a processor, implements the steps of the blood oxygen saturation detection method as described in any one of claims 1 to 4.

Images