A physiological signal acquisition method, system and wearable device

By using the serum electrical acquisition module and physiological signal acquisition module in wearable devices, combining preset serum electrical threshold and environmental monitoring data, the neural network model is used to optimize serum electrical threshold prediction, which solves the problem of equipment wear status detection and improves physiological signals Accuracy and calculation speed of acquisition.

CN117257270BActive Publication Date: 2024-11-08KINGFAR INTERNATIONAL INC
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202311402317.3
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2023-10-26
Publication Date
2024-11-08
Estimated Expiration
2043-10-26

AI Technical Summary

Technical Problem

Existing wearable devices are susceptible to environmental noise interference during physiological signal acquisition, resulting in inaccurate signal acquisition and difficulty in accurately detecting wearable state, resulting in reduced data stability and calculation speed.

Method used

By setting up a skin-electric acquisition module and a physiological signal acquisition module in the wearable device, preset the skin-electric threshold to determine whether the device is wearing normally. If it is abnormal, the acquisition will be stopped, clear the signal filter and zero, and adjust the threshold using environmental monitoring data. Optimize threshold prediction in combination with neural network models.

Benefits of technology

It improves the accuracy and calculation speed of physiological signal acquisition, reduces environmental noise interference, shortens data recovery time, and enhances the accuracy of device-skin contact judgment.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN117257270B_ABST
    Figure CN117257270B_ABST
Patent Text Reader

Abstract

The present application relates to a physiological signal acquisition method, system and wearable device, which belongs to the field of physiological signal acquisition technology and is applied to a wearable device equipped with a skin electricity acquisition module and a physiological signal acquisition module. The acquisition method includes: obtaining the skin electricity signal value acquired by the skin electricity acquisition module; determining the current wearing state of the wearable device according to the preset skin electricity threshold and the skin electricity signal value; in response to the wearing state being an abnormal state, controlling the physiological signal acquisition module to stop collecting physiological signals. The present application can effectively monitor and control the wearing condition of the wearable device by quickly judging and identifying the wearing condition, reducing the mixing of environmental noise into the sampled signal, and thus improving the accuracy of the collected signal.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the technical field of physiological signal acquisition, and in particular to a physiological signal acquisition method, system and wearable device. Background Art

[0002] With the continuous development of digital signal processing technology, the collection of human physiological signals has gradually shifted from mechanization to electronics, and is constantly developing towards automation and intelligence.

[0003] Physiological signal acquisition refers to the use of sensors and other equipment to detect and measure the physiological parameters of the human body. By continuously monitoring and analyzing the physiological parameters of the human body in real time, it can provide personalized health monitoring and management, promote health, prevent diseases, and improve exercise and rehabilitation effects. It has important significance and applications in the fields of medical care, health monitoring, sports training, etc.

[0004] At present, common wearable devices such as wristbands have many factors that interfere with signal collection when collecting physiological signals based on wristbands, making the collected signals inaccurate. Summary of the invention

[0005] In order to improve the accuracy of the sampled signal, the present application provides a physiological signal acquisition method, system and wearable device.

[0006] In a first aspect, the present application provides a physiological signal acquisition method, which adopts the following technical solution:

[0007] A physiological signal acquisition method is applied to a wearable device equipped with a skin electrical acquisition module and a physiological signal acquisition module, and the acquisition method comprises:

[0008] Obtain the skin electrical signal value collected by the skin electrical collection module;

[0009] Determining a current wearing state of the wearable device according to a preset skin electrical threshold and the skin electrical signal value;

[0010] In response to the wearing state being an abnormal state, the physiological signal acquisition module is controlled to stop acquiring physiological signals.

[0011] By adopting the above technical solution, since the numerical value of normal human skin electricity is relatively stable, usually between 0.01µSiemens and 100µSiemens; therefore, by pre-setting the skin electricity threshold, it is judged whether the wearable device is worn normally according to the collected skin electricity signal value, that is, whether the wearable device has been in good contact with the skin. When the wearing is abnormal, it is necessary to stop the collection of physiological signals immediately to reduce the mixing of environmental noise in the sampled signal. By quickly judging and identifying the wearing status, the wearing status of the wearable device can be effectively monitored and controlled, thereby improving the accuracy of the collected signal.

[0012] Optionally, determining the current wearing state of the wearable device according to a preset skin electrical threshold and the skin electrical signal value includes:

[0013] Comparing the skin electrical signal value with the preset skin electrical threshold;

[0014] In response to the electrical skin signal value being less than a preset electrical skin threshold, it is determined that the current wearing state of the wearable device is an abnormal state.

[0015] By adopting the above technical solution, the collected skin electrodermal signal value is compared with the preset skin electrodermal threshold value, so as to determine the wearing state of the wearable device, thereby realizing rapid judgment and identification of the wearing state. By effectively monitoring and controlling the wearing condition of the wearable device, the accuracy of physiological signal collection can be improved.

[0016] Optionally, determining the wearing state of the wearable device according to a preset skin electrical threshold and the skin electrical signal value further includes: in response to the skin electrical signal value being greater than or equal to the preset skin electrical threshold, determining that the current wearing state of the wearable device is a non-abnormal state;

[0017] The method further includes: in response to the wearing state of the wearable device being a non-abnormal state, controlling the physiological signal acquisition module to start or keep acquiring physiological signals.

[0018] By adopting the above technical solution, when no abnormality is detected in wearing, the physiological signal collection mode can be turned on or maintained.

[0019] Optionally, in response to the wearing state being an abnormal state, the method further includes: clearing a signal filter signal of the wearable device, and / or setting a sampling value of a physiological signal to zero.

[0020] By adopting the above technical solution, when an abnormal wearing state is detected, the signal filter is cleared and the collected data can be reset to zero. Since the signal filter has been cleared before each collection, abnormal jumps are significantly reduced, and the time it takes for the PPG signal to return to a reasonable state is also greatly shortened, thereby significantly improving the calculation speed and accuracy.

[0021] Optionally, the step of determining the preset skin electrical threshold includes:

[0022] Obtaining environmental monitoring data of the environment where the wearable device is located;

[0023] The preset skin conductance threshold is determined or adjusted according to the environmental monitoring data.

[0024] By adopting the above technical solution, by monitoring the changes in environmental monitoring data such as temperature and humidity, and adjusting the skin electrical threshold accordingly according to the high and low environmental monitoring data such as temperature and humidity, the accuracy and stability of the data can be ensured to adapt to different environmental conditions, and at the same time, the contact between the device and the skin can be judged more accurately.

[0025] Optionally, determining or adjusting the preset skin electricity threshold according to the environmental monitoring data includes: determining the skin electricity threshold corresponding to the environmental monitoring data as the preset skin electricity threshold according to a mapping relationship between the pre-acquired environmental monitoring data and the skin electricity threshold.

[0026] By adopting the above-mentioned technical solution, after monitoring and obtaining the environmental monitoring data, the corresponding preset skin electrical threshold can be determined according to the pre-set mapping relationship between the environmental monitoring data and the skin electrical threshold; specifically, the mapping relationship can be determined by collecting the experience and opinions of experts in related fields and conducting statistical analysis based on historical monitoring data. This mapping relationship can be used as a basis for setting the preset skin electrical threshold.

[0027] Optionally, determining or adjusting the preset skin charge threshold based on the environmental monitoring data includes: inputting the environmental monitoring data into a neural network model for skin charge threshold prediction processing, the skin charge threshold prediction value output by the neural network model is the preset skin charge threshold; wherein the neural network model is pre-supervised trained with a training set, and the training set includes historical environmental monitoring data and corresponding historical skin charge monitoring data.

[0028] By adopting the above technical solution, according to the individual's physiological characteristics and environmental conditions, by monitoring the changes in environmental monitoring data such as temperature and humidity, and inputting the environmental monitoring data such as temperature and humidity into a pre-trained neural network model, the skin electrode threshold is adjusted accordingly according to the temperature and humidity based on the model results, thereby ensuring the accuracy and stability of the data to adapt to different environmental conditions, and at the same time being able to more accurately judge the contact between the device and the skin.

[0029] Optionally, the training of the neural network model includes:

[0030] Get sample data set and preprocess it;

[0031] Perform feature extraction on the preprocessed sample data set;

[0032] Dividing the sample data set after feature extraction into the training set, the validation set and the test set; wherein the validation set and the test set include the environmental monitoring data of the future time period and the corresponding skin electrical monitoring data;

[0033] The neural network model is initially trained based on the training set, and the neural network model after the initial training is verified and model parameters are optimized and adjusted based on the verification set and the test set.

[0034] By adopting the above technical solution, using historical environmental monitoring data and the corresponding historical skin charge monitoring data as training sets to train the neural network model, it is helpful for the model to capture and learn the complex nonlinear relationship between environmental monitoring data and skin charge monitoring data. By using environmental monitoring data in future time periods and the corresponding skin charge monitoring data as verification sets and test sets, the parameters of the algorithm model are optimized and corrected to obtain the optimal skin charge threshold prediction model.

[0035] Optionally, performing initial training on the neural network model based on the training set, and performing verification and model parameter optimization adjustment on the initially trained neural network model based on the verification set and the test set, including:

[0036] Inputting the training set into a pre-built neural network model for training, optimizing the model parameters and calculating the loss function of the neural network model, until the loss function meets a preset condition or the number of iterations of the pre-built neural network model reaches a preset number, thereby obtaining the trained electrodermal threshold prediction model;

[0037] Verifying the skin electrode threshold prediction model based on the verification set, evaluating the performance of the skin electrode threshold prediction model and adjusting the hyperparameters of the model;

[0038] The predictive ability of the adjusted skin conductance threshold prediction model is tested based on the test set.

[0039] By adopting the above technical solution, the model parameters are iteratively updated based on the training set to gradually optimize the model's predictive ability, and the model performance is evaluated and the hyperparameters are adjusted based on the validation set. When the model performs well on the validation set, the test set is used to evaluate the final performance of the model, and the actual predictive ability of the prediction model is comprehensively determined by combining multiple test indicators, thereby further improving the accuracy of the prediction results.

[0040] In a second aspect, the present application provides a physiological signal acquisition system for executing the physiological signal acquisition method described in the first aspect, using the following technical solution:

[0041] A physiological signal acquisition system, used to execute the physiological signal acquisition method described in the first aspect, is applied to a wearable device equipped with a skin electrical acquisition module and a physiological signal acquisition module, and the acquisition system includes:

[0042] The skin electrical signal value acquisition module is used to obtain the skin electrical signal value collected by the skin electrical acquisition module;

[0043] A wearing state determination module, used to determine the current wearing state of the wearable device according to a preset skin electrical threshold and the skin electrical signal value;

[0044] The collection control module is used to control the physiological signal collection module to stop collecting physiological signals in response to the wearing state being an abnormal state.

[0045] By adopting the above technical solution, according to the preset skin electrical threshold, the skin electrical signal value collected in real time is used to judge whether the wearable device is worn normally, that is, to judge whether the wearable device has been in good contact with the skin. When the wearing is abnormal, it is necessary to stop the collection of physiological signals immediately to reduce the mixing of environmental noise in the sampled signal, thereby improving the accuracy of the collected signal.

[0046] In a third aspect, the present application provides a wearable device.

[0047] A wearable device comprises: a device body, an galvanic skin acquisition module, a physiological signal acquisition module and a physiological signal acquisition system as described in the second aspect, wherein the galvanic skin acquisition module and the physiological signal acquisition module are arranged on the device body, and the physiological signal acquisition system is respectively connected to the galvanic skin acquisition module and the physiological signal acquisition module, the galvanic skin acquisition module is used to collect the galvanic skin signal value of the user, and the physiological signal acquisition module is used to collect the physiological signal data of the user.

[0048] To summarize, the present application includes at least one of the following beneficial technical effects: by pre-setting the skin electricity threshold, judging whether the wearable device is worn normally according to the collected skin electricity signal value, that is, judging whether the wearable device has been in good contact with the skin. When it is worn normally, the physiological signal collection mode can be turned on; when it is worn abnormally, it is necessary to stop the collection of physiological signals immediately. By quickly judging and identifying the wearing status, the wearing condition of the wearable device can be effectively monitored and controlled to reduce the mixing of environmental noise in the sampled signal, thereby improving the accuracy of the collected signal. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 It is a first flow chart of a physiological signal acquisition method according to one embodiment of the present application.

[0050] Figure 2 It is a second flow chart of the physiological signal acquisition method according to one embodiment of the present application.

[0051] Figure 3 It is a third flow chart of the physiological signal acquisition method according to one embodiment of the present application.

[0052] Figure 4 It is a fourth flow chart of a physiological signal collection method according to one embodiment of the present application.

[0053] Figure 5 This is a fifth flow chart of a physiological signal collection method according to one embodiment of the present application.

[0054] Figure 6 It is a sixth flow chart of a physiological signal acquisition method according to one embodiment of the present application.

[0055] Figure 7 This is a test effect diagram of the physiological signal acquisition method of the present application during actual application.

[0056] Figure 8 It is a structural block diagram of a physiological signal acquisition system according to one embodiment of the present application.

[0057] Fig. 9 It is a structural schematic diagram of a wearable device according to one embodiment of the present application.

[0058] Explanation of the reference numerals: 1. Device body; 11. Skin electrical acquisition module; 12. Physiological signal acquisition module; 101. Skin electrical signal value acquisition module; 102. Wearing state determination module; 103. Acquisition control module. DETAILED DESCRIPTION

[0059] In order to make the purpose, technical solutions and advantages of this application more clear, the following Figure 1-9 It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0060] Currently, when users use wearable devices such as bracelets to collect physiological signals, when the bracelet is taken off the wrist, the optical signal transmitter is still switching and collecting according to the set frequency, which may introduce unnecessary ambient light, motion artifacts and other environmental noise; when the bracelet is worn normally again, it may be affected by the environmental noise in the filter before wearing, which may cause the time for the data to stabilize to a reasonable state to be prolonged, thereby reducing the calculation speed of the physiological signal.

[0061] It can be seen that the wearing detection accuracy and reliability of wearable physiological acquisition devices are the key to their application. However, most common wearable devices cannot accurately distinguish whether the current device is suitable for wearing. For example, the PPG value monitoring method used in some devices may cause misjudgment of wearing when an object blocks the receiver on the device or the receiver is placed on the table.

[0062] Therefore, how to accurately detect whether wearable devices are worn properly and improve the accuracy of physiological signal acquisition results is an urgent problem to be solved.

[0063] The embodiment of the present application discloses a method for collecting physiological signals.

[0064] Reference Figure 1 A physiological signal acquisition method is applied to a wearable device equipped with a skin electrical acquisition module and a physiological signal acquisition module, and the acquisition method includes:

[0065] Step S101, obtaining the skin electrical signal value collected by the skin electrical collection module;

[0066] Step S102, determining the current wearing state of the wearable device according to the preset skin electrical threshold and the skin electrical signal value;

[0067] Specifically, the current wearing state includes a normal wearing state and an abnormal wearing state. The normal wearing state means that the wearable device has been in good contact with the user's skin and can start collecting physiological signals; the abnormal wearing state means that the user may have an inappropriate wearing position or tightness, which may cause environmental noise to be mixed into the sampled signal.

[0068] Step S103, in response to the wearing state being an abnormal state, controlling the physiological signal acquisition module to stop acquiring physiological signals.

[0069] In the above implementation, since the value of normal human skin electricity is relatively stable, usually between 0.01µSiemens and 100µSiemens; therefore, by pre-setting the skin electricity threshold, it is judged whether the wearable device is worn normally according to the collected skin electricity signal value, that is, whether the wearable device has been in good contact with the skin. When the wearing is abnormal, it is necessary to immediately stop the collection of physiological signals to reduce the mixing of environmental noise in the sampled signal. By quickly judging and identifying the wearing status, the wearing condition of the wearable device can be effectively monitored and controlled, thereby improving the accuracy of the collected signal.

[0070] As a further implementation of step S103, after responding to the wearing state being an abnormal state in step S103, the step further includes:

[0071] Step S1031, clearing the signal filter of the wearable device; and / or, step S1032, setting the sampling value of the physiological signal to zero.

[0072] Among them, when an abnormal wearing state is detected, the signal filter is cleared on the basis of stopping the collection of physiological signal data, and the collected data can also be reset to zero. Since the signal filter has been cleared before each collection, the abnormal jump is significantly reduced, and the time for the PPG signal to return to a reasonable state is also greatly shortened, thereby significantly improving the calculation speed and accuracy.

[0073] It should be noted that after determining that the wearing state is an abnormal state, step S1031 and / or step S1031 can be executed, and when both need to be executed, they can be executed simultaneously, or they can be executed sequentially in no particular order.

[0074] Reference Figure 2 As an implementation of determining the current wearing state of the wearable device according to the preset skin electrical threshold and the skin electrical signal value in step S102, specifically includes:

[0075] Step S201, comparing the skin electrical signal value with a preset skin electrical threshold;

[0076] Step S202: In response to the skin electrical signal value being less than a preset skin electrical threshold, determining that the current wearing state of the wearable device is an abnormal state.

[0077] In the above implementation, by comparing the collected skin electrodermal signal value with the preset skin electrodermal threshold, the wearing state of the wearable device can be determined, thereby achieving rapid judgment and identification of the wearing state. By effectively monitoring and controlling the wearing condition of the wearable device, the accuracy of physiological signal collection can be improved.

[0078] Reference Figure 2 As a further implementation of the physiological signal acquisition method, after step S201, it further includes:

[0079] Step S301, in response to the skin electrical signal value being greater than or equal to the preset skin electrical threshold, determining that the current wearing state of the wearable device is a non-abnormal state;

[0080] Step S302: In response to the wearing state of the wearable device being a non-abnormal state, controlling the physiological signal acquisition module to start or keep acquiring physiological signals.

[0081] Among them, starting or maintaining the collection of physiological signals means controlling the physiological signal collection module to start or maintain the physiological signal collection mode, for example, starting or maintaining the light collection of PPG signals, blood oxygen signals, etc.

[0082] In the above implementation, when no abnormality is detected during wearing, the physiological signal collection mode can be started or maintained.

[0083] It is understandable that since the filter has been cleared at this time and the normal acquisition mode is turned on, the impact of environmental noise in the filter before wearing is reduced, the time for the data to recover to a reasonable state is shortened, and the calculation speed and accuracy of physiological signals such as blood oxygen and heart rate are improved.

[0084] Reference Figure 3 As an implementation of the physiological signal acquisition method, the step of determining a preset skin electrical threshold includes:

[0085] Step S401, obtaining environmental monitoring data of the environment where the wearable device is located;

[0086] Among them, environmental monitoring data can include temperature data and humidity data, which can be collected in real time through temperature and humidity sensors;

[0087] Step S402, determining or adjusting a preset skin conductance threshold according to environmental monitoring data.

[0088] In the above implementation, by monitoring the changes in environmental monitoring data such as temperature and humidity, and adjusting the skin electrode threshold accordingly according to the high and low environmental monitoring data such as temperature and humidity, the accuracy and stability of the data can be ensured to adapt to different environmental conditions, and at the same time, the contact between the device and the skin can be judged more accurately.

[0089] It is understandable that by measuring the temperature changes in the environment, we can understand the effect of temperature on skin conductivity. Since the increase in temperature will cause the skin's blood vessels to dilate, blood flow will accelerate, and the skin's conductivity will increase accordingly, the value of the skin electrical data will also increase. Similarly, changes in environmental humidity will also affect the conductivity of the skin. When the environmental humidity is high, the moisture content on the skin surface increases, and the conductivity will also increase accordingly. In the face of this situation, using a fixed skin electrical threshold may affect the accuracy of wearing detection.

[0090] Therefore, as one implementation of step S402, the step of determining or adjusting the preset skin electrical threshold according to the environmental monitoring data includes:

[0091] Step S4021, according to the mapping relationship between the pre-acquired environmental monitoring data and the skin electrical threshold, determine that the skin electrical threshold corresponding to the environmental monitoring data is the preset skin electrical threshold.

[0092] In the above implementation, after the environmental monitoring data is monitored, the corresponding preset skin electrical threshold can be determined based on the pre-set mapping relationship between the environmental monitoring data and the skin electrical threshold; specifically, the mapping relationship can be determined by collecting the experience and opinions of experts in related fields and conducting statistical analysis based on historical monitoring data. This mapping relationship can be used as a basis for setting the preset skin electrical threshold.

[0093] In addition, as another implementation of step S402, determining or adjusting the preset skin electrical threshold according to the environmental monitoring data includes:

[0094] Step S4022, input the environmental monitoring data into the neural network model for skin conduction threshold prediction processing, and the skin conduction threshold prediction value output by the neural network model is the preset skin conduction threshold; wherein the neural network model is pre-supervised trained with a training set, and the training set includes historical environmental monitoring data and corresponding historical skin conduction monitoring data.

[0095] In the above implementation, according to the individual's physiological characteristics and environmental conditions, by monitoring the changes in environmental monitoring data such as temperature and humidity, and inputting the environmental monitoring data such as temperature and humidity into a pre-trained neural network model, the skin electrode threshold is adjusted accordingly according to the temperature and humidity based on the model results, thereby ensuring the accuracy and stability of the data to adapt to different environmental conditions, and at the same time being able to more accurately judge the contact between the device and the skin.

[0096] Reference Figure 4 , as an implementation of the neural network model training in step S4022, specifically includes:

[0097] Step S501, obtaining a sample data set and performing preprocessing;

[0098] Specifically, the sample data set includes a large number of samples and diverse data, such as historical environmental monitoring data and corresponding skin electricity monitoring data, and also, for example, environmental monitoring data and corresponding skin electricity monitoring data for future time periods relative to the historical data; wherein the environmental monitoring data includes temperature data and humidity data, and the skin electricity monitoring data includes corresponding skin electricity signal values, and sensors or data recorders can be used for real-time data collection to ensure that the data has a timestamp to capture time dependencies.

[0099] In one embodiment of the present application, the step of data preprocessing includes data cleaning, such as processing missing values, outliers and duplicate values, etc., which can be done by interpolation, smoothing and denoising.

[0100] Step S502, extracting features from the preprocessed sample data set;

[0101] Specifically, the features related to the threshold of the electrodermal signal can be determined through domain knowledge. In the embodiment of the present application, feature extraction includes extracting time features, statistical features, hysteresis features, etc. Specifically, time features include time information such as year, month, day, hour, etc., statistical features include statistical information such as mean, variance, maximum value, minimum value, trend, etc., and hysteresis features include extracting environmental monitoring data at past time points to capture the autocorrelation of the time series.

[0102] Step S503, dividing the sample data set after feature extraction into a training set, a validation set and a test set; wherein the validation set and the test set include environmental monitoring data and corresponding skin electrical monitoring data in the future time period;

[0103] In one embodiment of the present application, 70-80% of the data may be divided into a training set, 10-15% of the data may be divided into a validation set, and 10-15% of the data may be divided into a test set;

[0104] It should be noted that for time series data, the validation set and test set should contain data from future time periods to simulate actual usage. Time series cross-validation should be performed to ensure the generalization performance of the model for future data.

[0105] Step S504, performing initial training on the neural network model based on the training set, and verifying and optimizing the model parameters of the initially trained neural network model based on the verification set and the test set.

[0106] In the above implementation, historical environmental monitoring data and corresponding historical skin charge monitoring data are used as training sets to train the neural network model, which helps the model capture and learn the complex nonlinear relationship between environmental monitoring data and skin charge monitoring data. Environmental monitoring data in future time periods and corresponding skin charge monitoring data are used as validation sets and test sets to optimize and correct the parameters of the algorithm model, thereby obtaining the optimal skin charge threshold prediction model.

[0107] Reference Figure 5 As an implementation method of step S504, the steps of initially training the neural network model based on the training set, and verifying the initially trained neural network model and optimizing the model parameters based on the verification set and the test set specifically include:

[0108] Step S601, inputting the training set into a pre-built neural network model for training, optimizing the model parameters and calculating the loss function of the neural network model, until the loss function meets a preset condition or the number of iterations of the pre-built neural network model reaches a preset number, and obtaining a trained skin electrodermal threshold prediction model;

[0109] In one embodiment of the present application, the neural network model can be trained by the Adam optimization algorithm and the mean square error (MSE) or cross-entropy loss function. The neural network model can capture the complex nonlinear relationship between the environmental monitoring data and the skin electrical signal, and can easily process various types of input data, including continuous value features such as temperature and humidity, and discrete classification of skin electrical signals;

[0110] Specifically, Adam is an optimization algorithm with adaptive learning rate, which has the advantages of fast convergence and good performance. It can adjust the learning rate according to the historical information of the gradient of each parameter, thereby effectively optimizing the model parameters.

[0111] Among them, the formula of the mean square error loss function is: MSE=(1 / n)*Σ(yi-Y_pred)², where n represents the number of training samples, yi represents the true value, and Y_pred represents the model prediction value. The square of the difference between the model prediction and the actual skin electrode signal is calculated to better fit the target value; the formula of the cross entropy loss function is: L=-∑(yi*log(y_pred)), where L represents the loss function, yi represents the true value, y_pred represents the label value predicted by the model, and log represents the natural logarithm function. The cross entropy loss function is used to measure the difference between the classification prediction of the model and the actual category, so as to enable the model to better adapt to the classification problem.

[0112] Step S602, verifying the skin electrical threshold prediction model based on the validation set, evaluating the performance of the skin electrical threshold prediction model and adjusting the hyperparameters of the model;

[0113] Specifically, the validation set is used to adjust the hyperparameters of the skin electrodermal threshold prediction model, such as the number of layers, number of neurons, and learning rate, in order to understand the generalization performance of the model on unseen data and adjust the model parameters. During the model tuning process, an early stopping strategy can be used to avoid overfitting.

[0114] Step S603, testing the prediction ability of the adjusted skin conductance threshold prediction model based on the test set;

[0115] Among them, the test indicators include but are not limited to root mean square error (RMSE), mean absolute error (MAE) and coefficient of determination (R-squared).

[0116] In the above implementation, the model parameters are iteratively updated based on the training set to gradually optimize the model's predictive ability, the model performance is evaluated based on the validation set and the hyperparameters are adjusted. When the model performs well on the validation set, the test set is used to evaluate the final performance of the model, and the actual predictive ability of the prediction model is comprehensively determined by combining multiple test indicators, thereby further improving the accuracy of the prediction results.

[0117] In other embodiments of the present application, after the model test is completed, a chart of the actual observed values ​​and the model predicted values ​​can be drawn, and the evaluation results can be visualized to intuitively compare the consistency between the model predictions and the actual observed values; when the model meets the performance requirements, the trained model can be deployed in the actual environment for real-time electrodermal signal threshold prediction; at the same time, an API or interface can be established to provide other systems with an interface for interacting with the model so as to receive input data and return prediction results.

[0118] Furthermore, during the use of the prediction model, the performance of the model can be tested regularly. If model drift or performance degradation occurs, the model may need to be retrained to improve the prediction accuracy by using the latest data. At the same time, the model can be updated according to new environmental conditions or changes in demand to adapt to new prediction tasks. By introducing a feedback mechanism, the prediction performance of the model can be improved based on actual feedback, and the accuracy and stability of the model can be continuously optimized.

[0119] In addition, human experience can be used to judge and take appropriate measures based on the threshold of the skin electrodermal signal output by the prediction model. For example, the model parameters can be adjusted according to the output results of the prediction model to keep the model in an ideal state; and when it is judged that the output results of the prediction model exceed the preset threshold range, an alarm can be issued based on this information to alert the operator or relevant personnel and take appropriate measures to deal with the abnormal situation in a timely manner.

[0120] Reference Figure 6 As an implementation of step S601, the training set is input into a pre-built neural network model for training, and the step of optimizing the model parameters includes:

[0121] Step S701, initializing model parameters; the model parameters include a learning rate α, an exponential decay rate β1 of a first-order moment estimate, an exponential decay rate β2 of a second-order moment estimate, and an initial value of a skin electrical threshold value Y;

[0122] In one of the embodiments of the present application, the learning rate α=0.001, the exponential decay rate of the first-order moment estimation β1=0.9, the exponential decay rate of the second-order moment estimation β2=0.999, and the initial value of the skin electrode threshold Y is set to 0. After the initialization is completed, the iterative update of the parameters can be started; it should be noted that the configuration and selection of each parameter can be adjusted according to the actual situation.

[0123] Step S702, inputting a plurality of training samples in the training set into a pre-built neural network model, and calculating a skin charge threshold prediction value of each training sample; wherein the training sample includes sample environment monitoring data and corresponding sample skin charge monitoring data, and the skin charge threshold prediction value is calculated according to a prediction function of the neural network model;

[0124] Specifically, the predicted value Y_pred of training sample i is obtained by formula (1):

[0125] Y_pred=f(E_i,T_i) (1)

[0126] Where f is the prediction function of the model, E_i is the error between the real-time skin electrode threshold and the predicted skin electrode threshold, and T_i is the specific time point.

[0127] Step S703, calculating the partial derivative dL / dY_pred of the skin electrical threshold prediction value Y_pred and the partial derivative dL / dY_pred*df / dY of the sample skin electrical monitoring data;

[0128] Among them, the partial derivative dL / dY_pred represents the impact of the predicted value on the loss function, and the partial derivative of the skin electrode threshold dL / dY_pred*df / d represents the impact of the parameter update on the loss function;

[0129] It can be understood that by calculating the partial derivatives of the loss function with respect to the model parameters, the gradient can be obtained, which represents the rate of change of the model parameters in each direction; the parameters of the neural network model can be adaptively adjusted and iteratively updated through optimization algorithms (such as the gradient descent method), and the update direction and step size of the parameters can be determined based on the calculated gradient to minimize the loss function.

[0130] Step S704: based on the partial derivative of the sample skin electrical monitoring data, update the first-order moment estimate m and the second-order moment estimate v of each model parameter:

[0131] Specifically, the first-order moment estimate m and the second-order moment estimate v of each model parameter are updated by formulas (2) and (3):

[0132] m=β1*m+(1-β1)*dL / dY_pred*df / dY (2),

[0133] v=β2*v+(1-β2)*(dL / dY_pred*df / dY)^2 (3);

[0134] Among them, the above estimated values ​​can be used to calculate the update direction and step size of the parameters in order to better optimize the model;

[0135] Step S705, performing deviation correction on the first-order moment estimate m and the second-order moment estimate v of each model parameter to obtain a corrected first-order moment estimate m_hat and a corrected second-order moment estimate v_hat;

[0136] Specifically, the first-order moment estimate m and the second-order moment estimate v of each model parameter are corrected for deviations by using formulas (4) and (5) to obtain the corrected first-order moment estimate m_hat and the corrected second-order moment estimate v_hat:

[0137] m_hat=m / (1-β1^t) (4)

[0138] v_hat=v / (1-β2^t) (5),

[0139] In the formula, t is the number of model iterations;

[0140] Among them, the accuracy of parameter updates is improved by reducing the deviation of estimated values ​​in the early stages of model training.

[0141] In the above implementation, in order to ensure that the model has a suitable initial state before starting the iterative update, the model parameters are initialized for better parameter optimization, and the partial derivatives are calculated to determine the impact of the loss function on the predicted value and the skin electrode threshold, so as to provide gradient information for subsequent parameter updates; by updating the first-order moment estimate and the second-order moment estimate, the parameter update direction and step size can be adjusted according to the historical gradient information to converge to the optimal solution more quickly; then, the deviation of the estimated value is reduced through deviation correction to improve the accuracy and stability of the parameter update; finally, the skin electrode threshold is iteratively updated to gradually optimize the prediction ability of the model, so that it can better adapt to the training data and improve the prediction accuracy.

[0142] Further, as an implementation method of obtaining the predicted value of the skin electrical threshold according to the neural network model, it specifically includes:

[0143] Based on the optimized learning rate α, the initial value of the skin electrical threshold Y, the corrected first-order moment estimate m_hat and the corrected second-order moment estimate v_hat, the skin electrical threshold Y is iteratively updated:

[0144] Y=Y-α*m_hat / (√v_hat+ε) (6),

[0145] Where ε is a constant, m_hat is the corrected first-order moment estimate, and v_hat is the corrected second-order moment estimate.

[0146] In an embodiment of the present application, the update of the skin electrode threshold Y can be achieved by optimizing the loss function. The mean square error (MSE) or cross-entropy loss function can be used to train the model. By iteratively updating the skin electrode threshold Y, we can minimize the loss function, thereby gradually optimizing the prediction ability of the model, making it better adapt to the training data and improving the prediction accuracy.

[0147] It is understandable that the skin electrical threshold Y is part of the model output, representing the model's response to the input data E_i and T_i. The update of the skin electrical threshold Y will directly affect the model's prediction results, thereby affecting the model's performance. By iteratively updating the skin electrical threshold Y, the model can better adapt to changes in environmental monitoring data and skin electrical signals, and gradually improve the model's performance by optimizing the loss function.

[0148] This application proposes a physiological signal acquisition method, which has improved accuracy, calculation speed and power saving. This method has been applied to relevant wearable devices and has been verified in relevant tests. Figure 7The figure shows the test effect diagram of the acquisition method of the present application in actual application. It can be clearly seen that after using the acquisition method of the present application, since the signal filter has been cleared before each acquisition, the abnormal jump is significantly reduced, and the time for the PPG signal to return to a reasonable state is also greatly shortened, thereby significantly improving the calculation speed and accuracy.

[0149] The embodiment of the present application also discloses a physiological signal acquisition system.

[0150] Reference Figure 8 A physiological signal acquisition system is used to execute the above-mentioned physiological signal acquisition method, and is applied to a wearable device equipped with a skin electricity acquisition module and a physiological signal acquisition module. The acquisition system includes:

[0151] The skin electrical signal value acquisition module 101 is used to acquire the skin electrical signal value collected by the skin electrical acquisition module;

[0152] A wearing state determination module 102, configured to determine a current wearing state of the wearable device according to a preset skin electrical threshold and a skin electrical signal value;

[0153] The collection control module 103 is used to control the physiological signal collection module to stop collecting physiological signals in response to the wearing state being an abnormal state.

[0154] In the above implementation, according to the preset skin electricity threshold, the skin electricity signal value collected in real time is used to determine whether the wearable device is worn normally, that is, to determine whether the wearable device has been in good contact with the skin. When it is worn normally, the collection of physiological signals can be started; when it is worn abnormally, the collection of physiological signals needs to be stopped immediately, which reduces the system's useless work and equipment loss, and at the same time reduces the situation of environmental noise mixed in the sampled signal, thereby improving the accuracy of the collected signal.

[0155] The physiological signal acquisition system of the embodiment of the present application can implement any of the above-mentioned physiological signal acquisition methods, and the specific working process of each module in the physiological signal acquisition system can refer to the corresponding process in the above-mentioned method embodiment.

[0156] The embodiment of the present application also discloses a wearable device.

[0157] Reference Fig. 9A wearable device includes: a device body 1, a skin electricity acquisition module 11, a physiological signal acquisition module 12 and the above-mentioned physiological signal acquisition system, the skin electricity acquisition module 11 and the physiological signal acquisition module 12 are arranged on the device body 1, and the physiological signal acquisition system is respectively connected to the skin electricity acquisition module 11 and the physiological signal acquisition module 12, the skin electricity acquisition module 11 is used to collect the skin electricity signal value of the user, and the physiological signal acquisition module 12 is used to collect the physiological signal data of the user, such as PPG signal, blood oxygen signal, etc.

[0158] In one of the embodiments of the present application, the skin electricity acquisition module 11 can be two skin electricity electrodes, and the two skin electricity electrodes and the physiological signal acquisition module 12 are all arranged on the side of the device body 1 that contacts the user's skin, and the two skin electricity electrodes are respectively arranged on both sides of the physiological signal acquisition module 12; specifically, the distance between the skin electricity electrodes can be set to 15mm, and the diameter of the raised part of the skin electricity electrode can be 30mm. Since the normal human wrist width is between 5-7cm, this size can ensure that the bracelet fits the wrist better, thereby providing a better wearing experience and accurate skin electricity signal acquisition effect.

[0159] In the several embodiments provided in this application, it should be understood that the provided methods and systems can be implemented in other ways. For example, the system embodiments described above are only illustrative; for example, the division of a certain module is only a logical function division, and there may be other division methods in actual implementation, such as multiple modules can be combined or integrated into another system, or some features can be ignored or not executed.

[0160] It should be noted that in the above embodiments, the description of each embodiment has different emphases, and for parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0161] The above are all preferred embodiments of the present application, and are not intended to limit the protection scope of the present application. Any feature disclosed in this specification (including the abstract and drawings), unless otherwise stated, can be replaced by other equivalent or alternative features with similar purposes. That is, unless otherwise stated, each feature is only an example of a series of equivalent or similar features.

Claims

1. A physiological signal acquisition method, applied to a wearable device equipped with a skin electrical acquisition module and a physiological signal acquisition module, characterized in that: The collection method comprises: Obtain the skin electrical signal value collected by the skin electrical collection module; Determining a current wearing state of the wearable device according to a preset skin electrical threshold and the skin electrical signal value; In response to the wearing state being an abnormal state, controlling the physiological signal acquisition module to stop acquiring physiological signals; Determining the current wearing state of the wearable device according to a preset skin electrical threshold and the skin electrical signal value includes: Comparing the skin electrical signal value with the preset skin electrical threshold; In response to the skin electrical signal value being less than a preset skin electrical threshold, determining that the current wearing state of the wearable device is an abnormal state; The step of determining the preset skin electrical threshold comprises: Acquire environmental monitoring data of the environment where the wearable device is located; the environmental monitoring data includes temperature data and humidity data; Determine or adjust the preset skin electrical threshold according to the environmental monitoring data; Determining or adjusting the preset skin electrical threshold according to the environmental monitoring data includes: determining the skin electrical threshold corresponding to the environmental monitoring data as the preset skin electrical threshold according to the mapping relationship between the environmental monitoring data and the skin electrical threshold obtained in advance; or, Determine or adjust the preset skin charge threshold according to the environmental monitoring data, including: input the environmental monitoring data into a neural network model to perform skin charge threshold prediction processing, the skin charge threshold prediction value output by the neural network model is the preset skin charge threshold; wherein the neural network model is pre-supervised trained with a training set, and the training set includes historical environmental monitoring data and corresponding historical skin charge monitoring data.

2. The method according to claim 1, characterized in that The determining the wearing state of the wearable device according to the preset skin electrical threshold and the skin electrical signal value further includes: in response to the skin electrical signal value being greater than or equal to the preset skin electrical threshold, determining that the current wearing state of the wearable device is a non-abnormal state; The method further includes: in response to the wearing state of the wearable device being a non-abnormal state, controlling the physiological signal acquisition module to start or keep acquiring physiological signals.

3. The method according to claim 1, characterized in that In response to the wearing state being an abnormal state, the method further includes: clearing a signal filter signal of the wearable device, and / or setting a sampling value of a physiological signal to zero.

4. The method according to claim 1, characterized in that: The training of the neural network model includes: Get a sample data set and preprocess it; Perform feature extraction on the preprocessed sample data set; Dividing the sample data set after feature extraction into the training set, the validation set and the test set; wherein the validation set and the test set include the environmental monitoring data of the future time period and the corresponding skin electrical monitoring data; The neural network model is initially trained based on the training set, and the neural network model after the initial training is verified and model parameters are optimized and adjusted based on the verification set and the test set.

5. The method according to claim 4, characterized in that The neural network model is initially trained based on the training set, and the neural network model after the initial training is verified and the model parameters are optimized and adjusted based on the verification set and the test set, including: Inputting the training set into a pre-built neural network model for training, optimizing the model parameters and calculating the loss function of the neural network model, until the loss function meets a preset condition or the number of iterations of the pre-built neural network model reaches a preset number, thereby obtaining a trained skin electrodermal threshold prediction model; Verifying the skin electrode threshold prediction model based on the verification set, evaluating the performance of the skin electrode threshold prediction model and adjusting the hyperparameters of the model; The predictive ability of the adjusted skin conductance threshold prediction model is tested based on the test set.

6. A physiological signal acquisition system, used to execute the physiological signal acquisition method according to any one of claims 1 to 5, applied to a wearable device equipped with a skin electrical acquisition module and a physiological signal acquisition module, characterized in that: The acquisition system comprises: The skin electrical signal value acquisition module (101) is used to acquire the skin electrical signal value collected by the skin electrical acquisition module; A wearing state determination module (102), configured to determine the current wearing state of the wearable device according to a preset skin electrical threshold and the skin electrical signal value; The collection control module (103) is used to control the physiological signal collection module to stop collecting physiological signals in response to the wearing state being an abnormal state.

7. A wearable device, characterized in that: include: A device body (1), an galvanometer acquisition module (11), a physiological signal acquisition module (12) and a physiological signal acquisition system as claimed in claim 6, wherein the galvanometer acquisition module (11) and the physiological signal acquisition module (12) are arranged on the device body (1), and the physiological signal acquisition system is connected to the galvanometer acquisition module (11) and the physiological signal acquisition module (12), respectively; the galvanometer acquisition module (11) is used to acquire the galvanometer signal value of the user, and the physiological signal acquisition module (12) is used to acquire the physiological signal data of the user.