Physiological state analysis methods, equipment, and storage media based on skin sweat glands and radio waves
By synchronously acquiring and preprocessing electrical signals from skin sweat glands and electrical conduction signals, and using a two-branch neural network model with an attention mechanism to establish a time synchronization and causal relationship model, the problem of time synchronization between electrical signals from skin sweat glands and electrical conduction signals is solved, thereby improving the accuracy and robustness of physiological state monitoring.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- SHENZHEN ABEL HEALTH TECH CO LTD
- Filing Date
- 2025-12-29
- Publication Date
- 2026-05-26
AI Technical Summary
Existing technologies struggle to establish time synchronization and causal correlation models between skin sweat gland electrical signals and electrical wave conduction signals, resulting in insufficient accuracy and robustness in physiological state monitoring.
By synchronously collecting electrical signals from skin sweat glands and electrical conduction signals, and after preprocessing, a time synchronization and causal association model is established using a dual-branch neural network model based on an attention mechanism. The two signal features are dynamically mapped and fused to generate physiological state classification results.
It improves the accuracy and robustness of physiological state monitoring, enabling more precise identification of emotional stress responses, physical thermal stimulation responses, or non-physiological disturbances.
Smart Images
Figure CN122074946A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of signal processing technology, and in particular to a method, device, and storage medium for analyzing the physiological state of skin sweat glands and radio waves. Background Technology
[0002] In fields such as wearable health monitoring, human-computer interaction, and affective computing, there is a growing demand for accurate and continuous monitoring of human physiological states. Existing technologies are mostly based on the collection and analysis of single types of physiological signals. While these have applications in specific scenarios, their inherent limitations severely restrict the accuracy and robustness of the monitoring results. Currently, some research attempts to fuse electrical signals from sweat glands with electromagnetic conduction signals, aiming to improve the accuracy of physiological state monitoring through collaborative analysis of the two signals. However, sweat gland electrical signals have a rapid response (milliseconds), while electromagnetic conduction signals reflect deeper changes relatively slowly. The significant differences in their temporal characteristics make it difficult to establish accurate time synchronization and causal correlation models, resulting in less than ideal collaborative analysis results. Therefore, establishing time synchronization and causal correlation models between sweat gland electrical signals and electromagnetic conduction signals, and collaboratively analyzing the two signals to improve the accuracy of physiological state monitoring, is a pressing technical problem that needs to be solved. Summary of the Invention
[0003] This application provides a method, device, and storage medium for analyzing the physiological state of skin sweat glands and electromagnetic waves. By using a time synchronization and causal correlation model between the electrical signals of skin sweat glands and the conductive signals of electromagnetic waves, the two signals are analyzed in a coordinated manner, aiming to improve the accuracy and robustness of physiological state monitoring.
[0004] In a first aspect, embodiments of this application provide a method for analyzing the physiological state based on skin sweat glands and electromagnetic waves, comprising: synchronously acquiring the user's skin sweat gland electrical signals and electromagnetic wave conduction signals; preprocessing the skin sweat gland electrical signals and electromagnetic wave conduction signals to extract skin sweat gland electrical signal features from the skin sweat gland electrical signals and electromagnetic wave conduction signal features from the electromagnetic wave conduction signals; inputting the skin sweat gland electrical signal features and electromagnetic wave conduction signal features into a pre-trained time synchronization and causal association model, establishing a dynamic mapping relationship between the skin sweat gland electrical signal features and electromagnetic wave conduction signal features in the time synchronization and causal association model, and performing feature fusion based on the dynamic mapping relationship; and generating a classification result of the user's physiological state based on the output of the time synchronization and causal association model.
[0005] In one embodiment, the time synchronization and causal association model includes a two-branch neural network model based on an attention mechanism. The two-branch neural network model based on an attention mechanism includes a first branch network, a second branch network, and an attention fusion layer. In the time synchronization and causal association model, a dynamic mapping relationship is established between the electrical signal features of skin sweat glands and the electrical signal features of electromagnetic waves. Feature fusion is performed based on the dynamic mapping relationship, including: processing the electrical signal features of skin sweat glands through the first branch network to obtain a first high-valence feature. The second branch network processes the characteristics of the electromagnetic wave conduction signal to obtain the second high-valence feature; through the attention fusion layer, the fusion weights of the first and second high-valence features are dynamically calculated and assigned at each time step; the first high-valence feature reflects the type and intensity of sympathetic nerve activity, and the second high-valence feature reflects the physiological state of deep tissues; feature fusion is performed on the first and second high-valence features based on the fusion weights.
[0006] In one embodiment, the attention fusion layer dynamically calculates and assigns fusion weights for the first higher-order feature and the second higher-order feature at each time step, including: assigning a first fusion weight to the first higher-order feature when the initial event of skin conduction response in the skin sweat gland electrical signal is detected; assigning a second fusion weight to the second higher-order feature when the skin conduction response enters its peak or recovery period; and reducing the first fusion weight when interference features are detected in the electromagnetic conduction signal feature.
[0007] In one embodiment, generating a classification result of a user's physiological state includes: classifying a first high-order feature based on the fused features to distinguish the classification result of its corresponding physiological state; the classification result of the physiological state includes: emotional stress response, physical thermal stimulation response, or non-physiological interference.
[0008] In one embodiment, the preprocessing of the skin sweat gland electrical signal and the electromagnetic conduction signal includes: baseline correction and motion artifact elimination processing of the skin sweat gland electrical signal; and demodulation processing of the electromagnetic conduction signal to extract its amplitude and phase information.
[0009] In one embodiment, the electrical signal characteristics of skin sweat glands include at least one of the following: skin conductivity level, peak value of skin conductivity response, rise time, recovery time, and frequency of occurrence; the electrical wave conductivity characteristics include at least one of the following: impedance amplitude, phase angle, and impedance spectrum slope at a specific frequency.
[0010] In one embodiment, the method further includes: identifying and filtering common-mode interference by utilizing the different response characteristics of electromagnetic wave conduction signals and skin sweat gland electrical signals to environmental interference.
[0011] Secondly, embodiments of this application provide a physiological state analysis device based on skin sweat glands and radio waves, comprising: The acquisition module is used to simultaneously acquire the user's skin sweat gland electrical signals and electromagnetic conduction signals; The processing module is used to preprocess the electrical signals of skin sweat glands and the conduction signals of electromagnetic waves to extract the electrical signal features of skin sweat glands from the electrical signals of skin sweat glands and the conduction signals of electromagnetic waves from the conduction signals of electromagnetic waves. The fusion module is used to input the electrical signal features of skin sweat glands and the electrical conduction signal features of electromagnetic waves into a pre-trained time synchronization and causal association model. In the time synchronization and causal association model, a dynamic mapping relationship is established between the electrical signal features of skin sweat glands and the electrical conduction signal features of electromagnetic waves, and feature fusion is performed based on the dynamic mapping relationship. The generation module is used to generate classification results of the user's physiological state based on the output of the time synchronization and causal association model.
[0012] In one embodiment, the time synchronization and causal association model includes an attention-based dual-branch neural network model, which includes a first branch network, a second branch network, and an attention fusion layer. The fusion module includes: The first processing unit is used to process the electrical signal characteristics of skin sweat glands through the first branch network to obtain the first high-valence feature; The second processing unit is used to process the electromagnetic conduction signal characteristics through the second branch network to obtain the second high-valence feature; The computational unit is used to dynamically calculate and assign fusion weights of the first high-valence feature and the second high-valence feature at each time step through the attention fusion layer; the first high-valence feature reflects the type and intensity of sympathetic nerve activity, and the second high-order feature reflects the physiological state of deep tissues. The fusion unit is used to perform feature fusion on the first high-order feature and the second high-order feature based on the fusion weight.
[0013] In one embodiment, the computing unit is specifically configured to: assign a first fusion weight to a first higher-order feature when the initiation event of skin conductance response is detected in the skin sweat gland electrical signal through the attention fusion layer; assign a second fusion weight to a second higher-order feature when the skin conductance response enters the peak or recovery period; and reduce the first fusion weight when interference features are detected in the electromagnetic conductance signal feature.
[0014] In one embodiment, the generation module is specifically used to: classify the first high-order features based on the fused features to distinguish the classification results of their corresponding physiological states; the classification results of physiological states include: emotional stress response, physical thermal stimulation response, or non-physiological interference.
[0015] In one embodiment, the processing module is specifically used for: performing baseline correction and motion artifact elimination processing on the electrical signals of skin sweat glands; and demodulating the electromagnetic wave conduction signals to extract their amplitude and phase information.
[0016] In one embodiment, the electrical signal characteristics of skin sweat glands include at least one of the following: skin conductivity level, peak value of skin conductivity response, rise time, recovery time, and frequency of occurrence; the electrical wave conductivity characteristics include at least one of the following: impedance amplitude, phase angle, and impedance spectrum slope at a specific frequency.
[0017] In one embodiment, the device further includes an identification module for identifying and filtering common-mode interference by utilizing the different response characteristics of electromagnetic conductive signals and skin sweat gland electrical signals to environmental interference.
[0018] Thirdly, embodiments of this application provide an electronic device, including: Memory and processing modules; The memory is used to store computer programs; The processing module is used to execute the computer program and, when executing the computer program, to implement the steps of the physiological state analysis method based on skin sweat glands and electrical waves as described in the first aspect above.
[0019] Fourthly, embodiments of this application provide a computer-readable storage medium storing a computer program; When the computer program is executed by one or more processing modules, the one or more processing modules perform the steps of the physiological state analysis method based on skin sweat glands and electrical waves as described in the first aspect above.
[0020] This application provides a method, device, and storage medium for analyzing physiological states based on skin sweat glands and electromagnetic waves. The method includes: simultaneously acquiring electrical signals from a user's skin sweat glands and conductive signals from electromagnetic waves; preprocessing the electrical and conductive signals to extract features from both signals; inputting the extracted features into a pre-trained time synchronization and causal association model to establish a dynamic mapping relationship between the features of the skin sweat gland electrical signals and the conductive signals, and performing feature fusion based on this dynamic mapping relationship; and generating a classification result of the user's physiological state based on the output of the time synchronization and causal association model. By using a time synchronization and causal association model to collaboratively analyze the two signals, the accuracy and robustness of physiological state monitoring are improved. Attached Figure Description
[0021] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0022] Figure 1 A flowchart illustrating the physiological state analysis method based on skin sweat glands and radio waves provided in this application embodiment; Figure 2 for Figure 1 A schematic diagram illustrating the specific implementation process of S103 in the middle; Figure 3 A schematic diagram of the physiological state analysis device based on skin sweat glands and radio waves provided in the embodiments of this application; Figure 4 A schematic block diagram of a physiological state analysis device based on skin sweat glands and radio waves provided in this application embodiment. Detailed Implementation
[0023] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0024] The flowchart shown in the attached diagram is for illustrative purposes only and does not necessarily include all content and operations / steps, nor does it necessarily have to be performed in the order described. For example, some operations / steps can be broken down, combined, or partially merged, so the actual execution order may change depending on the actual situation.
[0025] It should also be understood that the terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the scope of the application. As used in this specification and the appended claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise.
[0026] It should also be further understood that the term “and / or” as used in this application specification and the appended claims means any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.
[0027] The technical solution provided in this application will be described in detail below with reference to the accompanying drawings.
[0028] Please see Figure 1 , Figure 1 This is a flowchart illustrating the physiological state analysis method based on skin sweat glands and radio waves provided in this application embodiment. The physiological state analysis method based on skin sweat glands and radio waves provided in this application embodiment is implemented by a physiological state analysis device based on skin sweat glands and radio waves. This application embodiment does not limit the physiological state analysis device based on skin sweat glands and radio waves. Specifically, as... Figure 1 As shown, the physiological state analysis method based on skin sweat glands and electrical waves includes steps S101 to S104. Details are as follows: S101: Synchronously collects the user's skin sweat gland electrical signals and electromagnetic conduction signals.
[0029] Synchronous data acquisition is performed using a wearable device integrating multimodal sensors. Specifically, the wearable device includes an Ag / AgCl electrode pair and a radio frequency transceiver. The Ag / AgCl electrode pair is used to acquire electrical signals from skin sweat glands; the radio frequency transceiver (such as a chip operating in the 100kHz to 100MHz frequency range) is used to acquire electromagnetic conductivity signals (i.e., bioimpedance signals). The Ag / AgCl electrode pair and the radio frequency transceiver are triggered by the same main control chip to ensure that the acquired time-series data are strictly aligned in the time domain, with a time synchronization accuracy requirement of less than 10 milliseconds.
[0030] S102: Preprocess the electrical signals of the sweat glands and the conduction signals of the electromagnetic waves to extract the electrical signal features of the sweat glands from the sweat glands and the conduction signals of the electromagnetic waves from the conduction signals.
[0031] The electrical signals and electrical conductivity signals of the skin sweat glands obtained by synchronous application of the agent were preprocessed, and features characterizing the physiological state were extracted from them.
[0032] Specifically, the electrical signals from sweat glands in the skin are first subjected to high-pass filtering to eliminate baseline drift, and then an adaptive filtering algorithm based on a triaxial accelerometer is used to suppress motion artifacts. Peak detection is performed on the preprocessed electrical signals from sweat glands to extract features such as skin conductivity level (the average conductivity level of the signal per unit time), peak value of skin conductance response (SCR), rise time (the time from the starting point to the peak value), recovery time (the time required for the peak value to decrease to 63% of the peak value), and frequency of occurrence per unit time.
[0033] The radio wave conduction signal received by the RF transceiver is demodulated, and its amplitude (|Z|) and phase (∠Z) information are calculated. At multiple specific frequency points (e.g., 50kHz, 100kHz, 500kHz), the impedance amplitude (average impedance amplitude at a specific frequency), phase angle (phase value or its change at a specific frequency), and impedance spectrum slope (slope value obtained by linear fitting of impedance amplitude at different frequencies) are extracted from the demodulated signal.
[0034] Specifically, the electrical signal characteristics of skin sweat glands include at least one of the following: skin conductivity level, peak value of skin conductivity response, rise time, recovery time, and frequency of occurrence; the electrical wave conductivity characteristics include at least one of the following: impedance amplitude, phase angle, and impedance spectrum slope at a specific frequency.
[0035] Furthermore, since the responses of skin sweat gland electrical signals and electromagnetic conduction signals to environmental interference (such as instantaneous temperature changes) differ, the signal-to-noise ratio of the overall signal can be improved by identifying and filtering common-mode interference. Specifically, before preprocessing the skin sweat gland electrical signals and electromagnetic conduction signals, the process includes: real-time comparison of the responses of the skin sweat gland electrical signals and electromagnetic conduction signals to the same environmental interference (such as sudden changes in ambient humidity). If the skin sweat gland electrical signals show drastic changes while the electromagnetic conduction signals do not exhibit a corresponding change pattern, the change is determined to be common-mode interference, and this segment of the skin sweat gland electrical signal is filtered out or marked, thereby improving the signal-to-noise ratio of the overall signal.
[0036] S103: Input the electrical signal features of skin sweat glands and the electrical conduction signal features of electromagnetic waves into a pre-trained time synchronization and causal association model. Establish a dynamic mapping relationship between the electrical signal features of skin sweat glands and the electrical conduction signal features of electromagnetic waves in the time synchronization and causal association model, and perform feature fusion based on the dynamic mapping relationship.
[0037] The time synchronization and causal association model includes a two-branch neural network model based on an attention mechanism. This model comprises a first branch network, a second branch network, and an attention fusion layer. The first branch network processes the electrical signal features of the sweat glands; the second branch network processes the electrical conduction signal features; and the attention fusion layer dynamically fuses the outputs of the two branches. In this application, the preprocessed extracted electrical signal features of the sweat glands and the electrical conduction signal features are input into a pre-trained time synchronization and causal association model. The core function of this model is to establish a dynamic mapping relationship between the two types of features and to perform feature fusion based on this dynamic mapping relationship to generate a fused feature vector that comprehensively represents the user's physiological state.
[0038] For example, such as Figure 2 As shown, Figure 2 for Figure 1 A schematic diagram illustrating the specific implementation process of S103. Figure 2 As can be seen, in this embodiment, S103 includes: S1031: The first high-valence feature is obtained by processing the electrical signal characteristics of the sweat glands in the skin through the first branch network.
[0039] The first branch network (which may contain a one-dimensional convolutional CNN and a long short-term memory (LSTM) network) processes the electrical signal features of skin sweat glands. The CNN layers learn local patterns (such as steep rising edges) of the skin sweat gland electrical signals, while the LSTM layers capture long-term dependencies in the time series. This first branch network outputs a first high-order feature, which abstracts the spatiotemporal pattern information of sympathetic neural activity.
[0040] S1032: The second high-valence feature is obtained by processing the characteristics of the electromagnetic wave conduction signal through the second branch network.
[0041] A second branch network (which is independent of the first branch network) is used to process the electrical conductivity signal features and learn the underlying physiological semantics (such as vasodilation and increased tissue fluid). This second branch network outputs a second higher-order feature, which reflects the physiological state of the deep tissues.
[0042] S1033: Through the attention fusion layer, the fusion weights of the first high-valence feature and the second high-valence feature are dynamically calculated and assigned at each time step; the first high-valence feature reflects the type and intensity of sympathetic nerve activity, and the second high-order feature reflects the physiological state of deep tissues; feature fusion is performed on the first high-order feature and the second high-order feature based on the fusion weights.
[0043] The attention fusion layer receives the first high-order feature output from the first branch network and the second high-order feature output from the second branch network. It calculates an attention energy score using a small neural network or dot product operation. This attention energy score quantifies the importance of each feature at the current time step. The attention energy score is then normalized to fusion weights between 0 and 1 using the Softmax function. Finally, the first and second high-order features are multiplied by their respective weights and summed to obtain the final fused feature.
[0044] Through an attention fusion layer, the fusion weights of the first and second higher-order features are dynamically calculated and assigned at each time step. This includes: when an initiation event of skin conductance response (such as a sudden high-slope positive change) is detected in the skin sweat gland electrical signal, a first fusion weight is assigned to the first higher-order feature; when the skin conductance response enters its peak or recovery phase, a second fusion weight is assigned to the second higher-order feature; and when interference features are detected in the electro-conductivity signal, the first fusion weight is reduced. The first fusion weight is a relatively high weight (e.g., 0.8) to prioritize evidence of surface event triggering. The second higher-order feature also has a relatively high weight (e.g., 0.7) to determine event intensity and recovery speed.
[0045] In addition, the attention fusion layer continuously monitors the second higher-order features. When a motion interference pattern is identified from the second higher-order features (e.g., a specific impedance noise pattern related to accelerometer data matched by a preset interference feature library), the weight of the first higher-order feature is significantly reduced (e.g., from 0.8 to 0.2), thereby greatly suppressing false positive events caused by motion artifacts.
[0046] S104: Based on the output of the time synchronization and causal relationship model, generate classification results for the user's physiological state.
[0047] In one embodiment, generating a classification result of a user's physiological state includes: classifying a first high-order feature based on the fused features to distinguish the classification result of its corresponding physiological state; the classification result of the physiological state includes: emotional stress response, physical thermal stimulation response, or non-physiological interference.
[0048] Specifically, if the fusion features simultaneously include high-amplitude skin sweat gland electrical signals and slight electromagnetic conduction signals, it is classified as an emotional stress response; if the fusion features include skin sweat gland electrical signals and strong electromagnetic conduction signals, it is classified as a physical thermal stimulation response; if the fusion features only contain skin sweat gland electrical signals and no electromagnetic conduction signals, it is classified as a non-physiological interference.
[0049] As can be seen from the above analysis, the physiological state analysis method based on skin sweat glands and electromagnetic waves provided in this application includes: simultaneously acquiring the user's skin sweat gland electrical signals and electromagnetic wave conduction signals; preprocessing the skin sweat gland electrical signals and electromagnetic wave conduction signals to extract features of the skin sweat gland electrical signals and electromagnetic wave conduction signals; inputting the extracted features into a pre-trained time synchronization and causal association model, establishing a dynamic mapping relationship between the features of the skin sweat gland electrical signals and the features of the electromagnetic wave conduction signals in the time synchronization and causal association model, and performing feature fusion based on the dynamic mapping relationship; and generating a classification result of the user's physiological state based on the output of the time synchronization and causal association model. By using the time synchronization and causal association model of skin sweat gland electrical signals and electromagnetic wave conduction signals to collaboratively analyze the two signals, the aim is to improve the accuracy and robustness of physiological state monitoring.
[0050] Please see Figure 3 As shown, Figure 3 This is a schematic diagram of the physiological state analysis device based on skin sweat glands and radio waves provided in an embodiment of this application. Figure 3 As can be seen, the physiological state analysis device 300 based on skin sweat glands and radio waves provided in this application embodiment includes: Acquisition module 301 is used to simultaneously acquire the user's skin sweat gland electrical signals and electromagnetic conduction signals; Processing module 302 is used to preprocess the electrical signals of skin sweat glands and the conduction signals of electromagnetic waves, so as to extract the electrical signal features of skin sweat glands from the electrical signals of skin sweat glands and the conduction signal features of electromagnetic waves from the conduction signals of electromagnetic waves. The fusion module 303 is used to input the electrical signal features of skin sweat glands and the electrical conduction signal features of electromagnetic waves into a pre-trained time synchronization and causal association model, establish a dynamic mapping relationship between the electrical signal features of skin sweat glands and the electrical conduction signal features of electromagnetic waves in the time synchronization and causal association model, and perform feature fusion based on the dynamic mapping relationship. The generation module 304 is used to generate classification results of the user's physiological state based on the output of the time synchronization and causal association model.
[0051] In one embodiment, the time synchronization and causal association model includes an attention-based dual-branch neural network model, which includes a first branch network, a second branch network, and an attention fusion layer. Fusion module 303 includes: The first processing unit is used to process the electrical signal characteristics of skin sweat glands through the first branch network to obtain the first high-valence feature; The second processing unit is used to process the electromagnetic conduction signal characteristics through the second branch network to obtain the second high-valence feature; The computational unit is used to dynamically calculate and assign fusion weights of the first high-valence feature and the second high-valence feature at each time step through the attention fusion layer; the first high-valence feature reflects the type and intensity of sympathetic nerve activity, and the second high-order feature reflects the physiological state of deep tissues. The fusion unit is used to perform feature fusion on the first high-order feature and the second high-order feature based on the fusion weight.
[0052] In one embodiment, the computing unit is specifically configured to: assign a first fusion weight to a first higher-order feature when the initiation event of skin conductance response is detected in the skin sweat gland electrical signal through the attention fusion layer; assign a second fusion weight to a second higher-order feature when the skin conductance response enters the peak or recovery period; and reduce the first fusion weight when interference features are detected in the electromagnetic conductance signal feature.
[0053] In one embodiment, the generation module 304 is specifically used to: classify the first high-order features based on the fused features to distinguish the classification results of their corresponding physiological states; the classification results of physiological states include: emotional stress response, physical thermal stimulation response, or non-physiological interference.
[0054] In one embodiment, the processing module 302 is specifically used for: performing baseline correction and motion artifact elimination processing on the electrical signals of skin sweat glands; and demodulating the electromagnetic wave conduction signals to extract their amplitude and phase information.
[0055] In one embodiment, the electrical signal characteristics of skin sweat glands include at least one of the following: skin conductivity level, peak value of skin conductivity response, rise time, recovery time, and frequency of occurrence; the electrical wave conductivity characteristics include at least one of the following: impedance amplitude, phase angle, and impedance spectrum slope at a specific frequency.
[0056] In one embodiment, the device further includes an identification module for identifying and filtering common-mode interference by utilizing the different response characteristics of electromagnetic conductive signals and skin sweat gland electrical signals to environmental interference.
[0057] It should be noted that the specific implementation process of each module or unit mentioned above can be referred to the specific implementation process of each step in the previous method embodiment, and will not be repeated here.
[0058] Please see Figure 4 As shown, Figure 4 A schematic block diagram of a physiological state analysis device based on skin sweat glands and radio waves provided in an embodiment of this application.
[0059] For example, a physiological state analysis device 400 based on skin sweat glands and radio waves includes a processing module 401 and a memory 402.
[0060] For example, the processing module 401 and the memory 402 are connected via a bus 403, such as an I2C (Inter-integrated Circuit) bus.
[0061] Specifically, the processing module 401 can be a microcontroller unit (MCU), a central processing unit (CPU), or a digital signal processor (DSP), etc.
[0062] Specifically, the memory 402 can be a Flash chip, a read-only memory (ROM) disk, an optical disk, a USB flash drive, or a portable hard drive, etc.
[0063] The processing module 401 is used to run the computer program stored in the memory 402, and implements the steps of the above-mentioned physiological state analysis method based on skin sweat glands and radio waves when executing the computer program.
[0064] For example, the processing module 401 is used to run a computer program stored in the memory 402, and performs the following steps when executing the computer program: The system synchronously collects the user's skin sweat gland electrical signals and electromagnetic conduction signals; preprocesses these signals to extract features from both the sweat gland electrical signals and the electromagnetic conduction signals; inputs these features into a pre-trained time synchronization and causal association model to establish a dynamic mapping relationship between the sweat gland electrical signal features and the electromagnetic conduction signals, and performs feature fusion based on this dynamic mapping relationship; and generates a classification result of the user's physiological state based on the output of the time synchronization and causal association model.
[0065] In one embodiment, the time synchronization and causal association model includes a two-branch neural network model based on an attention mechanism. The two-branch neural network model based on an attention mechanism includes a first branch network, a second branch network, and an attention fusion layer. In the time synchronization and causal association model, a dynamic mapping relationship is established between the electrical signal features of skin sweat glands and the electrical signal features of electromagnetic waves. Feature fusion is performed based on the dynamic mapping relationship, including: processing the electrical signal features of skin sweat glands through the first branch network to obtain a first high-valence feature. The second branch network processes the characteristics of the electromagnetic wave conduction signal to obtain the second high-valence feature; through the attention fusion layer, the fusion weights of the first and second high-valence features are dynamically calculated and assigned at each time step; the first high-valence feature reflects the type and intensity of sympathetic nerve activity, and the second high-valence feature reflects the physiological state of deep tissues; feature fusion is performed on the first and second high-valence features based on the fusion weights.
[0066] In one embodiment, the attention fusion layer dynamically calculates and assigns fusion weights for the first higher-order feature and the second higher-order feature at each time step, including: assigning a first fusion weight to the first higher-order feature when the initial event of skin conduction response in the skin sweat gland electrical signal is detected; assigning a second fusion weight to the second higher-order feature when the skin conduction response enters its peak or recovery period; and reducing the first fusion weight when interference features are detected in the electromagnetic conduction signal feature.
[0067] In one embodiment, generating a classification result of a user's physiological state includes: classifying a first high-order feature based on the fused features to distinguish the classification result of its corresponding physiological state; the classification result of the physiological state includes: emotional stress response, physical thermal stimulation response, or non-physiological interference.
[0068] In one embodiment, the preprocessing of the skin sweat gland electrical signal and the electromagnetic conduction signal includes: baseline correction and motion artifact elimination processing of the skin sweat gland electrical signal; and demodulation processing of the electromagnetic conduction signal to extract its amplitude and phase information.
[0069] In one embodiment, the electrical signal characteristics of skin sweat glands include at least one of the following: skin conductivity level, peak value of skin conductivity response, rise time, recovery time, and frequency of occurrence; the electrical wave conductivity characteristics include at least one of the following: impedance amplitude, phase angle, and impedance spectrum slope at a specific frequency.
[0070] In one embodiment, the method further includes: identifying and filtering common-mode interference by utilizing the different response characteristics of electromagnetic wave conduction signals and skin sweat gland electrical signals to environmental interference.
[0071] The specific principles and implementation methods of the physiological state analysis device based on skin sweat glands and radio waves provided in this application embodiment are similar to those of the physiological state analysis method based on skin sweat glands and radio waves in the foregoing embodiments, and will not be repeated here.
[0072] This application embodiment also provides a computer-readable storage medium storing a computer program, which, when executed by a processing module, causes the processing module to perform the following steps: The system synchronously collects the user's skin sweat gland electrical signals and electromagnetic conduction signals; preprocesses these signals to extract features from both the sweat gland electrical signals and the electromagnetic conduction signals; inputs these features into a pre-trained time synchronization and causal association model to establish a dynamic mapping relationship between the sweat gland electrical signal features and the electromagnetic conduction signals, and performs feature fusion based on this dynamic mapping relationship; and generates a classification result of the user's physiological state based on the output of the time synchronization and causal association model.
[0073] In one embodiment, the time synchronization and causal association model includes a two-branch neural network model based on an attention mechanism. The two-branch neural network model based on an attention mechanism includes a first branch network, a second branch network, and an attention fusion layer. In the time synchronization and causal association model, a dynamic mapping relationship is established between the electrical signal features of skin sweat glands and the electrical signal features of electromagnetic waves. Feature fusion is performed based on the dynamic mapping relationship, including: processing the electrical signal features of skin sweat glands through the first branch network to obtain a first high-valence feature. The second branch network processes the characteristics of the electromagnetic wave conduction signal to obtain the second high-valence feature; through the attention fusion layer, the fusion weights of the first and second high-valence features are dynamically calculated and assigned at each time step; the first high-valence feature reflects the type and intensity of sympathetic nerve activity, and the second high-valence feature reflects the physiological state of deep tissues; feature fusion is performed on the first and second high-valence features based on the fusion weights.
[0074] In one embodiment, the attention fusion layer dynamically calculates and assigns fusion weights for the first higher-order feature and the second higher-order feature at each time step, including: assigning a first fusion weight to the first higher-order feature when the initial event of skin conduction response in the skin sweat gland electrical signal is detected; assigning a second fusion weight to the second higher-order feature when the skin conduction response enters its peak or recovery period; and reducing the first fusion weight when interference features are detected in the electromagnetic conduction signal feature.
[0075] In one embodiment, generating a classification result of a user's physiological state includes: classifying a first high-order feature based on the fused features to distinguish the classification result of its corresponding physiological state; the classification result of the physiological state includes: emotional stress response, physical thermal stimulation response, or non-physiological interference.
[0076] In one embodiment, the preprocessing of the skin sweat gland electrical signal and the electromagnetic conduction signal includes: baseline correction and motion artifact elimination processing of the skin sweat gland electrical signal; and demodulation processing of the electromagnetic conduction signal to extract its amplitude and phase information.
[0077] In one embodiment, the electrical signal characteristics of skin sweat glands include at least one of the following: skin conductivity level, peak value of skin conductivity response, rise time, recovery time, and frequency of occurrence; the electrical wave conductivity characteristics include at least one of the following: impedance amplitude, phase angle, and impedance spectrum slope at a specific frequency.
[0078] In one embodiment, the method further includes: identifying and filtering common-mode interference by utilizing the different response characteristics of electromagnetic wave conduction signals and skin sweat gland electrical signals to environmental interference.
[0079] The computer-readable storage medium can be an internal storage unit of the physiological state analysis device based on skin sweat glands and radio waves in the aforementioned embodiments, such as a hard drive or memory of the physiological state analysis device based on skin sweat glands and radio waves. The computer-readable storage medium can also be an external storage device of the physiological state analysis device based on skin sweat glands and radio waves, such as a plug-in hard drive, SmartMedia Card (SMC), Secure Digital (SD) card, or Flash Card equipped on the physiological state analysis device based on skin sweat glands and radio waves.
[0080] It should be understood that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the scope of the application.
[0081] It should also be understood that the term “and / or” as used in this application and the appended claims means any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.
[0082] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in this application, and these modifications or substitutions should all be covered 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.
Claims
1. A method for analyzing a physiological state of a sweat gland of skin based on an electric wave, characterized by, include: Simultaneously collect the user's skin sweat gland electrical signals and electromagnetic conduction signals; The skin sweat gland electrical signal and the electromagnetic wave conduction signal are preprocessed to extract skin sweat gland electrical signal features from the skin sweat gland electrical signal and electromagnetic wave conduction signal features from the electromagnetic wave conduction signal; The electrical signal features of the sweat glands and the electrical conduction signal features are input into a pre-trained time synchronization and causal association model. A dynamic mapping relationship between the electrical signal features of the sweat glands and the electrical conduction signal features is established in the time synchronization and causal association model, and feature fusion is performed based on the dynamic mapping relationship. Based on the output of the time synchronization and causal relationship model, a classification result of the user's physiological state is generated.
2. The method according to claim 1, wherein the method is characterized by, The time synchronization and causal correlation model includes a two-branch neural network model based on an attention mechanism, comprising a first branch network, a second branch network, and an attention fusion layer; The step of establishing a dynamic mapping relationship between the electrical signal features of the skin sweat glands and the electrical conduction signal features in the time synchronization and causal correlation model, and performing feature fusion based on the dynamic mapping relationship, includes: The first high-valence feature is obtained by processing the electrical signal characteristics of the skin sweat glands through the first branch network; The second branch network is used to process the electromagnetic wave conduction signal characteristics to obtain the second high-valence characteristic; Through the attention fusion layer, the fusion weights of the first high-valence feature and the second high-valence feature are dynamically calculated and assigned at each time step; the first high-valence feature reflects the type and intensity of sympathetic nerve activity, and the second high-order feature reflects the physiological state of deep tissues. Based on the fusion weights, feature fusion is performed on the first higher-order feature and the second higher-order feature.
3. The method according to claim 2, wherein the method is characterized by, The step of dynamically calculating and allocating fusion weights for the first high-order feature and the second high-order feature at each time step through the attention fusion layer includes: Through the attention fusion layer, when the initiation event of skin conduction response is detected in the skin sweat gland electrical signal, a first fusion weight is assigned to the first high-order feature; when the skin conduction response enters the peak or recovery period, a second fusion weight is assigned to the second high-order feature; when interference features are detected in the electromagnetic conduction signal feature, the first fusion weight is reduced.
4. The method according to claim 3, wherein the method is characterized by, The generation of classification results for the user's physiological state includes: Based on the fused features, the first higher-order features are classified to distinguish the classification results of their corresponding physiological states; the classification results of the physiological states include: emotional stress response, physical thermal stimulation response, or non-physiological interference.
5. The skin sweat gland and electric wave-based physiological state analyzer according to claim 1, wherein, The preprocessing of the electrical signals from the skin sweat glands and the electrical conduction signals includes: Baseline correction and motion artifact elimination were performed on the electrical signals of the skin sweat glands; The electromagnetic wave conduction signal is demodulated to extract its amplitude and phase information.
6. The method according to claim 5, wherein the method is characterized by, The electrical signal characteristics of the skin sweat glands include at least one of the following: skin conductivity level, peak value of skin conductivity response, rise time, recovery time, and frequency of occurrence; the electrical wave conduction signal characteristics include at least one of the following: impedance amplitude, phase angle, and impedance spectrum slope at a specific frequency.
7. The method according to claim 1, wherein the method is characterized by, The method further includes: By utilizing the different response characteristics of the electromagnetic wave conduction signal and the skin sweat gland electrical signal to environmental interference, common-mode interference can be identified and filtered out.
8. A physiological state analysis device based on skin sweat glands and electrical waves, characterized in that, include: The acquisition module is used to simultaneously acquire the user's skin sweat gland electrical signals and electromagnetic conduction signals; The processing module is used to preprocess the skin sweat gland electrical signal and the electromagnetic wave conduction signal to extract skin sweat gland electrical signal features from the skin sweat gland electrical signal and electromagnetic wave conduction signal features from the electromagnetic wave conduction signal; The fusion module is used to input the electrical signal features of the skin sweat glands and the electrical conduction signal features into a pre-trained time synchronization and causal association model, establish a dynamic mapping relationship between the electrical signal features of the skin sweat glands and the electrical conduction signal features in the time synchronization and causal association model, and perform feature fusion based on the dynamic mapping relationship. The generation module is used to generate classification results of the user's physiological state based on the output of the time synchronization and causal association model.
9. An electronic device, characterized in that, include: Memory and processing modules; The memory is used to store computer programs; The processing module is used to execute the computer program and, when executing the computer program, to implement the steps of the physiological state analysis method based on skin sweat glands and radio waves as described in any one of claims 1 to 7.
10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program; When the computer program is executed by one or more processing modules, the one or more processing modules perform the steps of the physiological state analysis method based on skin sweat glands and electrical waves as described in any one of claims 1 to 7.