An underwater health risk assessment method and device, diving mask and medium

Abstract

The present application relates to the medical health technical field, especially underwater health risk assessment method, device, diving mirror and medium. By fusing multi-channel multispectral photoelectric volume pulse wave signal, eye movement image sequence and three kinds of heterogeneous information of motion data, a multi-dimensional physiological state monitoring system in underwater environment is constructed. The adaptive artifact suppression algorithm driven by motion data is adopted, and the signal fusion strategy combined with channel quality index weighting is adopted, which effectively improves the signal quality under the condition of low signal-to-noise ratio; At the same time, the eye movement feature analysis is introduced as the supplementary criterion of cognitive load and fatigue state, which makes up for the insufficient representation ability of single physiological parameter in complex underwater task. The method realizes signal preprocessing, physiological parameter prediction and comprehensive state evaluation, can output continuous health risk quantization results under the real-time constraint, provides timely safety warning for diving personnel, and significantly reduces the probability of underwater operation accident.

Description

Technical Field

[0001] This invention relates to the field of medical and health technology, and in particular to an underwater health risk assessment method, device, diving mask, and medium. Background Technology

[0002] With the increasing popularity of diving and the growing frequency of marine resource development activities, the safety of underwater workers has become a growing concern. In the underwater environment, the human body endures multiple physiological stresses, including high pressure, low temperature, and hypoxia. These stressors can lead to health risks such as nitrogen narcosis, oxygen toxicity, and decompression sickness, which can even be life-threatening in severe cases. Current technologies assess a diver's condition using physiological indicators such as heart rate and blood oxygen saturation. However, the underwater environment significantly affects sensor performance; signal acquisition is easily interfered with by water flow, pressure, and temperature, resulting in insufficient data reliability and low accuracy in health risk assessment. Therefore, improving the accuracy of health risk assessment for underwater workers is an urgent problem to be solved. Summary of the Invention

[0003] In view of this, embodiments of this application provide an underwater health risk assessment method, device, diving mask, and medium to solve the problem of low accuracy in health risk assessment for underwater workers in underwater environments.

[0004] In a first aspect, embodiments of this application provide an underwater health risk assessment method, the underwater health risk assessment method comprising: Acquire multi-channel, multispectral photoplethysmography (PPG) signals, eye-tracking image sequences, and motion data of the person being tested within any time window in an underwater environment; Based on the motion data, artifact suppression is performed on the multi-channel multispectral photoplethysmography signal to obtain the artifact-suppressed photoplethysmography signal. The channel quality index of each channel multispectral photoplethysmography (PPI) signal is determined. Based on the channel quality index, the artifact-suppressed PPI signals are weighted and fused to obtain the fused PPI signals. Based on the fused photoplethysmography (PPG) signal, the multi-channel multispectral PPG signal, and the motion data, the physiological parameters of the person to be tested are predicted to obtain the physiological parameter prediction results. Feature extraction is performed on each frame of the eye-tracking image to obtain the pupil features of each frame of the eye-tracking image. Based on the pupil features of each frame of the eye-tracking image, the eye-tracking feature vector is calculated. Based on the physiological parameter prediction results within the time window and the eye movement feature vector, the comprehensive state index of the person to be tested is predicted, and the comprehensive state index prediction result and confidence level are obtained. Based on the prediction results of the comprehensive status index within all time windows and the confidence level, the health risk of the person to be tested is assessed, and the health risk assessment result is obtained.

[0005] Secondly, embodiments of this application provide an underwater health risk assessment device, the underwater health risk assessment device comprising: The acquisition module is used to acquire multi-channel multispectral photoplethysmography (PPG) signals, eye-tracking image sequences, and motion data of the person under test in an underwater environment at any time window. The suppression module is used to perform artifact suppression on the multi-channel multispectral photoplethysmography signal based on the motion data, so as to obtain the artifact-suppressed photoplethysmography signal. The fusion module is used to determine the channel quality index of the multispectral photoplethysmography (PPG) signal for each channel, and to perform weighted fusion of the artifact-suppressed PPG signals based on the channel quality index to obtain the fused PPG signal. The physiological parameter prediction module is used to predict the physiological parameters of the person to be tested based on the fused photoplethysmography signal, the multi-channel multispectral photoplethysmography signal and the motion data, and obtain the physiological parameter prediction result. The eye movement feature extraction module is used to extract features from each frame of the eye movement image to obtain the pupil features of each frame of the eye movement image. Based on the pupil features of each frame of the eye movement image, the eye movement feature vector is calculated. The comprehensive state index prediction module is used to predict the comprehensive state index of the person to be tested based on the physiological parameter prediction results within the time window and the eye movement feature vector, and obtain the comprehensive state index prediction result and confidence level. The assessment module is used to assess the health risk of the person to be tested based on the prediction results of the comprehensive status index and the confidence level in all time windows, and to obtain the health risk assessment result.

[0006] Thirdly, embodiments of this application provide a diving mask, which includes a diving mask body, a multispectral photoplethysmography (PPG) sensor module, an eye-tracking module, a motion sensing module, and a processing module; The multispectral photoplethysmography (PPG) sensor module, the eye-tracking module, the motion sensing module, and the processing module are disposed on the diving mask body; The processing module is electrically connected to the multispectral photoplethysmography pulse wave sensing module, the eye-tracking module, and the motion sensing module, respectively. The multispectral photoplethysmography (PPG) sensing module is used to acquire multispectral PPG signals and transmit the multispectral PPG signals to the processing module. The eye-tracking module is used to acquire eye-tracking image sequences and transmit the eye-tracking image sequences to the processing module; The motion sensing module is used to collect motion data and transmit the motion data to the processing module. The processing module is used to execute any of the underwater health risk assessment methods described above.

[0007] Fourthly, embodiments of this application provide a computer-readable storage medium storing a computer program that, when executed by a processor, implements the underwater health risk assessment method as described in any of the preceding claims.

[0008] The advantages of this application compared to the prior art are: This application constructs a multi-dimensional physiological state monitoring system for underwater environments by fusing three types of heterogeneous information: multi-channel multispectral photoplethysmography (PPG) signals, eye-tracking image sequences, and motion data. An adaptive artifact suppression algorithm driven by motion data, combined with a channel quality index-weighted signal fusion strategy, effectively improves signal quality under low signal-to-noise ratio conditions. Simultaneously, eye-tracking feature analysis is introduced as a supplementary criterion for cognitive load and fatigue state, compensating for the insufficient characterization ability of single physiological parameters in complex underwater tasks. This method realizes the entire process from signal preprocessing and physiological parameter prediction to comprehensive state assessment, and can output continuous quantitative results of health risks under real-time constraints, providing timely safety warnings for divers and significantly reducing the probability of underwater accidents. Attached Figure Description

[0009] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments of the present invention will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0010] Figure 1 This is a flowchart illustrating an underwater health risk assessment method provided in one embodiment of this application; Figure 2 This is a schematic diagram of the structure of an underwater health risk assessment device provided in one embodiment of this application; Figure 3 This is a schematic diagram of the structure of a diving mask provided in one embodiment of this application; Figure 4 This is another structural schematic diagram of a diving mask provided in one embodiment of this application; Figure 5 This is a schematic diagram of the structure of five sensing node positions acquired by a multispectral photoplethysmography pulse wave sensing module according to an embodiment of this application; Figure 6 This is a schematic diagram of the structure of a computer device provided in one embodiment of this application.

[0011] The following are the labeling elements in the figure: The diving mask body 10, multispectral photoplethysmography pulse wave sensor module 20, eye tracking module 30, motion sensor module 40, processing module 50, alarm module 60, supplementary lighting strip 70, power supply module 80, and wireless communication module 90. Detailed Implementation

[0012] 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.

[0013] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of this application. However, those skilled in the art will understand that this application may also be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods have been omitted so as not to obscure the description of this application with unnecessary detail.

[0014] It should be understood that, when used in this application specification and the appended claims, the term "comprising" indicates the presence of the described features, integrals, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or a collection thereof.

[0015] It should also be 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.

[0016] As used in this application specification and the appended claims, the term "if" may be interpreted, depending on the context, as "when," "once," "in response to determination," or "in response to detection." Similarly, the phrase "if determined" or "if detected [the described condition or event]" may be interpreted, depending on the context, as meaning "once determined," "in response to determination," "once detected [the described condition or event]," or "in response to detection [the described condition or event]."

[0017] Furthermore, in the description of this application and the appended claims, the terms "first," "second," "third," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0018] References to "one embodiment" or "some embodiments" as described in this specification mean that one or more embodiments of this application include a specific feature, structure, or characteristic described in connection with that embodiment. Therefore, the phrases "in one embodiment," "in some embodiments," "in other embodiments," "in still other embodiments," etc., appearing in different parts of this specification do not necessarily refer to the same embodiment, but rather mean "one or more, but not all, embodiments," unless otherwise specifically emphasized. The terms "comprising," "including," "having," and variations thereof mean "including but not limited to," unless otherwise specifically emphasized.

[0019] It should be understood that the sequence number of each step in the following embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0020] To illustrate the technical solution of this application, specific embodiments are described below.

[0021] like Figure 1 As shown, Figure 1 This is a flowchart illustrating an underwater health risk assessment method according to an embodiment of this application, as shown below. Figure 1 As shown, the underwater health risk assessment method may include the following steps.

[0022] S101: Acquire multi-channel multispectral photoplethysmography (PPG) signals, eye-tracking image sequences, and motion data of the person under test in an underwater environment at any time window.

[0023] In step S101, the time window is a preset fixed-duration time window, which includes multiple sampling points. The multi-channel, multispectral photoplethysmography (PPG) signal characterizes the subcutaneous tissue blood volume changes of the person under test in an underwater environment under illumination by different wavelengths of light. "Multi-channel" indicates simultaneous acquisition at multiple different body parts, and "multispectral" indicates the use of at least two different wavelengths of light for detection. The eye-tracking image sequence reflects the eye movement trajectory, pupil diameter changes, and fixation point distribution of the person under test in the underwater environment. Motion data includes the person under test's three-dimensional acceleration, angular velocity, and attitude angle information in the underwater environment.

[0024] In this embodiment, the person being tested acquires multi-channel, multispectral photoplethysmography (PPG) signals, eye-tracking image sequences, and motion data in an underwater environment using a worn diving mask. The diving mask includes a PPG sensing module, an eye-tracking module, and a motion sensing module. The PPG sensing module acquires PPG signals using a multispectral PPG sensor, such as a reflective PPG sensor, which collects signals of three wavelengths: green, red, and near-infrared. The reflective PPG sensor includes five sensing nodes, such as those in the center of the forehead, the bilateral temporal regions, and both sides of the nose. Each sensing node integrates an LED light source group containing wavelengths emitting green, red, and near-infrared light, and a corresponding photodetector. The eye-tracking module acquires eye-tracking image sequences using a near-infrared image sensor located inside the lens and frame above the nose pad of the diving mask, capturing monocular or binocular motion image sequences at a sampling rate of approximately 90 Hz. The motion sensing module collects motion data through a three-axis accelerometer and a three-axis gyroscope. The motion sensing module obtains the linear acceleration and angular velocity information of the head by setting the output data rate of 200-400 Hz.

[0025] S102: Based on motion data, artifact suppression is performed on the multi-channel multispectral photoplethysmography (PPG) signal to obtain the artifact-suppressed PPG signal.

[0026] In step S102, artifact suppression is to eliminate spurious components generated by interference in the multi-channel multispectral photoplethysmography signal.

[0027] In this embodiment, for a multi-channel multispectral photoplethysmography (PPG) signal and motion data within any time window, artifact suppression is performed on the PPG signal based on the motion data to obtain the artifact-suppressed PPG signal. An adaptive noise cancellation algorithm can be used to process the multi-channel multispectral PPG signal. Using motion data as a reference input, a minimum mean square error adaptive filter is used to estimate and eliminate motion artifact components.

[0028] In this embodiment, by introducing head motion information collected by the motion sensing module as a reference input, the true physiological components and motion-induced interference components in the photoplethysmography (PPG) signal can be effectively distinguished. An adaptive noise cancellation algorithm is employed, using triaxial acceleration and triaxial gyroscope data as reference signals. The filter coefficients are adjusted in real time using the minimum mean square error criterion to dynamically estimate and cancel motion artifacts. This adaptive processing method exhibits better robustness to changes in motion intensity, effectively suppressing large artifacts during vigorous motion while avoiding the loss of true pulse signals due to over-filtering in static or slightly moving states. This significantly improves the quality and reliability of the PPG signal in complex underwater motion scenarios.

[0029] Optionally, based on motion data, artifact suppression is performed on the multi-channel multispectral photoplethysmography (PPG) signal to obtain an artifact-suppressed PPG signal, including: The trained denoising network is obtained, and the multi-channel multispectral photoplethysmography pulse wave signal and motion data are input into the trained denoising network to output the artifact estimation component. Based on the multi-channel multispectral photoplethysmography (PPG) signal and the artifact estimation component, the artifact-suppressed PPG signal is calculated.

[0030] In this embodiment, the motion data includes triaxial acceleration and triaxial angular velocity. A trained denoising network is obtained, and the multi-channel multispectral photoplethysmography (PPG) signal and motion data are input into the trained denoising network to output artifact estimation components, as shown in the following formula: Where i is the channel index, n is the sampling point index within the time window, and k is the time window index. For the i-th channel, the artifact estimation component at the n-th sampling point within the k-th time window. This refers to the multi-channel, multispectral photoplethysmography (PPG) signal of the nth sampling point within the kth time window of the i-th channel. For the motion data of the nth sampling point within k time windows, For a well-trained denoising network, These are the network parameters. The trained denoising network can be obtained by training samples, which include multi-channel multispectral photoplethysmography (PPG) signals, motion data, and PPG signal labels after artifact suppression. The denoising network is then trained under supervised supervision using these training samples to obtain the trained denoising network.

[0031] Based on the multi-channel, multispectral photoplethysmography (PPG) signal and artifact estimation components, the artifact-suppressed PPG signal is calculated. The calculation formula is as follows: in, This represents the image permeasurand pulse wave signal after artifact suppression at the nth sampling point within the kth time window of the i-th channel. This represents the multispectral photoplethysmography (PPG) signal of the nth sampling point within the kth time window of the i-th channel. For the i-th channel, the artifact estimation component of the n-th sampling point within the k-th time window.

[0032] S103: Determine the channel quality index of the multispectral photoplethysmography (PPG) signal for each channel, and perform weighted fusion of the artifact-suppressed PPG signals based on the channel quality index to obtain the fused PPG signal.

[0033] In step S103, the channel quality index characterizes the reliability of the multispectral photoplethysmography (PPG) signal within the channel.

[0034] In this embodiment, the channel quality index of the multispectral photoplethysmography (PPG) signal for each channel is determined. The spectral energy distribution of the PPG signal after artifact suppression can be calculated. The signal quality is evaluated by the energy concentration in the main physiological frequency bands. For example, a fast Fourier transform is performed on the artifact-suppressed signal to calculate the proportion of its energy in the 0.5Hz to 4Hz frequency band to the total energy. The proportion is determined as the channel quality index. The higher the proportion, the more significant the pulse wave component and the better the channel quality.

[0035] After determining the channel quality index of each channel of the multispectral photoplethysmography (PPG) signal, the channel quality index is used as a weight value to perform weighted fusion of all channel signals in the artifact-suppressed PPG signal to obtain the fused PPG signal. The calculation formula is as follows: in, The fused photoplethysmogram signal is the result of sampling points t. Let be the channel quality index of the i-th channel. Let I be the image pulse wave signal with artifact suppression at the t-th sampling point within the k-th time window of the i-th channel, where I is the number of channels.

[0036] Optionally, the channel quality index of each channel of the multispectral photoplethysmography (MPA) signal is determined, including: The physiological characteristics of the multispectral photoplethysmography (PPG) signal of each channel are obtained, and the physiological signal quality index of the corresponding channel PPG signal is calculated based on the physiological characteristics. Underwater coupling characteristics of each channel's multispectral photoplethysmography (PPG) signal are obtained, and the underwater optical coupling index of the corresponding channel's PPG signal is calculated based on the underwater coupling characteristics. Based on the physiological signal quality index and underwater optical coupling index of each channel, the channel quality index of the corresponding channel photoplethysmography (PPG) wave signal is calculated.

[0037] In this embodiment, the physiological characteristics of the multispectral photoplethysmography (PPG) signal of each channel are acquired, and the physiological signal quality index of the corresponding channel's PPG signal is calculated based on these physiological characteristics. The physiological signal quality index is used to characterize the discernibility of the physiological pulsation in that channel.

[0038] Physiological characteristics of the multispectral photoplethysmography (MPP) wave signal for each channel were acquired. These characteristics included the normalized signal-to-noise ratio (SNR), waveform morphological similarity between adjacent pulse cycles, and perfusion index. The SNR characterizes the strength of the effective pulsatile component relative to background noise within the heart rate-related frequency band. The waveform morphological similarity between adjacent pulse cycles characterizes the morphological consistency and stability of consecutive pulse cycles within the same channel, reflecting the rhythmic regularity of the physiological pulsation signal. The perfusion index characterizes the ratio of the pulsatile component to the DC component, reflecting the intensity of the pulsatile AC component relative to the DC baseline.

[0039] Based on physiological characteristics, the physiological signal quality index of the corresponding channel photoplethysmography (PPG) wave signal is calculated using the following formula: in, is the physiological signal quality index of the i-th channel, used to characterize the discernibility of physiological pulsations in that channel. Let be the normalized signal-to-noise ratio of the i-th channel. Let represent the waveform similarity between adjacent pulse cycles in the i-th channel. Let i be the perfusion index of the i-th channel. , , The weight coefficients are non-negative and satisfy the following conditions: The sum of is 1.

[0040] Underwater coupling characteristics of each channel's multispectral photoplethysmography (PPG) signal were acquired. These characteristics included DC baseline drift, saturation clipping, discernibility of heart rate-related AC pulsation components, and relative contrast stability between AC and DC. Specifically, DC baseline drift characterizes the degree of low-frequency drift caused by changes in sensor-skin coupling; saturation clipping characterizes the strength of signal overexposure or clipping; discernibility of heart rate-related AC pulsation components characterizes the ability to accurately extract and identify the characteristic pulsation waveform generated by periodic changes in blood volume driven by the cardiac cycle from complex background interference; and relative contrast stability between AC and DC characterizes the degree of fluctuation in the ratio of the AC pulsation component amplitude to the DC baseline component amplitude.

[0041] Based on the underwater coupling characteristics, the underwater optical coupling index of the corresponding channel photoplethysmography pulse wave signal is calculated; the calculation formula is as follows: in, denoted as the underwater optical coupling index of the i-th channel, used to characterize the effects of underwater coupling variations on DC drift, saturation clipping, and relative contrast stability. Let be the DC baseline drift of the i-th channel. The saturation clipping level of the i-th channel. The distinguishability of the heart rate-related communication pulsation component in the i-th channel. The relative contrast stability between AC and DC for the i-th channel. , , and The weight coefficients are non-negative and satisfy the following conditions: , , and The sum of is 1.

[0042] Based on the physiological signal quality index and underwater optical coupling index of each channel, the channel quality index of the corresponding channel's photoplethysmography (PPG) wave signal is calculated using the following formula: in, For the first i Channel quality index. denoted as the physiological signal quality index of the i-th channel. For the first i The underwater optical coupling index of the channel, and The weight coefficients are non-negative and satisfy the following conditions: and The sum of is 1, and I is the number of channels.

[0043] S104: Based on the fused photoplethysmography (PPG) signal, multi-channel multispectral PPG signal, and motion data, the physiological parameters of the person to be tested are predicted, and the physiological parameter prediction results are obtained.

[0044] In step S104, physiological parameters characterize the body's state, such as heart rate, blood oxygen saturation, and blood pressure trends.

[0045] In this embodiment, when predicting the physiological parameters of the person to be tested based on the fused photoplethysmography (PPG) signal, the multi-channel multispectral PPG signal, and motion data, a physiological parameter prediction model can be used. The fused PPG signal, the multi-channel multispectral PPG signal, and the motion data are input into the physiological parameter prediction model, which outputs the physiological parameter prediction results. The physiological parameter prediction model is a deep neural network architecture based on multi-task learning. This model adopts an encoder-decoder structure. The encoder part consists of a cascaded one-dimensional convolutional neural network and a bidirectional long short-term memory network, used to extract the time-frequency features of the fused PPG signal, the spatial spectral features of the multi-channel multispectral PPG signal, and the temporal dynamic features of the motion data, respectively. The decoder part adopts a multi-task output structure, setting independent prediction branches for different physiological parameter prediction tasks.

[0046] Optionally, physiological parameters include heart rate, blood oxygen saturation, and blood pressure trends; the physiological parameter prediction results include heart rate prediction results, blood oxygen saturation prediction results, and blood pressure trend prediction results. Based on the fused photoplethysmography (PPG) signal, multi-channel multispectral PPG signal, and motion data, the physiological parameters of the person being tested are predicted, yielding the predicted physiological parameters, including: Based on motion data, motion artifact suppression is performed on the fused photoplethysmography (PPG) signal to obtain the suppressed PPG signal. A candidate peak set of the suppressed PPG signal is extracted, and the heart rate prediction result is determined based on the candidate peak set. Based on the multi-channel multispectral photoplethysmography pulse wave signal, the red light signal and near-infrared signal of each channel are extracted, and the blood oxygen saturation prediction result is determined based on the red light signal and near-infrared signal of each channel. Feature extraction is performed on the fused photoplethysmography (PPG) signal to obtain signal features. Blood pressure trend prediction is then performed based on these signal features to obtain the blood pressure trend prediction results.

[0047] In this embodiment, the motion state of the person being tested is determined based on motion data, resulting in a motion state determination result. Based on the motion state determination result, motion artifact suppression is applied to the fused photoplethysmography (PPG) signal to obtain a suppressed PPG signal. A candidate peak set of the suppressed PPG signal is extracted, and a heart rate prediction result is determined based on the candidate peak set. The motion state includes a stationary state, a uniform motion state, and a vigorous motion state. For example, the motion data includes triaxial acceleration and triaxial angular velocity. Based on the triaxial acceleration, the acceleration amplitude is synthesized using spatial vectors. If the synthesized acceleration amplitude is less than a preset first motion intensity threshold, the person being tested is determined to be in a stationary state, and a stationary state determination result is output. If the synthesized acceleration amplitude is greater than or equal to the first motion intensity threshold, the synthesized angular velocity amplitude is obtained by synthesizing the triaxial angular velocity using spatial vectors. If the synthesized angular velocity amplitude is less than a preset second motion intensity threshold, the person being tested is determined to be in a uniform motion state, and a uniform motion state determination result is output. If the amplitude of the synthesized angular velocity is greater than or equal to the second motion intensity threshold, the person being tested is determined to be in a state of intense motion, and the result of the determination is output. Other methods can also be used for determination, and this embodiment does not limit them.

[0048] When the motion state determination result is a stationary state, a low-pass filter is used to filter the fused photoplethysmography (PPG) signal to obtain a suppressed PPG signal. When the motion state determination result is a uniform motion state, an adaptive noise cancellation algorithm is used to process the fused PPG signal to obtain an artifact-suppressed PPG signal. When the motion state determination result is a violent motion state, a multi-scale wavelet decomposition and reconstruction algorithm is used to suppress motion artifacts in the fused PPG signal to obtain an artifact-suppressed PPG signal.

[0049] A candidate spectral peak set is extracted from the suppressed photoplethysmography (PPG) signal. Based on this set, the heart rate prediction result is determined. Specifically, the suppressed PPG signals at all sampling points within a corresponding time window are identified. These suppressed PPG signals are then subjected to frequency domain transformation to obtain the signal spectrum within that time window. Based on this spectrum, a candidate spectral peak set is extracted. The heart rate prediction result is then determined based on this candidate spectral peak set. The calculation formula is as follows: in, Let be the spectrum of the t-th time window, and let the candidate spectral peak set be . , As a priori, Let be the likelihood function, describing the likelihood at a heart rate of [missing information]. At that time, the current spectrum was observed. The probability, The posterior probability is used to determine the main heart rate peak from the candidate spectral peak set by maximizing the posterior probability, and the heart rate prediction result is determined based on the frequency of the main heart rate peak.

[0050] It should be noted that the likelihood function integrates the amplitude, significance, signal-to-noise ratio, and harmonic consistency of candidate spectral peaks, and introduces a proximity penalty term between the candidate frequency and the acceleration spectrum features to suppress motion-frequency spurious peaks. For candidate spectral peaks adjacent to the dominant acceleration frequency and its harmonics, the function reduces their posterior probability, thereby reducing the interference of motion-frequency spurious peaks on the selection of the dominant peak. This can be expressed in log-likelihood form as follows: in, For the amplitude of the spectral peak, Signal-to-noise ratio of candidate spectral peaks. Energy at the harmonic level, To accelerate the main frequency, As a proximity penalty term, , , , These are the weighting coefficients. Represents the second and third harmonics.

[0051] Select the peak heart rate based on the maximum a posteriori criterion: in, For the t-th time window, the peak heart rate frequency selected using the maximum a posteriori criterion is... To convert the corresponding heart rate frequency, multiply the peak heart rate frequency by 60 to the clinically commonly used heart rate unit (beats per minute, BPM).

[0052] Simultaneously, the confidence level of the main peak is calculated. When the confidence level is lower than a preset threshold, the asymmetric smoothing prediction output heart rate estimate based on the proximity penalty term is enabled to ensure the continuity and physiological rationality of the heart rate output.

[0053] Based on multi-channel, multispectral photoplethysmography (PPG) signals, the red and near-infrared signals of each channel are extracted. The predicted blood oxygen saturation is then determined based on these signals. The predicted blood oxygen saturation can be determined using the AC and DC components of each signal. Specifically, the DC component is determined based on the red and near-infrared signals, and the formula for calculating the DC component is as follows: in, The DC component, The length of the time window is the number of sampling points within the time window. For the j-th channel at wavelength The discrete sampling sequence below, The wavelength of red light. This refers to the wavelength of near-infrared light.

[0054] To suppress low-frequency drift, the AC component uses "narrowband energy after DC removal," where the heart rate fundamental frequency is used. Centered on the signal, the AC pulsating energy is extracted as the AC component. First, the DC component is removed within each time window, then the signal is windowed and subjected to a Discrete Fourier Transform. The calculation formula is as follows: in, For the j-th channel at wavelength The discrete sampling sequence below, The DC component, For time window function, , is the imaginary unit. For a complex exponent, Here, n is the frequency index, and n is the sampling point index. The length of the time window is the number of sampling points within the time window. For the frequency index corresponding to the fundamental frequency, take the nearest integer. Sampling rate, For the purpose of exchanging quantities, For the j-th channel, wavelength , No. Frequency domain spectrum values ​​at each frequency point Here, H represents the harmonic weight, and H is the highest order harmonic. The harmonic order is... For the j-th channel, wavelength , No. The frequency point, the first Spectral amplitude at the second harmonic.

[0055] Calculate the ratio of the AC component to the DC component of each signal using the following formula: in, Let be the ratio of the red light signal to the near-infrared light signal in the j-th channel. For the AC component of the red light in the j-th channel, Let be the DC component of the red light in the j-th channel. Let j be the AC component of the near-infrared light in the j-th channel. denoted as the DC component of the near-infrared light in the j-th channel.

[0056] The ratios of multiple channels are weighted and fused. A joint signal quality index is constructed for each paired channel, and the joint signal quality index is determined as the weight value of the corresponding channel. The joint signal quality index is calculated based on spectral concentration, perfusion index, waveform consistency, and motion penalty, as shown in the following formula: in: Let be the spectral concentration of the j-th channel, used to measure the concentration of the center-slap correlation energy of the near-infrared channel across the entire frequency band, and Ω be the normalized set of frequency band indices. For the j-th channel, near-infrared wavelength , No. spectral amplitude at each frequency point; For the j-th channel, near-infrared wavelength , No. The frequency point, the first Spectral amplitude at the second harmonic; Harmonics ; This is the perfusion index; the other parameters are the same as above. For the time delay of red light and near-infrared light in channel j The maximum correlation coefficient is used to characterize alignment consistency; For the correlation coefficient operator; For the j-th channel red light signal, Let be the mean value of the red light signal in the j-th channel. The near-infrared light signal of channel j is shifted to the right as a whole. One time unit Let be the mean value of the near-red light signal in the j-th channel. Indicates a time delay operation. The time step for the delay; It is a function of standard deviation. In for , In for ; Let j be the motion penalty factor for the j-th channel. This is the penalty coefficient; The root mean square intensity of the IMU acceleration within the time window; Let j be the quality index of the j-th channel; , , , , where is greater than or equal to 0, is the exponential weight; ∈[0,1] represents the channel weight; The ratio after fusion; This is a joint signal quality index, where I is the number of channels. The ratio of the red light signal to the near-infrared light signal in the j-th channel is denoted as . The value is a weighted fusion of the ratios from multiple channels.

[0057] After fusion Substituting the values ​​into a preset quadratic polynomial calibration function, the blood oxygen saturation value is calculated and output. Boundary constraints are then applied to obtain the predicted blood oxygen saturation result. The calculation formula is as follows: in: This refers to blood oxygen saturation values. , , The parameters of the polynomial calibration function are to be determined by... Substitute the value into In the process, blood oxygen saturation values ​​are obtained, and boundary constraints are applied to limit the results to a clinically reliable range. For blood oxygen saturation prediction results, The clipping function restricts the input x to a certain value. Within the range, that is, the predicted blood oxygen saturation is limited to between 70 and 100.

[0058] Feature extraction is performed on the fused photoplethysmography (PPG) signal to obtain signal features. Blood pressure trend prediction is then performed based on these signal features to obtain the blood pressure trend prediction results.

[0059] Feature extraction is performed on the fused photoplethysmography (PPG) signal to obtain signal features. Specifically, the fused PPG signal is preprocessed to suppress baseline drift and high-frequency noise. The preprocessed fused PPG signal is then segmented into pulse cycles, and a set of blood pressure-related features, including at least time-domain features, derivative-domain features, and frequency-domain features, is extracted to obtain signal features.

[0060] It should be noted that before feature extraction, a correlation coefficient-based CFS algorithm is used to initially screen the features, obtaining a feature subset for blood pressure trend prediction. In this embodiment, the feature subset includes time-domain features, derivative-domain features, and frequency-domain features. The time-domain features, derivative-domain features, and frequency-domain features obtained from feature extraction of the fused photoplethysmography (PPG) signal are input into the blood pressure trend parameters output by the regression model. The output results are then smoothed using a time window to obtain the blood pressure trend prediction result. The regression model can be an ensemble learning model based on the random forest algorithm. This model reduces the risk of overfitting of a single model through a voting mechanism of multiple decision trees, improving the stability and generalization ability of blood pressure trend prediction. The blood pressure trend parameters output by the regression model are smoothed using a time window to eliminate instantaneous noise interference and enhance the continuity of the prediction results.

[0061] S105: Extract features from each frame of the eye-tracking image to obtain the pupil features of each frame of the eye-tracking image. Calculate the eye-tracking feature vector based on the pupil features of each frame of the eye-tracking image.

[0062] In step S105, the eye-tracking feature vector is used to characterize the visual attention allocation pattern and cognitive load state in the underwater environment.

[0063] In this embodiment, feature extraction is performed on each frame of the eye-tracking image to obtain the pupil features of each frame. The iris boundary detection algorithm can be applied to locate the pupil region, and the ellipse fitting method is used to extract geometric parameters such as the pupil center coordinates, the length of the pupil major axis and minor axis, the pupil area, and the pupil ellipticity.

[0064] Based on the pupil features of each frame of eye-tracking image, an eye-tracking feature vector is calculated. The temporal change of the pupil center coordinates can be calculated based on a continuous frame sequence to extract eye-tracking trajectory parameters such as fixation point drift velocity, saccade amplitude, and fixation duration. The eye-tracking feature vector is then constructed based on these parameters.

[0065] Optionally, feature extraction is performed on each frame of the eye-tracking image to obtain the pupil features of each frame, including: The eye-tracking images are segmented to determine the pupil region of the person being tested; Based on the pupil region, the pupil center coordinates, pupil diameter, and pupil area are calculated. The pupil center coordinates, pupil diameter, and pupil area are the pupil characteristics.

[0066] In this embodiment, eye-tracking images are segmented to determine the pupil region of the person being detected. A combination of image thresholding and edge detection can be used to achieve precise localization of the pupil region. For example, the eye-tracking image is converted to a grayscale image. Utilizing the grayscale differences between the pupil region and the iris and sclera, an adaptive thresholding algorithm is used to initially separate candidate regions with low grayscale values. Considering the potential for local contrast variations due to uneven underwater lighting, a local adaptive thresholding strategy is introduced. The segmentation threshold is dynamically adjusted based on the grayscale distribution of different regions of the image to enhance robustness under complex lighting conditions. Based on the initial segmentation, the Canny edge detection operator is applied to extract the contour information of the candidate regions. Since the pupil is approximately elliptical, the Hough transform ellipse detection algorithm is used to fit the edge points, selecting closed contours that conform to elliptical geometry and have an area within a reasonable range as the pupil boundary. Morphological opening operations are used to remove small noise points, and connected component analysis is used to retain the largest and most regularly shaped connected regions, ultimately determining the precise location of the pupil region.

[0067] Based on the pupil region, the pupil center coordinates, pupil diameter, and pupil area are calculated. These three parameters constitute the pupil features. The pupil center coordinates are the centroid coordinates of the pupil region. The diameter of the equivalent circle is calculated from the pupil area, yielding the pupil diameter. The pupil area represents the total number of pixels in the pupil region.

[0068] Optionally, an eye movement feature vector is calculated based on the pupil features of each frame of the eye movement image, including: Based on the pupil center coordinates, pupil diameter, and pupil area, the average pupil diameter, standard deviation of pupil diameter fluctuation, mean pupil diameter change rate, number of fixation frames, average fixation time, number of saccade frames, saccade frequency, saccade path length, and blink frequency are calculated. The average pupil diameter, standard deviation of pupil diameter fluctuation, mean pupil diameter change rate, number of fixation frames, average fixation time, number of saccade frames, saccade frequency, saccade path length, and blink frequency constitute the eye movement feature vector.

[0069] In this embodiment, based on the pupil diameter of each frame of eye-tracking image, the average pupil diameter, the standard deviation of pupil diameter fluctuation, and the mean rate of change of pupil diameter are calculated. The formulas for calculating the average pupil diameter, the standard deviation of pupil diameter fluctuation, and the mean rate of change of pupil diameter are as follows: in, The average pupil diameter, Let N be the pupil diameter of the eye-tracking image in the t-th frame within the corresponding time window, and N be the total number of eye-tracking images within the corresponding time window. The standard deviation of pupil diameter fluctuation. The mean rate of change of pupil diameter. The frame interval duration.

[0070] Based on the pupil center coordinates and pupil area of ​​each frame of the eye-tracking image, the saccade frequency, saccade path length, and blink frequency of the subject are calculated. The angular velocity of each frame of the eye-tracking image is calculated using the following formula: in, For the first t The angular velocity of a frame of eye-tracking image, that is, the magnitude of the displacement vector of the pupil center per unit time, is the instantaneous angular velocity of eye movement. , For the first t The pupil center coordinates of a frame of eye-tracking image, i.e., the x and y coordinates of the pupil center. , Let be the coordinates of the pupil center in the (t-1)th frame of the eye-tracking image, i.e., the x and y coordinates of the pupil center. The frame interval duration.

[0071] Each frame is categorized by label based on its angular velocity. When the value is less than a preset threshold, it is marked as a gaze frame. A frame not less than a preset threshold is marked as a saccade frame. The number of frames in a fixation is determined as the fixation frame count. The total fixation duration is calculated based on the number of fixation frames and the frame interval duration. The average fixation time is obtained by dividing the total fixation duration by the number of fixation frames. The number of frames in a saccade frame is determined as the saccade frame count. The saccade frequency is calculated based on the ratio of the saccade frame count to the duration of the corresponding time window. The formula for calculating the saccade path length is as follows: in, To scan the path length, Let be the i-th consecutive saccade interval, i.e., the number of frames in the i-th consecutive saccade interval. Let t be the image of the i-th consecutive saccade interval. This represents the change in the x-coordinate of the pupil center between frame t and frame (t-1). This represents the change in the vertical coordinate of the pupil center between frame t and frame t-1.

[0072] Blinking events can be identified by detecting a sudden decrease in the pixel area of ​​the pupil region within consecutive frames. For example, when the pixel area of ​​the pupil region decreases by more than half relative to the reference area, it is determined as a blinking event. Based on the blinking events, the blinking frequency is calculated, which is the ratio of the number of blinking events to the duration of the corresponding time window.

[0073] S106: Based on the physiological parameter prediction results and eye movement feature vector within the time window, predict the comprehensive state index of the person to be tested, and obtain the comprehensive state index prediction results and confidence level.

[0074] In step S106, the comprehensive state index is used to quantitatively represent the stress level, cognitive load, or panic state.

[0075] In this embodiment, a preset comprehensive state index prediction model is obtained. The physiological parameter prediction results and eye movement feature vectors within the corresponding time window are input into the preset comprehensive state index prediction model, and the comprehensive state index prediction results and confidence scores are output. The preset comprehensive state index prediction model is a regression model. The comprehensive state index prediction result S∈[0,1].

[0076] S107: Based on the prediction results and confidence levels of the comprehensive status indicators within all time windows, assess the health risks of the personnel to be tested and obtain the health risk assessment results.

[0077] In step S107, the health risk assessment results characterize the overall health risk level of the person being tested in the underwater environment.

[0078] In this embodiment, based on the prediction results and confidence levels of the comprehensive status index, the effective window-level risk value for the corresponding time window is calculated. Based on the effective window-level risk value, the cumulative health risk score is calculated using the following formula: in, The effective window level risk value for the k-th time window. Let be the confidence level for the k-th time window. The result is the prediction of the comprehensive state index for the k-th time window. The cumulative health risk score for the (k-1)th time window. The cumulative health risk score for the k-th time window. The forgetting factor has a value of This is used to characterize the degree to which historical risks retain the current risk score. For persistent anomalies, The weighting coefficients for persistent outliers; The duration of continuous anomalies up to the k-th time window. The duration normalization threshold, This means that the result is restricted to the interval [0,1].

[0079] It should be noted that when the effective window level risk value When the value exceeds a preset threshold, the current time window is determined to be in an abnormal state, and the duration of the continuous abnormality is set to [a certain value]. Increasing, when the effective window level risk value When the duration of continuous anomalies is not greater than a preset threshold, set the duration of continuous anomalies to... Reset to 0, thereby reducing the cumulative health risk score. It also reflects the current time window status, the historical status of the preceding time window, and the duration of the anomaly.

[0080] Based on cumulative health risk score The current health level is determined by comparing the result with a preset level threshold. For example, when... When, it is judged as normal; when When, it is judged as being followed; when When, it is judged as an early warning; when When the condition is met, it is determined to be an emergency; where θ1, θ2, and θ3 are the attention threshold, warning threshold, and emergency threshold, respectively, and satisfy the following conditions: .

[0081] When the health level reaches the attention threshold, a prompt message is output and the status is reported at a low frequency; when the health level reaches the warning threshold, a visual warning message is output and the status reporting frequency is increased; when the health level reaches the emergency threshold, a strong visual alarm is output and an emergency alarm frame is generated. The emergency alarm frame includes a timestamp, health level, cumulative health risk score, blood oxygen saturation prediction result, heart rate prediction result, comprehensive status index prediction result, and confidence level.

[0082] Please see Figure 2 , Figure 2 This is a schematic diagram of an underwater health risk assessment device according to an embodiment of this application. This underwater health risk assessment device corresponds one-to-one with the underwater health risk assessment method described in the above embodiments. Please refer to [link / reference] for details. Figure 2 as well as Figure 2 The relevant descriptions in the corresponding embodiments are shown below. For ease of explanation, only the parts relevant to this embodiment are shown. See also... Figure 2 The underwater health risk assessment device 200 includes: an acquisition module 201, an inhibition module 202, a fusion module 203, a physiological parameter prediction module 204, an eye movement feature extraction module 205, a comprehensive state index prediction module 206, and an assessment module 207.

[0083] The acquisition module 201 is used to acquire multi-channel multispectral photoplethysmography (PPG) signals, eye-tracking image sequences, and motion data of the person to be tested within any time window in the underwater environment.

[0084] The suppression module 202 is used to suppress artifacts in multi-channel multispectral photoplethysmography (PPG) signals based on motion data, so as to obtain the PPG signal after artifact suppression.

[0085] The fusion module 203 is used to determine the channel quality index of the multispectral photoplethysmography (PPG) signal for each channel, and to perform weighted fusion of the artifact-suppressed PPG signals based on the channel quality index to obtain the fused PPG signal.

[0086] The physiological parameter prediction module 204 is used to predict the physiological parameters of the person to be tested based on the fused photoplethysmography signal, the multi-channel multispectral photoplethysmography signal and motion data, and obtain the physiological parameter prediction results. The eye movement feature extraction module 205 is used to extract features from each frame of the eye movement image to obtain the pupil features of each frame of the eye movement image, and to calculate the eye movement feature vector based on the pupil features of each frame of the eye movement image.

[0087] The comprehensive state index prediction module 206 is used to predict the comprehensive state index of the person to be tested based on the physiological parameter prediction results and eye movement feature vector within the time window, and obtain the comprehensive state index prediction result.

[0088] The assessment module 207 is used to assess the health risk of the personnel to be tested based on the prediction results of the comprehensive status indicators within all time windows, and obtain the health risk assessment results.

[0089] Optionally, the suppression module 202 includes: The first acquisition unit is used to acquire the trained denoising network, inputting the multi-channel multispectral photoplethysmography pulse wave signal and motion data into the trained denoising network, and outputting artifact estimation components.

[0090] The first calculation unit is used to calculate the artifact-suppressed photoplethysmography (PPG) signal based on the multi-channel multispectral PPG signal and the artifact estimation component.

[0091] Optionally, the fusion module 203 includes: The second calculation unit is used to acquire the physiological characteristics of the multispectral photoplethysmography (PPG) signal of each channel, and calculate the physiological signal quality index of the corresponding channel PPG signal based on the physiological characteristics.

[0092] The third calculation unit is used to obtain the underwater coupling characteristics of the multispectral photoplethysmography (PPG) signal of each channel, and calculate the underwater optical coupling index of the corresponding channel PPG signal based on the underwater coupling characteristics.

[0093] The fourth calculation unit is used to calculate the channel quality index of the corresponding channel photoplethysmography signal based on the physiological signal quality index and the underwater optical coupling index of each channel.

[0094] Optionally, the physiological parameter prediction module 204 includes: The first determining unit is used to perform motion artifact suppression on the fused photoplethysmography (PPG) signal based on motion data to obtain the suppressed PPG signal, extract the candidate spectral peak set of the suppressed PPG signal, and determine the heart rate prediction result based on the candidate spectral peak set.

[0095] The second determining unit is used to extract the red light signal and near-infrared signal of each channel based on the multi-channel multispectral photoplethysmography pulse wave signal, and to determine the blood oxygen saturation prediction result based on the red light signal and near-infrared signal of each channel.

[0096] The prediction unit is used to extract features from the fused photoplethysmography pulse wave signal, obtain signal features, and predict blood pressure trends based on the signal features to obtain blood pressure trend prediction results.

[0097] Optionally, the eye-tracking feature extraction module 205 includes: The segmentation unit is used to segment the eye-tracking image and determine the pupil region of the person being tested.

[0098] The fifth calculation unit is used to calculate the pupil center coordinates, pupil diameter, and pupil area based on the pupil region. The pupil center coordinates, pupil diameter, and pupil area are pupil characteristics.

[0099] Optionally, the eye-tracking feature extraction module 205 includes: The sixth calculation unit is used to calculate the average pupil diameter, standard deviation of pupil diameter fluctuation, mean pupil diameter change rate, number of fixation frames, average fixation time, number of saccade frames, saccade frequency, saccade path length, and blink frequency based on the pupil center coordinates, pupil diameter, and pupil area. The average pupil diameter, standard deviation of pupil diameter fluctuation, mean pupil diameter change rate, number of fixation frames, average fixation time, number of saccade frames, saccade frequency, saccade path length, and blink frequency constitute the eye movement feature vector.

[0100] Please see Figure 3 , Figure 3This is a schematic diagram of the structure of a diving mask provided in one embodiment of this application. Figures (a) and (b) are schematic diagrams of the diving mask from different perspectives. The diving mask includes a diving mask body 10, a multispectral photoplethysmography (PPG) sensor module 20, an eye-tracking module 30, a motion sensing module 40, and a processing module 50. The PPG sensor module 20, eye-tracking module 30, motion sensing module 40, and processing module 50 are disposed on the diving mask body 10. The processing module 50 is electrically connected to the PPG sensor module 20, eye-tracking module 30, and motion sensing module 40, respectively. The PPG sensor module 20 is used to collect multispectral PPG signals and transmit them to the processing module 50. The eye-tracking module 30 is used to collect eye-tracking image sequences and transmit them to the processing module 50. The motion sensing module 40 is used to collect motion data and transmit it to the processing module 50. The processing module 50 is used to execute the underwater health risk assessment method described in any embodiment.

[0101] Please see Figure 4 , Figure 4 This is another structural schematic diagram of a diving mask provided in one embodiment of this application. Figures (a) and (b) are structural schematic diagrams of the diving mask from different perspectives. The diving mask also includes an alarm module 60, a supplementary lighting strip 70, a power module 80, and a wireless communication module 90. The alarm module 60, supplementary lighting strip 70, power module 80, and wireless communication module 90 are disposed on the diving mask body 10. The processing module 50 is electrically connected to the alarm module 60, supplementary lighting strip 70, and wireless communication module 90 respectively. The alarm module 60 receives the drive signal generated by the processing module 50 based on the health risk assessment result and outputs a visual alarm prompt. The supplementary lighting strip 70 receives the trigger command sent by the processing module 50 and adjusts the color temperature and brightness. The wireless communication module 90 transmits the health risk assessment result received by the processing module 50. The power module 80 provides power to each module.

[0102] The sensor in the multispectral photoplethysmography (PPG) sensor module 20 is a multispectral PPG sensor, specifically a reflective PPG sensor. It includes a light-emitting unit and a photodetector unit. The light-emitting unit comprises LEDs emitting green, red, and near-infrared light wavelengths. The photodetector unit receives the light signal reflected from the facial tissue of the person being tested and outputs an electrical signal related to blood flow changes. The optical physiological signal acquisition circuit of the multispectral PPG sensor uses an optical pulse oximetry and heart rate sensing front-end chip. The multispectral PPG sensor module 20 includes five sensing nodes; please refer to [link to relevant documentation]. Figure 5 , Figure 5This is a schematic diagram illustrating the structural positions of five sensing nodes acquired by a multispectral photoplethysmography (PPG) sensing module according to an embodiment of this application. These nodes are embedded in the contact areas between the diving mask and the center of the forehead, the bilateral temples, and the sides of the nose. Each sensing node integrates an LED light source group emitting green, red, and near-infrared light wavelengths, along with a corresponding photodetector.

[0103] The eye-tracking module 30 includes two near-infrared image sensors, which are located on the inside of the lens where the nose pad of the diving mask is embedded with the lens, and capture eye movement image sequences of the user's single eye or both eyes at a sampling rate of approximately 90 Hz.

[0104] Preferably, the near-infrared image sensor uses a CMOS image sensor chip with global shutter and near-infrared enhancement characteristics. By synchronously controlling the image acquisition timing and near-infrared emission pulses, pupil localization, gaze point estimation and blink detection at high frame rates can be achieved.

[0105] The motion sensing module 40 is used to collect head motion information of the person being tested. It is set on the frame in the center of the forehead and includes a three-axis accelerometer and a three-axis gyroscope. By setting the output data rate of 200 to 400 Hz, it obtains the linear acceleration and angular velocity information of the head, which is used for motion state classification and adaptive suppression of motion artifacts in multispectral photoplethysmography pulse wave signals.

[0106] The processing module 50 is based on a microcontroller. It is electrically connected to the multispectral photoplethysmography (PPG) sensor module 20, eye-tracking module 30, and motion sensor module 40 via I²C, SPI, UART, a digital camera interface, and several general-purpose input / output ports. This allows for unified management and scheduling of the multimodal signals output by each sensor module. The processing module 50 is configured to execute program instructions stored in its internal or external memory. Within a preset time window, it performs synchronous acquisition, preprocessing, feature extraction, and fusion modeling of multi-channel multispectral PPG signals, eye-tracking image sequences, and motion data. Through multiple collaborative algorithm modules, it performs joint analysis of heart rate, blood oxygen saturation, blood pressure trend parameters, and eye-tracking behavior characteristics, generating a comprehensive state index characterizing stress levels, cognitive load, and panic levels. Based on this, it outputs corresponding control signals to the alarm module 60.

[0107] The wireless communication module 90 is preferably a magnetically coupled communication module, integrating a transmitting coil, a receiving coil, and a modulation / demodulation circuit, achieving underwater short-range data transmission through low-frequency magnetic field coupling. Optionally, the wireless communication module 90 is a capacitively coupled communication module, achieving underwater data transmission through electric field coupling. The wireless communication module 90 performs feature-level compression and frame encapsulation on the transmitted content and can employ error correction and retransmission mechanisms; in emergency situations, it can increase the reporting frequency or repeatedly send SOS alarm frames to improve reachability.

[0108] It should be noted that the information interaction and execution process between the above-mentioned units are based on the same concept as the method embodiments of this application. For details on their specific functions and technical effects, please refer to the method embodiments section, which will not be repeated here.

[0109] Figure 6 This is a schematic diagram of the structure of a computer device provided in one embodiment of this application. For example... Figure 6 As shown, the computer device of this embodiment includes: at least one processor ( Figure 6 Only one is shown in the diagram), a memory, and a computer program stored in the memory and executable on at least one processor, wherein the processor executes the computer program to implement the steps in any of the above embodiments of the underwater health risk assessment methods.

[0110] This computer device may include, but is not limited to, a processor and memory. Those skilled in the art will understand that... Figure 6 The examples of computer devices are merely examples and do not constitute a limitation on computer devices. Computer devices may include more or fewer components than shown in the illustration, or combinations of certain components, or different components, such as network interfaces, displays, and input devices.

[0111] The processor referred to can be a CPU, but it can also be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor can be a microprocessor or any conventional processor.

[0112] Memory includes readable storage media, internal memory, etc., wherein internal memory can be the RAM of a computer device, providing an environment for the operation of the operating system and computer-readable instructions stored in the readable storage media. The readable storage media can be the hard drive of a computer device, or in other embodiments, it can be an external storage device of the computer device, such as a plug-in hard drive, Smart Media Card (SMC), Secure Digital (SD) card, or Flash Card. Furthermore, memory can include both internal storage units and external storage devices of the computer device. Memory is used to store the operating system, applications, bootloader, data, and other programs, such as program code for computer programs. Memory can also be used to temporarily store data that has been output or will be output.

[0113] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is used as an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiments can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit. Furthermore, the specific names of the functional units and modules are only for easy differentiation and are not intended to limit the scope of protection of this application. The specific working process of the units and modules in the above device can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here. If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments of this application can be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the above method embodiments. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. A computer-readable medium can include at least: any entity or device capable of carrying computer program code, a recording medium, a computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media. Examples include USB flash drives, portable hard drives, magnetic disks, or optical disks. In some jurisdictions, according to legislation and patent practice, computer-readable media cannot be electrical carrier signals or telecommunication signals.

[0114] The implementation of all or part of the processes in the methods of the above embodiments can also be accomplished by a computer program product. When the computer program product is run on a computer device, it causes the computer device to execute the steps in the above method embodiments.

[0115] In the above embodiments, the descriptions of each embodiment have different focuses. For parts that are not described in detail or recorded in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0116] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0117] In the embodiments provided in this application, it should be understood that the disclosed apparatus / computer devices and methods can be implemented in other ways. For example, the apparatus / computer device embodiments described above are merely illustrative. For instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0118] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0119] The above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should all be included within the protection scope of this application.

Claims

1. A method for underwater health risk assessment, characterized in that, The underwater health risk assessment method includes: Acquire multi-channel, multispectral photoplethysmography (PPG) signals, eye-tracking image sequences, and motion data of the person being tested within any time window in an underwater environment; Based on the motion data, artifact suppression is performed on the multi-channel multispectral photoplethysmography signal to obtain the artifact-suppressed photoplethysmography signal. The channel quality index of each channel multispectral photoplethysmography (PPI) signal is determined. Based on the channel quality index, the artifact-suppressed PPI signals are weighted and fused to obtain the fused PPI signals. Based on the fused photoplethysmography (PPG) signal, the multi-channel multispectral PPG signal, and the motion data, the physiological parameters of the person to be tested are predicted to obtain the physiological parameter prediction results. Feature extraction is performed on each frame of the eye-tracking image to obtain the pupil features of each frame of the eye-tracking image. Based on the pupil features of each frame of the eye-tracking image, the eye-tracking feature vector is calculated. Based on the physiological parameter prediction results within the time window and the eye movement feature vector, the comprehensive state index of the person to be tested is predicted, and the comprehensive state index prediction result and confidence level are obtained. Based on the prediction results of the comprehensive status index within all time windows and the confidence level, the health risk of the person to be tested is assessed, and the health risk assessment result is obtained.

2. The underwater health risk assessment method as described in claim 1, characterized in that, The process of performing artifact suppression on the multi-channel multispectral photoplethysmography (PPG) signal based on the motion data to obtain an artifact-suppressed PPG signal includes: The trained denoising network is obtained, and the multi-channel multispectral photoplethysmography pulse wave signal and the motion data are input into the trained denoising network to output the artifact estimation component. Based on the multi-channel multispectral photoplethysmography (PPG) signal and the artifact estimation component, the artifact-suppressed PPG signal is calculated.

3. The underwater health risk assessment method as described in claim 1, characterized in that, The determination of the channel quality index for each channel of the multispectral photoplethysmography (MPA) signal includes: The physiological characteristics of the multispectral photoplethysmography (PPG) signal of each channel are obtained, and the physiological signal quality index of the corresponding channel PPG signal is calculated based on the physiological characteristics. The underwater coupling characteristics of the multispectral photoplethysmography (PPG) signal for each channel are obtained, and the underwater optical coupling index of the corresponding channel PPG signal is calculated based on the underwater coupling characteristics. Based on the physiological signal quality index of each channel and the underwater optical coupling index, the channel quality index of the corresponding channel photoplethysmography signal is calculated.

4. The underwater health risk assessment method as described in claim 1, characterized in that, The physiological parameters include heart rate, blood oxygen saturation, and blood pressure trend; the physiological parameter prediction results include heart rate prediction results, blood oxygen saturation prediction results, and blood pressure trend prediction results. Based on the fused photoplethysmography (PPG) signal, the multi-channel multispectral PPG signal, and the motion data, the physiological parameters of the person to be tested are predicted, and the physiological parameter prediction results are obtained, including: Based on the motion data, motion artifact suppression is performed on the fused photoplethysmography (PPG) signal to obtain a suppressed PPG signal. A candidate peak set of the suppressed PPG signal is extracted, and the heart rate prediction result is determined based on the candidate peak set. Based on the multi-channel multispectral photoplethysmography pulse wave signal, the red light signal and near-infrared signal of each channel are extracted, and the blood oxygen saturation prediction result is determined based on the red light signal and near-infrared signal of each channel. Feature extraction is performed on the fused photoplethysmography (PPG) signal to obtain signal features. Blood pressure trend prediction is then performed based on these signal features to obtain the blood pressure trend prediction result.

5. The underwater health risk assessment method as described in claim 1, characterized in that, The step of extracting features from each frame of the eye-tracking image to obtain the pupil features of each frame includes: The eye-tracking images are segmented to determine the pupil region of the person to be tested; Based on the pupil region, the pupil center coordinates, pupil diameter, and pupil area are calculated, and the pupil center coordinates, pupil diameter, and pupil area are the pupil features.

6. The underwater health risk assessment method as described in claim 5, characterized in that, The step of calculating the eye movement feature vector based on the pupil features of each frame of eye movement image includes: Based on the pupil center coordinates, pupil diameter, and pupil area, the average pupil diameter, standard deviation of pupil diameter fluctuation, mean pupil diameter change rate, number of fixation frames, average fixation time, number of saccade frames, saccade frequency, saccade path length, and blink frequency are calculated. The average pupil diameter, standard deviation of pupil diameter fluctuation, mean pupil diameter change rate, number of fixation frames, average fixation time, number of saccade frames, saccade frequency, saccade path length, and blink frequency constitute an eye movement feature vector.

7. An underwater health risk assessment device, characterized in that, The underwater health risk assessment device includes: The acquisition module is used to acquire multi-channel multispectral photoplethysmography (PPG) signals, eye-tracking image sequences, and motion data of the person under test in an underwater environment at any time window. The suppression module is used to perform artifact suppression on the multi-channel multispectral photoplethysmography signal based on the motion data, so as to obtain the artifact-suppressed photoplethysmography signal. The fusion module is used to determine the channel quality index of the multispectral photoplethysmography (PPG) signal for each channel, and to perform weighted fusion of the artifact-suppressed PPG signals based on the channel quality index to obtain the fused PPG signal. The physiological parameter prediction module is used to predict the physiological parameters of the person to be tested based on the fused photoplethysmography signal, the multi-channel multispectral photoplethysmography signal and the motion data, and obtain the physiological parameter prediction result. The eye movement feature extraction module is used to extract features from each frame of the eye movement image to obtain the pupil features of each frame of the eye movement image. Based on the pupil features of each frame of the eye movement image, the eye movement feature vector is calculated. The comprehensive state index prediction module is used to predict the comprehensive state index of the person to be tested based on the physiological parameter prediction results within the time window and the eye movement feature vector, and obtain the comprehensive state index prediction result and confidence level. The assessment module is used to assess the health risk of the person to be tested based on the prediction results of the comprehensive status index and the confidence level in all time windows, and to obtain the health risk assessment result.

8. A diving mask, characterized in that, The diving mask includes a diving mask body, a multispectral photoplethysmography (PPG) sensor module, an eye-tracking module, a motion sensing module, and a processing module. The multispectral photoplethysmography (PPG) sensor module, the eye-tracking module, the motion sensing module, and the processing module are disposed on the diving mask body; The processing module is electrically connected to the multispectral photoplethysmography pulse wave sensing module, the eye-tracking module, and the motion sensing module, respectively. The multispectral photoplethysmography (PPG) sensing module is used to acquire multispectral PPG signals and transmit the multispectral PPG signals to the processing module. The eye-tracking module is used to acquire eye-tracking image sequences and transmit the eye-tracking image sequences to the processing module; The motion sensing module is used to collect motion data and transmit the motion data to the processing module. The processing module is used to execute the underwater health risk assessment method according to any one of claims 1 to 6.

9. The diving mask as described in claim 8, characterized in that, The diving mask also includes an alarm module, a supplementary lighting strip, a wireless communication module, and a power module; The alarm module, the supplementary lighting strip, the power module, and the wireless communication module are disposed on the diving mask body; The processing module is electrically connected to the alarm module, the fill light strip is electrically connected to the wireless communication module; The alarm module receives the drive signal generated by the processing module based on the health risk assessment results and outputs a visual alarm prompt. The supplementary light strip receives trigger commands sent by the processing module to adjust the color temperature and brightness; The wireless communication module transmits the health risk assessment results received from the processing module. The power module provides electrical energy to each module.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the underwater health risk assessment method as described in any one of claims 1 to 6.

Images