Physiological parameter detection method, device and wearable device

Abstract

This application relates to a method, apparatus, and wearable device for detecting physiological parameters. It can simultaneously acquire initial physiological signals and multiple inertial measurement signals, then fuse and analyze the multiple inertial measurement signals to obtain a synthetic inertial measurement signal. Next, adaptive filtering analysis is performed on the initial physiological signal and the synthetic inertial measurement signal to obtain a pure physiological signal, which is then used for extraction and analysis to determine physiological parameters. This scheme, by simultaneously acquiring initial physiological signals and multiple inertial measurement signals, and fusing the multiple inertial measurement signals during the data processing stage to obtain a synthetic inertial measurement signal with a higher signal-to-noise ratio, is then input into subsequent adaptive filtering analysis. By combining hardware denoising through multi-inertial measurement unit fusion with software denoising through adaptive filtering analysis, motion artifact noise generated by user movement is removed from the initial physiological signal, significantly improving the accuracy of physiological parameter acquisition.

Description

Technical Field

[0001] This application relates to the field of wearable device technology, and in particular to a method, apparatus and wearable device for detecting physiological parameters. Background Technology

[0002] PPG (Photoplethysmography), a non-invasive optical measurement technique, is widely used in wearable devices (such as smart bracelets and watches) to monitor physiological parameters such as heart rate and blood oxygen saturation. Its basic principle utilizes the characteristic that the absorption of specific wavelengths of light by skin tissue changes with blood volume and pulse rate. A photoelectric sensor detects the intensity of reflected or transmitted light to obtain the pulse wave signal.

[0003] However, physiological parameter detection is highly susceptible to interference from user movement, producing noise known as motion artifacts. In severe cases, these artifacts can completely drown out weak physiological signals, leading to measurement failure or a significant decrease in accuracy. Summary of the Invention

[0004] Therefore, it is necessary to provide a physiological parameter detection method, device, and wearable device to address the problem that physiological parameter detection is easily affected by user movement, leading to measurement failure or a significant decrease in accuracy.

[0005] In a first aspect, this application provides a method for detecting physiological parameters, comprising: simultaneously acquiring initial physiological signals and multiple inertial measurement signals; wherein the multiple inertial measurement signals are synchronously acquired by multiple independently configured inertial measurement units, each inertial measurement unit having the same zero-bias instability range and different microelectromechanical processes; performing fusion analysis on the multiple inertial measurement signals to determine a synthetic inertial measurement signal; performing adaptive filtering analysis based on the initial physiological signal and the synthetic inertial measurement signal to determine a pure physiological signal; and performing extraction analysis based on the pure physiological signal to determine physiological parameters.

[0006] Secondly, this application provides a physiological parameter detection device, comprising: a synchronous acquisition module for synchronously acquiring initial physiological signals and multiple inertial measurement signals; wherein the multiple inertial measurement signals are synchronously acquired by multiple independently configured inertial measurement units, each inertial measurement unit having the same zero-bias instability range and different microelectromechanical processes; a fusion module for performing fusion analysis on the multiple inertial measurement signals to determine a synthetic inertial measurement signal; a filtering module for performing adaptive filtering analysis based on the initial physiological signal and the synthetic inertial measurement signal to determine a pure physiological signal; and a parameter identification module for extracting and analyzing based on the pure physiological signal to determine physiological parameters.

[0007] Thirdly, this application provides a wearable device, including a physiological parameter acquisition unit, a processor, and at least two inertial measurement units (IMUs). Each IMU and the physiological parameter acquisition unit are respectively connected to the processor. The physiological parameter acquisition unit is used to acquire initial physiological signals. Each IMU acquires one IMU signal. The zero-bias instability range of each IMU is the same, and their microelectromechanical processes are different. The processor is used to determine physiological parameters based on the initial physiological signals and the multiple IMU signals.

[0008] The aforementioned physiological parameter detection method, device, and wearable device are equipped with multiple inertial measurement units (IMUs) for inertial measurement signal acquisition. They can simultaneously acquire initial physiological signals and multiple inertial measurement signals, then fuse and analyze these signals to obtain a synthetic inertial measurement signal. Next, adaptive filtering analysis is performed on the initial physiological signal and the synthetic inertial measurement signal to obtain a pure physiological signal. This pure physiological signal is then used for extraction and analysis to determine the physiological parameters. This scheme, by simultaneously acquiring initial physiological signals and multiple inertial measurement signals, and fusing the multiple inertial measurement signals during the data processing stage to obtain a synthetic inertial measurement signal with a higher signal-to-noise ratio, is then input into subsequent adaptive filtering analysis.

[0009] Thus, by combining hardware denoising through the fusion of multiple inertial measurement units with software denoising through adaptive filtering analysis, motion artifact noise generated by user movement is removed from the initial physiological signal, significantly improving the accuracy of physiological parameter acquisition and effectively solving the problem that physiological parameter detection is easily affected by user movement, leading to measurement failure or a significant decrease in accuracy. Attached Figure Description

[0010] To more clearly illustrate the technical solutions in the embodiments of this application or the conventional technology, the drawings used in the description of the embodiments or the conventional technology will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0011] Figure 1 This is a schematic diagram of a physiological parameter detection method in one embodiment of this application;

[0012] Figure 2 This is a schematic diagram of a physiological parameter detection method in another embodiment of this application;

[0013] Figure 3 This is a schematic diagram of a pure physiological signal analysis process in one embodiment of this application;

[0014] Figure 4This is a schematic diagram of the pure physiological signal analysis process in another embodiment of this application;

[0015] Figure 5 This is a schematic diagram of the physiological parameter detection device in one embodiment of this application;

[0016] Figure 6 This is a schematic diagram of the physiological parameter detection device in another embodiment of this application;

[0017] Figure 7 This is a schematic diagram of the wearable device structure in one embodiment of this application;

[0018] Figure 8 This is a schematic diagram of the arrangement of the inertial measurement unit in one embodiment of this application;

[0019] Figure 9 This is a schematic diagram of the packaging structure of a wearable device in one embodiment of this application. Detailed Implementation

[0020] To facilitate understanding of this application, a more complete description will be provided below with reference to the accompanying drawings. Preferred embodiments of this application are shown in the drawings. However, this application can be implemented in many different forms and is not limited to the embodiments described herein. Rather, these embodiments are provided to provide a thorough and complete understanding of the disclosure of this application.

[0021] The physiological parameter detection method provided in this application is applied to wearable devices. The specific type of wearable device is not unique. It can be a smartwatch, smart bracelet, smart glasses, etc. In other embodiments, it can also be a wearable monitoring patch, wearable armband, wearable chest strap, etc., which are not limited here.

[0022] The wearable device provided in this application includes a physiological parameter acquisition unit for monitoring physiological signals, and multiple inertial measurement units for monitoring inertial measurement signals (which can also be understood as motion parameters) generated by the user's movement. The physiological parameter acquisition unit and the inertial measurement units are both connected to a processor, which is used to execute the physiological parameter detection method proposed in this application.

[0023] It is understood that the type of physiological parameter acquisition device is not unique. In one embodiment, it includes a PPGAFE (Analog Front End) chip for monitoring physiological indicators such as heart rate and blood oxygen. The type of inertial measurement unit (IMU) is also not unique; as long as it includes a three-axis velocitiesight and a three-axis gyroscope, it is acceptable and is not limited here.

[0024] Please see Figure 1In one aspect, this application provides a method for detecting physiological parameters, including steps 102, 104, 106 and 108.

[0025] Step 102: Simultaneously acquire initial physiological signals and multiple inertial measurement signals.

[0026] Multiple inertial measurement signals are synchronously acquired by several independently configured inertial measurement units. The initial physiological signal refers to the raw, unprocessed physiological electrical signal directly acquired via photoplethysmography (PPG). Typically, the initial physiological signal comprises two parts: a useful physiological component reflecting changes in vascular volume caused by heartbeats, and motion artifact noise introduced by user body movements (such as arm swings and muscle tremors). In wearable devices, the initial physiological signal is usually obtained by illuminating the skin with a green / red LED (light emitting diode) in the physiological parameter acquisition unit, and then receiving changes in reflected light intensity through the photodiode of the physiological parameter acquisition unit.

[0027] An inertial measurement unit (IMU) is a microelectromechanical system (MEMS) sensor that integrates a triaxial accelerometer and a triaxial gyroscope to measure the linear acceleration and angular velocity of an object in three-dimensional space. Multiplexed inertial measurement signals refer to multiple sets of motion data synchronously acquired and output by two or more independent IMUs. Each IMU signal corresponds to the raw output of one IMU, typically including time-series data of triaxial acceleration and triaxial angular velocity. The purpose of multiplexed IMUs is to acquire motion information from different locations in space, providing redundancy and complementarity for subsequent fusion to improve motion estimation accuracy.

[0028] In one embodiment, the zero-bias instability range of each inertial measurement unit is the same, and the microelectromechanical processes are different for each unit.

[0029] Zero-bias instability refers to the degree to which the output signal of an inertial measurement unit (IMU) fluctuates around its mean value under constant conditions (such as constant temperature, no external acceleration or angular velocity input). It is a core indicator for measuring the long-term stability and measurement accuracy of an IMU. This indicator is usually expressed as the rate of change of value per unit time. The smaller the value, the more stable the output of the IMU is in a stationary state. By configuring the zero-bias instability range of each IMU to be the same, it is possible to ensure a high degree of consistency in the accuracy of each IMU.

[0030] The term "different MEMS processes" refers to the fact that multiple selected inertial measurement units (IMUs) employ different micromechanical structures, manufacturing processes, or operating principles for their internal MEMS (Micro-Electro-Mechanical Systems) sensing chips. For example, gyroscopes may use different vibration modes (tuning fork type, vibrating wheel type) and different comb-tooth drive / detection structures; accelerometers may use different mass-spring structures or capacitive sensing methods. The core objective is to ensure that the inherent noise floor of each IMU and its response characteristics to specific environmental disturbances (such as temperature gradients or vibrations at specific frequencies) are statistically uncorrelated or weakly correlated.

[0031] Thus, by employing a heterogeneous selection strategy using inertial measurement units (IMUs) with consistent zero-bias instability ranges but different microelectromechanical processes, it can be ensured that while each IMU has the same measurement range and sensitivity, its internal noise (such as thermomechanical noise and electronic noise) is statistically uncorrelated or weakly correlated. During data fusion, uncorrelated noise components are suppressed through averaging, thereby reducing noise at the physical level and improving the accuracy of motion noise.

[0032] In one embodiment, a PPG AFE chip is used as a physiological parameter acquisition device. The PPG AFE chip, also known as a photoplethysmography (PPG) analog front-end chip, is a dedicated mixed-signal integrated circuit integrated into a wearable device. It is responsible for driving the optical emission components and receiving and initially processing the returned photoelectric signals, converting them into digital physiological data that can be read by the processor.

[0033] Step 104: Perform fusion analysis on the multi-channel inertial measurement signals to determine the composite inertial measurement signal.

[0034] Fusion analysis refers to the algorithmic process of integrating and processing multiple inertial measurement signals to generate a synthetic motion signal with a higher signal-to-noise ratio and greater stability, i.e., a synthetic inertial measurement signal. The synthetic inertial measurement signal is a high-quality reference signal representing the user's overall motion state, obtained through fusion analysis of multiple inertial measurement signals. As the reference input for subsequent adaptive filters, the quality of the synthetic inertial measurement signal directly affects the effectiveness of motion artifact elimination. Ideally, it should retain, to the greatest extent possible, components related to motion artifacts in the initial physiological signal while suppressing the noise of the inertial measurement unit itself.

[0035] It is understood that there is no single way to perform fusion analysis on multiple inertial measurement signals, as long as the final synthesized inertial measurement signal has a higher signal-to-noise ratio. This can be one or more methods such as weighted average fusion analysis, Kalman filter fusion analysis, principal component fusion analysis, and median fusion analysis, and is not limited here. For example, in one embodiment, the fusion analysis includes steps such as time synchronization alignment, weighted averaging, and principal component analysis to suppress independent random noise from each inertial measurement unit and retain common real motion components, ultimately outputting a more representative synthesized inertial measurement signal. In another embodiment, only weighted average fusion analysis can be used to process the multiple inertial measurement signals. This method has low computational complexity and can effectively improve fusion efficiency.

[0036] Step 106: Perform adaptive filtering analysis based on the initial physiological signal and the synthetic inertial measurement signal to determine the pure physiological signal.

[0037] Adaptive filtering analysis is a signal processing technique that automatically adjusts filter parameters based on the characteristics of the input signal, used to separate the target component from a mixed signal. A pure physiological signal refers to a physiological signal after adaptive filtering, in which motion artifacts are significantly suppressed or removed.

[0038] In more detail, the technical solution of this application adopts an adaptive noise cancellation architecture, using the synthetic inertial measurement signal as the reference input and the initial physiological signal as the main input, thereby estimating and eliminating (or suppressing) motion interference components to obtain a pure physiological signal.

[0039] Step 108: Extract and analyze the pure physiological signals to determine the physiological parameters.

[0040] Extraction analysis refers to the algorithmic process of feature identification and parameter calculation from pure physiological signals. Physiological parameters refer to quantitative indicators that reflect the physiological state of the human body, ultimately calculated from pure physiological signals, such as heart rate, blood oxygen saturation, and respiratory rate.

[0041] It is understood that the extraction and analysis methods are not unique. For example, in one embodiment, methods such as peak detection, spectrum analysis, and waveform morphology analysis may be included. These methods are mainly used to calculate specific and readable physiological indicators, such as heart rate, heart rate variability, and blood oxygen saturation, from pure physiological signals.

[0042] The aforementioned physiological parameter detection method is equipped with multiple inertial measurement units (IMUs) for inertial measurement signal acquisition. It can simultaneously acquire initial physiological signals and multiple inertial measurement signals, then fuse and analyze these signals to obtain a synthetic inertial measurement signal. Next, adaptive filtering analysis is performed on the initial physiological signal and the synthetic inertial measurement signal to obtain a purified physiological signal. This purified physiological signal is then used for extraction and analysis to determine the physiological parameters. This scheme, by simultaneously acquiring initial physiological signals and multiple inertial measurement signals, and fusing the multiple inertial measurement signals during the data processing stage to obtain a synthetic inertial measurement signal with a higher signal-to-noise ratio, is then input into subsequent adaptive filtering analysis.

[0043] Thus, by combining hardware denoising through the fusion of multiple inertial measurement units with software denoising through adaptive filtering analysis, motion artifact noise generated by user movement is removed from the initial physiological signal, significantly improving the accuracy of physiological parameter acquisition and effectively solving the problem that physiological parameter detection is easily affected by user movement, leading to measurement failure or a significant decrease in accuracy.

[0044] Please see Figure 2 In one embodiment, step 106 includes steps 202 and 204.

[0045] Step 202: Perform bandpass filtering on the initial physiological signal to determine the preprocessed physiological signal.

[0046] Step 204: Perform adaptive filtering analysis based on the preprocessed physiological signal and the synthetic inertial measurement signal to determine the pure physiological signal.

[0047] Bandpass filtering means allowing only components within a specific frequency range (passband) of the initial physiological signal to pass through, while significantly attenuating low-frequency and high-frequency components outside the passband. Preprocessed physiological signal refers to the intermediate signal obtained after the initial physiological signal has undergone bandpass filtering.

[0048] In this embodiment, the initial physiological signal typically includes low-frequency and high-frequency interference components. Low-frequency interference components are usually baseline drift caused by respiration, changes in body temperature, and slow changes in ambient light, while high-frequency interference components are usually electromyographic noise, power supply frequency interference, and high-frequency noise introduced by rapid fluctuations in ambient light. Therefore, to improve the accuracy of the physiological signal, the initial physiological signal needs to be bandpass filtered to remove both low-frequency and high-frequency interference components.

[0049] It is understood that the passband range of bandpass filtering is not unique. Based on the characteristics of the initial physiological signal, in one embodiment, the passband range can be set to 0.5 Hz (Hertz) to 5 Hz. In another embodiment, it can be set to other sizes, such as 0.6 Hz to 6 Hz, etc., without any specific limitation.

[0050] The above scheme preprocesses the initial physiological signal with bandpass filtering before adaptive filtering, effectively filtering out low-frequency baseline drift caused by respiration, body temperature changes, and high-frequency environmental noise, resulting in a cleaner signal input to the adaptive filter. This improves the targeting of subsequent motion noise estimation, reduces the impact of non-motion interference on the filtering process, and further enhances the success rate and accuracy of physiological parameter detection under complex motion conditions.

[0051] Please see Figure 3 In one embodiment, step 204 includes steps 302 and 304.

[0052] Step 302: Perform drift removal and amplitude normalization processing on the synthetic inertial measurement signal to determine the reference inertial measurement signal.

[0053] Step 304: Perform adaptive filtering analysis based on the preprocessed physiological signal and the reference inertial measurement signal to determine the pure physiological signal.

[0054] Drift correction is a signal correction technique for data acquired by inertial measurement units (IMUs). It aims to eliminate or significantly reduce slow-changing, non-motion-induced error components in the signal, primarily zero bias and temperature drift. Zero bias refers to the fixed or slowly changing bias of the IMU's output when stationary and without external force; this bias varies with time, temperature, and stress. In wearable devices, the zero bias of the IMU can drift due to device temperature changes and stress accumulation from long-term wear, causing acceleration or angular velocity readings to be non-zero even when stationary, thus contaminating the motion reference signal.

[0055] Amplitude normalization refers to the process of mapping the amplitude range of a synthetic inertial measurement signal to a preset standard interval (such as [0, 1] or [-1, 1]) through mathematical transformation, so that its dynamic range matches the amplitude range of the preprocessed physiological signal. This operation takes into account the different physical dimensions of the preprocessed physiological signal and the synthetic inertial measurement signal, resulting in significant differences in their original amplitude ranges. Normalization is necessary to eliminate these dimensional differences, creating conditions for stable and efficient operation of adaptive filtering analysis.

[0056] The reference inertial measurement signal refers to the synthetic inertial measurement signal that has undergone drift removal processing, followed by amplitude normalization and other processing, ultimately generating a signal used as a reference input to the adaptive filter. Normalization ensures its dynamic range is on the same order of magnitude as the preprocessed physiological signal, facilitating efficient updating and convergence of the filter coefficients.

[0057] The above scheme performs de-drift processing on the fused synthetic inertial measurement signal, eliminating the slow signal drift caused by zero-bias instability and temperature changes. The processed reference inertial measurement signal more accurately reflects the real, real-time inertial parameter information, making it highly correlated with the motion artifacts embedded in the initial physiological signal in both the time and frequency domains. This provides a higher-quality reference input for the adaptive filter, significantly accelerating the algorithm's convergence speed and reducing steady-state error, ensuring that the system can still quickly track and effectively compensate for artifacts even when the motion state changes rapidly.

[0058] Please see Figure 4 In one embodiment, step 304 includes steps 402 and 404.

[0059] Step 402: Calculate the noise based on the reference inertial measurement signal and the filter weighting coefficients to determine the motion noise signal.

[0060] Step 404: Perform artifact elimination calculations based on the preprocessed physiological signal and motion noise signal to determine the pure physiological signal.

[0061] In an adaptive filter, the filter weight coefficients are a set of adjustable parameters used to multiply the reference input signal with samples taken at different times. Their initial values ​​can be set to zero or a small random value, and are then continuously and dynamically updated during algorithm execution based on feedback from the error signal to minimize the residual motion artifacts. They represent a mathematical model of the dynamic relationship between the current motion and the physiological signal.

[0062] Noise calculation specifically refers to the process of convolving a reference inertial measurement signal with the current filter weights to estimate the motion artifact components contained in the initial physiological signal at the current moment. The motion noise signal refers to the specific estimate of the motion interference components mixed in the preprocessed physiological signal at the current moment, obtained through noise calculation. Artifact removal calculation refers to the operation of directly subtracting the estimated motion noise signal from the noisy preprocessed physiological signal to obtain an output signal (i.e., a clean physiological signal) with the estimated noise component removed.

[0063] In one embodiment, noise calculation can be expressed as y(n) = w(n) * Ref_IMU(n), where y(n) represents the motion noise signal, Ref_IMU(n) represents the reference inertial measurement signal, and w(n) represents the filter weight coefficient. Artifact removal calculation can be expressed as e(n) = d(n) - y(n), where e(n) represents the pure physiological signal, and d(n) represents the preprocessed physiological signal.

[0064] The above scheme employs an adaptive noise cancellation structure based on an Nth-order finite impulse response filter. Utilizing a high signal-to-noise ratio reference inertial measurement signal, it estimates the motion noise component contained in the initial physiological signal and then calculates and removes it from the noisy preprocessed physiological signal. This achieves active and precise elimination of motion artifacts, rather than simple signal filtering, thus maximizing the preservation of useful physiological information components in the initial physiological signal. This results in a pure physiological signal with extremely low distortion, laying a solid foundation for subsequent accurate calculations of physiological parameters such as heart rate and blood oxygen saturation.

[0065] In one embodiment, the method further includes: using a pure physiological signal as an error signal feedback to update the filter weight coefficients with the goal of minimizing the mean square value of the error signal.

[0066] Update refers to the process of recalculating and adjusting the filter weight coefficients according to the algorithm rules, using the latest calculated error signal (i.e., the pure physiological signal) and other inputs. Update reflects the self-learning capability of adaptive filtering. Its purpose is to minimize the mean square value of the error signal. Each update moves the filter coefficients toward a better solution, enabling the system to automatically track and adapt to changes in motion patterns, slow drifts in sensor characteristics, or changes in wearing conditions.

[0067] It should be noted that there is no single way to update the filter weight coefficients. In one embodiment, the filter weight coefficients can be dynamically adjusted using the LMS (Least Mean Square) or RLS (Recursive Least Squares) algorithm.

[0068] The above scheme updates the filter weight coefficients in real time by filtering the output error (i.e., the pure physiological signal), enabling the adaptive filter to continuously learn and adjust. This dynamic adjustment mechanism allows the system to automatically adapt to different users' movement patterns, changes in device tightness, and sensor performance drift over time, thereby maintaining excellent motion compensation performance throughout the device's entire lifespan and achieving a stable monitoring experience that requires no long-term calibration.

[0069] In one embodiment, the method further includes: performing zero-bias calibration based on the inertial measurement signal of the inertial measurement unit when the device is stationary.

[0070] The device's static state refers to the state in which the device used in the physiological parameter detection method is stationary. More specifically, in one embodiment, taking a wearable device as an example, the device's static state refers to a stable physical state where the wearable device is worn on the user's body, but the user's limbs do not undergo active movement or only exhibit barely perceptible physiological activity (such as breathing or slight tremors). In the static state, the device's casing, internal printed circuit board, and sensors are all macroscopically stationary. Ideally, the inertial measurement unit should only output a fixed signal consisting of gravitational acceleration (for the accelerometer) and its own zero-bias error, without including any dynamic acceleration or angular velocity components generated by the user's active movement.

[0071] Zero bias calibration refers to the signal correction process for inertial measurement units. It involves measuring and calculating the average output value of the unit under known conditions without external dynamic input, using this average value as an estimate of the system error (zero bias), and subtracting it in real time during subsequent dynamic measurements.

[0072] Zero bias is an inherent systematic error of inertial measurement units (IMUs), which causes measured values ​​to continuously deviate from the true value. The purpose of calibration is to obtain an accurate zero bias value under the current environment to eliminate its contamination of the motion reference signal. The zero bias calibration in this solution specifically refers to a dynamic, autonomous, and online calibration process, distinct from the one-time calibration at the factory, and capable of adapting to drift caused by wearable devices during use.

[0073] In a more detailed embodiment, raw signals from all inertial measurement units (IMUs) can be continuously monitored in the background. When a stationary condition is met (e.g., measurement data meets certain conditions and remains unchanged), the device is considered to be in a stationary state. Afterward, through a synchronization mechanism, all IMUs are controlled to synchronously acquire raw data for a fixed duration (e.g., 2-4 seconds) while stationary. For each measurement axis (3-axis acceleration + 3-axis angular velocity, a total of 6 channels) of each IMU, the following calculations are performed independently. Finally, the latest zero-bias estimate is updated and stored in the processor, overwriting the old calibration parameters, thus achieving the update.

[0074] The above solution introduces a dynamic self-calibration mechanism based on static states, periodically updating the zero bias of each inertial measurement unit. This resolves the measurement reference drift caused by long-term use, temperature cycling, or mechanical stress. It ensures the long-term accuracy of inertial measurement signals, improving reliability throughout the entire lifecycle from the source, eliminating the need for any manual calibration by the user.

[0075] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages in other steps. It is understood that the steps in different embodiments can be freely combined as needed, and all non-contradictory solutions formed by such combinations are within the scope of protection of this application.

[0076] Based on the same inventive concept, this application also provides a physiological parameter detection device for implementing the physiological parameter detection method described above. The solution provided by this device is similar to the implementation scheme described in the above method; therefore, the specific limitations in one or more physiological parameter detection device embodiments provided below can be found in the limitations of the physiological parameter detection method described above, and will not be repeated here.

[0077] Please see Figure 5 Secondly, this application provides a physiological parameter detection device, including a synchronous acquisition module 502, a fusion module 504, a filtering module 506, and a parameter recognition module 508.

[0078] The synchronous acquisition module 502 is used to synchronously acquire initial physiological signals and multiple inertial measurement signals; the fusion module 504 is used to perform fusion analysis on the multiple inertial measurement signals to determine the synthetic inertial measurement signal; the filtering module 506 is used to perform adaptive filtering analysis based on the initial physiological signals and the synthetic inertial measurement signals to determine the pure physiological signals; and the parameter identification module 508 is used to extract and analyze the pure physiological signals to determine the physiological parameters.

[0079] In one embodiment, the filtering module 506 is further configured to perform bandpass filtering on the initial physiological signal to determine the preprocessed physiological signal; and to perform adaptive filtering analysis based on the preprocessed physiological signal and the synthetic inertial measurement signal to determine the pure physiological signal.

[0080] In one embodiment, the filtering module 506 is further configured to perform drift removal and amplitude normalization processing on the synthetic inertial measurement signal to determine the reference inertial measurement signal; and to perform adaptive filtering analysis on the preprocessed physiological signal and the reference inertial measurement signal to determine the pure physiological signal.

[0081] In one embodiment, the filtering module 506 is further configured to perform noise calculation based on the reference inertial measurement signal and the filtering weight coefficient to determine the motion noise signal; and to perform artifact elimination calculation based on the preprocessed physiological signal and the motion noise signal to determine the pure physiological signal.

[0082] In one embodiment, the filtering module 506 is further configured to use the pure physiological signal as an error signal as feedback, and update the filtering weight coefficients with the goal of minimizing the mean square value of the error signal.

[0083] Please see Figure 6 In one embodiment, the device further includes a static calibration module 602.

[0084] The static calibration module 602 is used to perform zero-bias calibration based on the inertial measurement signal of the inertial measurement unit when the equipment is stationary.

[0085] Each module in the aforementioned physiological parameter detection device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in the processor of a computer device in hardware form or independent of it, or stored in the memory of a computer device in software form, so that the processor can call and execute the operations corresponding to each module.

[0086] Please see Figure 7 Thirdly, this application provides a wearable device, including a physiological parameter acquisition unit 73, a processor 72, and at least two inertial measurement units 71. Each inertial measurement unit 71 and the physiological parameter acquisition unit 73 are respectively connected to the processor 72. The physiological parameter acquisition unit 73 is used to acquire initial physiological signals, and one inertial measurement unit 71 acquires one inertial measurement signal. The zero-bias instability range of each inertial measurement unit is the same, and the microelectromechanical processes are different from each other. The processor is used to determine physiological parameters based on the initial physiological signals and multiple inertial measurement signals.

[0087] The implementation of the physiological parameter detection method is as shown in the above embodiments and accompanying drawings. The physiological parameter acquisition device 73 refers to a hardware module in a wearable device specifically responsible for driving the light source and receiving and converting physiological optical signals. For example, this application uses the PPG AFE chip as an example for explanation. The inertial measurement unit 71 is a microelectromechanical system sensor integrating a triaxial accelerometer and a triaxial gyroscope, used to measure the linear acceleration and angular velocity of an object in three-dimensional space.

[0088] In this embodiment, the zero-bias instability range of each inertial measurement unit 71 is the same, ensuring that the long-term stability of the measurement reference is comparable; at the same time, the microelectromechanical processes are different, ensuring that the background noise of each inertial measurement unit 71 is statistically uncorrelated, providing a physical basis for subsequent fusion and noise reduction.

[0089] The processor 72 refers to the microcontroller or digital signal processor in the wearable device, which is responsible for receiving data from the physiological parameter acquisition unit 73 and each inertial measurement unit 71, and executing motion compensation algorithms to calculate physiological parameters.

[0090] This wearable device is equipped with multiple inertial measurement units 71 to acquire inertial measurement signals. It can simultaneously acquire initial physiological signals and multiple inertial measurement signals, and then determine physiological parameters using the initial physiological signals and multiple inertial measurement signals. In this way, by combining hardware denoising through the fusion of multiple inertial measurement units 71 with software denoising within the processor, motion artifact noise generated by user movement is removed from the initial physiological signals, significantly improving the accuracy of physiological parameter acquisition. This effectively solves the problem that physiological parameter detection is easily affected by user movement, leading to measurement failure or a significant decrease in accuracy.

[0091] In some embodiments, the processor 72 is further configured to: perform fusion analysis on multiple inertial measurement signals to determine a synthetic inertial measurement signal; perform adaptive filtering analysis based on the initial physiological signal and the synthetic inertial measurement signal to determine a pure physiological signal; and perform extraction analysis based on the pure physiological signal to determine physiological parameters.

[0092] In this embodiment, the processor 72 can receive multiple synchronously acquired inertial measurement signals, and fuse the multiple inertial measurement signals using methods such as weighted averaging or principal component analysis to generate a synthetic inertial measurement signal with a higher signal-to-noise ratio. Then, adaptive filtering analysis is performed on the initial physiological signal and the synthetic inertial measurement signal to determine the pure physiological signal, which is then used to acquire physiological parameters.

[0093] Please refer to the following: Figure 8 In one embodiment, the distance between any two adjacent inertial measurement units 71 (i.e., H in the figure) is greater than a preset distance, wherein the preset distance indicates that the resonant frequencies between adjacent inertial measurement units 71 are not interconnected.

[0094] The preset spacing refers to the minimum physical distance threshold between the center points of two adjacent inertial measurement units 71, set in the hardware layout design of wearable devices to ensure the normal and independent operation of multiple inertial measurement units 71. The resonant frequencies being mutually disconnected describes a physical state where the mechanical vibration energy of the high-frequency resonant mass blocks within two or more inertial measurement units 71 (specifically, their internal gyroscopes) cannot be effectively transferred and coupled to each other through media such as circuit boards, packaging materials, or air due to a sufficiently large installation spacing.

[0095] The preset spacing is not arbitrarily set; it can be determined through simulation or experiment based on mechanical wave propagation theory and the physical characteristics of MEMS devices (the inertial measurement unit 71 is a type of MEMS device). Its core purpose is to physically isolate each inertial measurement unit 71 to avoid harmful mutual interference caused by them being too close together.

[0096] It is understood that the size of the preset spacing is not unique. For example, in one embodiment, the preset spacing can be configured to be greater than or equal to 20 mm.

[0097] In a more detailed embodiment, the preset spacing can be configured as such as 20 mm, 25 mm, 30 mm, etc., without any specific limitation.

[0098] The above scheme, by scientifically setting the installation spacing between the inertial measurement units 71 and based on the mechanical wave propagation theory, effectively avoids the mutual mechanical coupling and resonance interference caused by the close proximity of the internal high-frequency resonant mass blocks of multiple inertial measurement units 71. This physical isolation measure ensures that the inertial measurement signals acquired by each inertial measurement unit 71 are independent and authentic, providing a high-quality independent signal source for subsequent data fusion, and is a key hardware guarantee for obtaining high signal-to-noise ratio synthetic inertial measurement signals.

[0099] Please see Figure 9 In one embodiment, the wearable device further includes a circuit board 91 and a reinforcing member 92. An inertial measurement unit 71 is disposed on a first surface of the circuit board 91, and the reinforcing member 92 is disposed on a second surface of the circuit board 91. The second surface is disposed opposite to the first surface. The first vertical projection of the reinforcing member 92 on the first surface at least partially overlaps with the second vertical projection of the inertial measurement unit 71 on the first surface.

[0100] The circuit board 91 refers to the carrier in wearable devices that supports and electrically connects various electronic components (such as chips, sensors, resistors, and capacitors), typically a flexible circuit board or a rigid printed circuit board. The reinforcing member 92 refers to a structural component attached to a specific area of ​​the circuit board 91, designed to locally increase the mechanical stiffness and strength of that area. The reinforcing member 92 is usually made of a material with much higher rigidity than the circuit board 91 itself, such as stainless steel sheets, aluminum sheets, thickened FR4 plates, or special engineering plastics. Its core function is localized reinforcement, forming a robust miniature platform beneath the mounting area of ​​the inertial measurement unit 71 to resist and disperse external mechanical stress, preventing unintended microscopic deformation in that area.

[0101] The outline of the reinforcing member 92 is vertically projected onto the first surface (mounting surface) of the circuit board 91 to form a first vertical projection; the outline of the inertial measurement unit 71 body is vertically projected onto the same surface to form a second vertical projection. The first vertical projection and the second vertical projection partially overlap, meaning that at least a portion of the two projection areas are shared in the plane. This geometric requirement ensures that the reinforcing effect of the reinforcing member 92 can be directly and effectively transmitted and act on the precise area where the inertial measurement unit 71 is located.

[0102] It is understood that at least partial overlap is the minimum requirement to ensure the effectiveness of reinforcement. In some embodiments, the first vertical projection of the reinforcement 92 on the first surface partially overlaps with the second vertical projection of the inertial measurement unit 71 on the first surface. In this way, the overlapping area is reinforced, and even if the overlapping area may still experience micro-deformation, the reinforcement 92 can still significantly reduce the degree of deformation transmission to the inertial measurement unit 71 by increasing the average stiffness of the area as a whole.

[0103] In some embodiments, the first projection of the reinforcing member 92 should completely cover or the second projection of the inertial measurement unit 71. That is, the first projection and the second projection can completely overlap, or the projection area formed by the second projection is within the projection area formed by the first projection. In this way, the reinforcing member 92 directly supports the entire packaging area of ​​the inertial measurement unit 71, resulting in the most uniform stress distribution, the strongest deformation suppression effect on the circuit board 91, and the best protection effect.

[0104] It should be noted that in one embodiment, the number of reinforcing members 92 can be one. In another embodiment, multiple reinforcing members 92 can be configured, such as one reinforcing member 92 corresponding to one inertial measurement unit 71. There is no limitation here, and the configuration can be made according to actual needs.

[0105] The above-described solution adds a rigid reinforcement to the back of the circuit board 91 on which the inertial measurement unit (IMU) is mounted, effectively enhancing the mechanical rigidity of this local area and suppressing minor bending or deformation of the circuit board 91 caused by the user's daily wrist movements, external impacts, or long-term stress. This structure protects the IMU, which is extremely sensitive to its adhesion stress, preventing the introduction of additional false acceleration or angular velocity signals (i.e., zero-bias changes) due to deformation of the circuit board 91, thus ensuring the long-term stability and reliability of the IMU's measurement data.

[0106] In one embodiment, the physiological parameter acquisition unit 73 is connected to the external interrupt pin of each inertial measurement unit 71 via a synchronization signal acquisition line. The physiological parameter acquisition unit 73 is used to synchronously trigger the inertial measurement unit 71 to acquire inertial measurement signals during the acquisition time window.

[0107] The synchronization signal acquisition line refers to a dedicated physical wire that extends from the physiological parameter acquisition unit 73 and connects to a specific pin of each inertial measurement unit 71. The acquisition time window refers to the specific time interval during which the physiological parameter acquisition unit 73 emits LED light, receives data from photodiodes, and completes analog-to-digital conversion in order to acquire a valid pulse wave data point.

[0108] The acquisition time window is very short (typically a few hundred microseconds) and is repeatedly opened at a fixed frequency (e.g., 100Hz). Through synchronous control, it can be ensured that the sampling point of the inertial measurement unit 71 falls precisely within this optical measurement window, so that the acquired motion data (inertial measurement signal) corresponds completely with the initial physiological signal in time, eliminating the correlation loss caused by the misalignment of the sampling time.

[0109] In this embodiment, the synchronous signal acquisition line can broadcast a synchronization pulse immediately when the physiological parameter acquisition unit 73 begins a valid optical sampling, forcing all inertial measurement units 71 to latch their sensor data at the same electrical moment, thus eliminating the millisecond-level random delay caused by traditional software polling or interrupt response.

[0110] The above scheme employs a hardware synchronization triggering mechanism with the physiological parameter acquisition unit 73 as the main device. The physiological parameter acquisition unit 73 emits a synchronization pulse the instant it acquires the photoelectric signal within the acquisition time window, directly triggering all inertial measurement units 71 to simultaneously latch data. This hardware-level synchronization method completely eliminates the time uncertainty caused by traditional software polling or interrupt responses, achieving strict temporal synchronization between the initial physiological signal and multiple inertial measurement signals. This ensures perfect temporal alignment between the artifacts in the inertial measurement signal and the initial physiological signal, providing a crucial timing foundation for the effective operation of adaptive filtering.

[0111] Please continue reading. Figure 9 In one embodiment, the wearable device also includes a housing 93, and the physiological parameter acquisition unit 73, the inertial measurement unit 71 and the processor 72 are all disposed inside the housing 93. An air gap 94 is provided between the inertial measurement unit 71 and the housing 93.

[0112] The housing 93 refers to the structural component that forms the external outline and physical body of a wearable device (such as a smart bracelet or smartwatch), housing and protecting the internal electronic components and providing the user's wearing interface. The housing 93 is typically made of plastic, metal, or composite materials and is the part of the wearable device that comes into direct contact with the external environment and the user's body. During daily use, the housing 93 undergoes macroscopic or microscopic elastic or even plastic deformation due to wearing pressure, wrist flexion, external impacts, or temperature changes. This deformation is the primary pathway for transmitting external mechanical stress.

[0113] Air gap 94 refers to a physical space, not filled with any solid material, intentionally reserved between the top surface (or upper surface of the package) of the inertial measurement unit 71 and the inner wall of the housing 93 in the internal assembly structure of a wearable device. Air gap 94 is filled with air, providing mechanical decoupling and buffering. When the housing 93 deforms due to external forces, air gap 94 acts as a soft insulating layer, effectively absorbing and dissipating deformation energy, preventing direct hard contact or continuous compressive stress between the inner wall of the housing 93 and the inertial measurement unit 71, thereby protecting the sensor from external mechanical interference.

[0114] The aforementioned design incorporates a sufficient air gap 94 between the top of the inertial measurement unit 71 and the inner wall of the wearable device's housing 93, forming a mechanically isolated structure with the top suspended. This design ensures that when the housing 93 is subjected to external pressure or impact, the impact force is not directly transmitted to the fragile inertial measurement unit 71, but is instead buffered and dissipated through the air layer. This significantly enhances the inertial measurement unit 71's ability to withstand everyday accidental collisions and assembly stresses, protects its delicate mechanical structure, extends the sensor's lifespan, and ensures its detection accuracy.

[0115] To facilitate understanding of the technical solution of this application, the following detailed embodiments will be used to explain and illustrate this application.

[0116] Hardware architecture design:

[0117] (1) Overall layout: At least two inertial measurement units 71 (two units, namely IMU1 and IMU2) are set on the circuit board 91 of the wearable device. The straight-line distance between the two inertial measurement units 71 in physical space is greater than a preset distance (such as 20 mm, which is set based on the acoustic and mechanical wave propagation characteristics to ensure that the resonant frequencies of the MEMS gyroscopes of the two IMUs are not connected).

[0118] (2) IMU selection: The accuracy is consistent, and the zero-bias instability of IMU1 and IMU2 is consistent; the process difference: IMU1 and IMU2 adopt different MEMS microstructure designs (e.g., different comb structures, different resonant frequency designs). Due to the different processes, the response characteristics (fusion) of the two to thermal response and vibration at a specific frequency are not related to noise.

[0119] Packaging structure design:

[0120] (1) Back reinforcement: A rigid reinforcement (such as a stainless steel sheet or a thickened FR4 board) is mounted on the back (bottom layer) of the circuit board 91 where the IMU is located. The reinforcement 92 completely covers the IMU projection area.

[0121] (2) Top suspension: The IMU is a MEMS device that is fragile and sensitive to stress. Sufficient air gap 94 is left between the top surface of the main body and the inner wall of the wearable device housing 93 to ensure that the IMU will not come into contact with the mechanical impact deformation of the wristband.

[0122] Signal synchronization:

[0123] Employing AFE source triggering mode, the PPG AFE chip (physiological parameter acquisition unit 73) is configured as the master device, with a synchronization signal line leading out and connected to the processor 72 (such as an MCU) and the external interrupt pins of all IMUs. When the PPG AFE chip acquires physiological parameters within the acquisition time window, the synchronization signal line outputs pulses synchronously, forcing all IMUs to latch data at the same microsecond.

[0124] Data processing:

[0125] (1) Dynamic self-calibration: update zero bias in a static state.

[0126] (2) IMU fusion: Weighted fusion of data from multiple IMUs to output a synthetic inertial measurement signal with high signal-to-noise ratio.

[0127] (3) Bandpass filtering (e.g., 0.5Hz-5Hz) is applied to the initial physiological signal to remove baseline drift and high-frequency noise, resulting in a preprocessed physiological signal, denoted as d(n).

[0128] (4) Normalize and de-drift the synthetic inertial measurement signal so that its amplitude range matches the dynamic range of the preprocessed physiological signal to obtain the reference inertial measurement signal Ref_IMU(n).

[0129] (5) Using the noisy preprocessed physiological signal d(n) as the main input and the reference inertial measurement signal Ref_IMU(n) as the reference input, an Nth-order finite impulse response (FIR) filter is constructed. The filter weight coefficient w(n) is dynamically adjusted using the LMS (least mean square) or RLS (recursive least square) algorithm. The pure physiological signal e(n) is obtained by calculating y(n)=w(n)*Ref_IMU(n) and e(n)=d(n)-y(n).

[0130] (6) Perform spectrum analysis (FFT) or peak detection on the pure physiological signal e(n) to calculate accurate physiological parameters (such as heart rate / blood oxygen value).

[0131] The above solution can achieve the following effects:

[0132] Faster and more accurate algorithm convergence: Due to the improved signal-to-noise ratio (1 / √N) of the IMU signal used as the reference input, the LMS / RLS algorithm can lock the motion frequency faster and has excellent tracking performance in non-steady-state motions such as variable speed running.

[0133] Ultimate signal-to-noise ratio: a combination of hardware denoising (multi-IMU fusion) and software denoising (adaptive filtering).

[0134] Physical anti-interference: The spacing between the inertial measurement units 71 is greater than the preset distance, which eliminates the risk of multi-gyroscope resonance; the back reinforcement and top suspension structure effectively isolate mechanical impact and protect the precision MEMS structure.

[0135] Leading throughout the entire lifecycle: By dynamically adjusting the device when it is inactive, the problem of accuracy degradation after long-term use of wearable devices is solved without the need for manual intervention by the user.

[0136] Manual zero-phase: Hardware-level synchronization eliminates software scheduling and ensures a strict correspondence between motion data and photoelectric data.

[0137] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0138] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this patent application should be determined by the appended claims.

Claims

1. A method for detecting physiological parameters, characterized in that, include: Initial physiological signals and multiple inertial measurement signals are acquired simultaneously; wherein, the multiple inertial measurement signals are acquired synchronously by multiple independently set inertial measurement units, and the zero-bias instability range of each inertial measurement unit is the same, and the microelectromechanical processes are different from each other; The multiple inertial measurement signals are fused and analyzed to determine the composite inertial measurement signal; Adaptive filtering analysis is performed based on the initial physiological signal and the synthetic inertial measurement signal to determine the pure physiological signal; Physiological parameters are determined by extracting and analyzing the purified physiological signals.

2. The method according to claim 1, characterized in that, The step of performing adaptive filtering analysis based on the initial physiological signal and the synthetic inertial measurement signal to determine the pure physiological signal includes: The initial physiological signal is subjected to bandpass filtering to determine the preprocessed physiological signal; Adaptive filtering analysis is performed based on the preprocessed physiological signal and the synthetic inertial measurement signal to determine the pure physiological signal.

3. The method according to claim 2, characterized in that, The step of performing adaptive filtering analysis based on the preprocessed physiological signal and the synthetic inertial measurement signal to determine the pure physiological signal includes: Based on the synthesized inertial measurement signal, drift removal and amplitude normalization are performed to determine the reference inertial measurement signal; Adaptive filtering analysis is performed based on the preprocessed physiological signal and the reference inertial measurement signal to determine the pure physiological signal.

4. The method according to claim 3, characterized in that, The step of performing adaptive filtering analysis based on the preprocessed physiological signal and the reference inertial measurement signal to determine the pure physiological signal includes: The motion noise signal is determined by calculating the noise based on the reference inertial measurement signal and the filter weighting coefficients. Based on the preprocessed physiological signal and the motion noise signal, artifact elimination calculations are performed to determine the pure physiological signal.

5. The method according to claim 4, characterized in that, The method further includes: The pure physiological signal is used as an error signal feedback, and the filter weight coefficients are updated with the goal of minimizing the mean square value of the error signal.

6. The method according to any one of claims 1-5, characterized in that, The method further includes: When the device is stationary, zero-bias calibration is performed based on the inertial measurement signal from the inertial measurement unit.

7. A physiological parameter detection device, characterized in that, include: The synchronous acquisition module is used to synchronously acquire initial physiological signals and multiple inertial measurement signals; wherein, the multiple inertial measurement signals are synchronously acquired by multiple independently set inertial measurement units, and the zero-bias instability range of each inertial measurement unit is the same, and the microelectromechanical processes are different from each other; The fusion module is used to perform fusion analysis on the multiple inertial measurement signals to determine the composite inertial measurement signal; The filtering module is used to perform adaptive filtering analysis based on the initial physiological signal and the synthetic inertial measurement signal to determine the pure physiological signal; The parameter identification module is used to extract and analyze the purified physiological signals to determine physiological parameters.

8. A wearable device, characterized in that, include: Physiological parameter acquisition device, used to acquire initial physiological signals; At least two inertial measurement units are provided, and each inertial measurement unit acquires one inertial measurement signal; wherein, the zero-bias instability range of each inertial measurement unit is the same, and the microelectromechanical processes are different from each other; The processor is connected to each of the inertial measurement units and the physiological parameter acquisition units. The processor is used to determine physiological parameters based on the initial physiological signal and multiple inertial measurement signals.

9. The wearable device according to claim 8, characterized in that, The processor is also used for: The multiple inertial measurement signals are fused and analyzed to determine the synthetic inertial measurement signal. Adaptive filtering analysis is performed based on the initial physiological signal and the synthetic inertial measurement signal to determine the pure physiological signal. Physiological parameters are determined based on the pure physiological signal.

10. The wearable device according to claim 8, characterized in that, The spacing between any two adjacent inertial measurement units is greater than a preset spacing, wherein the preset spacing indicates that the resonant frequencies of adjacent inertial measurement units are not interconnected.

11. The wearable device according to claim 10, characterized in that, The preset spacing is greater than or equal to 20 mm.

12. The wearable device according to claim 8, characterized in that, It also includes a circuit board and a reinforcing member. The inertial measurement unit is disposed on a first surface of the circuit board, and the reinforcing member is disposed on a second surface of the circuit board. The second surface is disposed opposite to the first surface. The first vertical projection of the reinforcing member on the first surface at least partially overlaps with the second vertical projection of the inertial measurement unit on the first surface.

13. The wearable device according to claim 8, characterized in that, The physiological parameter acquisition unit is connected to the external interrupt pin of each inertial measurement unit via a synchronization signal acquisition line. The physiological parameter acquisition unit is used to synchronously trigger the inertial measurement unit to acquire inertial measurement signals within the acquisition time window.

14. The wearable device according to claim 8, characterized in that, It also includes a housing, in which the physiological parameter acquisition device, the inertial measurement unit and the processor are all disposed inside the housing, and an air gap is provided between the inertial measurement unit and the housing.

Images