Multi-sensor fused vital sign and motion posture monitoring system and method

Abstract

The invention relates to a multi-sensor fused vital sign and motion posture monitoring system and method. The system comprises a photoelectric volume pulse wave acquisition module, an electrocardio acquisition module, an inertia measurement module, a synchronization module, an artifact suppression module, a contact state recognition module, a data quality evaluation module, a vital sign extraction module and an anomaly detection module. Synchronously acquiring a photoelectric volume pulse wave signal, an electrocardiosignal and motion information; suppressing motion artifacts in the photoplethysmography signal and / or the electrocardiosignal based on the motion information; identifying the contact state and generating a data quality indicator related to the measurement reliability in combination with the de-artifact physiological signal; when the data quality meets a preset condition, heart rate, oxyhemoglobin saturation and / or breathing information are extracted, and vital sign parameters are formed; and carrying out conjoint analysis on the artifact-removed physiological signal and the motion information, and marking heart rate rhythm abnormity and / or posture abnormity. According to the invention, the reliability and availability of vital sign monitoring can be improved.

Description

Technical Field

[0001] This invention relates to the fields of vital sign monitoring, motion posture monitoring and embedded system technology, specifically to a multi-sensor fusion system and method for monitoring vital signs and motion posture. Background Technology

[0002] Currently, continuous monitoring of vital signs (such as heart rate, blood oxygen saturation, and respiratory information) and human movement posture monitoring are gradually expanding from traditional medical institution scenarios to everyday applications, such as exercise and fitness monitoring, home rehabilitation, long-term management of chronic diseases, and fall warning for the elderly. To obtain these physiological and postural parameters, a common practice is to attach or wear detection devices on the human body. These devices acquire blood flow pulsation signals through photoplethysmography (PPG) sensors, cardiac electrical activity signals through electrocardiogram (ECG) circuitry, and acceleration and / or angular velocity information, etc., through inertial measurement units (IMUs), to describe motion state and posture. These contact-based measurement solutions typically feature low power consumption, compact structure, and long-term wearability, making them suitable for continuous monitoring.

[0003] However, during users' daily activities and even slight walking, PPG and ECG signals are often significantly interfered with. Specifically, PPG signals are susceptible to motion artifacts caused by factors such as minute slippage between the probe and skin, local tissue deformation under pressure, ambient light leakage, and perfusion changes, leading to unstable heart rate and blood oxygen readings. ECG signals are easily affected by factors such as electrode-skin contact impedance fluctuations, baseline drift, electromyography interference, and power frequency noise, resulting in problems such as inaccurate R-wave detection and discontinuous heart rhythm. If there is a lack of unified time synchronization between physiological signals and IMU signals, the device also has difficulty distinguishing which waveform changes correspond to real physiological abnormalities and which are merely spurious signals introduced by posture or movement.

[0004] Existing monitoring solutions often rely on fixed thresholds or filters to suppress noise. For example, heart rate output may be directly blocked during strenuous exercise, or preset bandpass or notch filters may be used to smooth the signal. Some devices also rely on a single indicator (such as signal amplitude or signal-to-noise ratio) to roughly determine the reliability of the current data. However, such solutions often lack accurate identification of the wearing contact status (e.g., good contact, looseness, or detachment). The system struggles to distinguish between situations where data is still being collected but of poor quality and situations where effective contact has been largely lost, thus affecting the stability of continuous monitoring. Furthermore, these devices typically only output physiological parameters such as heart rate and blood oxygenation, lacking the ability to identify, classify, and label abnormal states, making it difficult to distinguish between genuine cardiac arrhythmias and false anomalies caused by large-amplitude limb movements.

[0005] Therefore, it is necessary to develop a new multi-sensor fusion system and method for monitoring vital signs and motion posture. Summary of the Invention

[0006] The purpose of this invention is to provide a multi-sensor fusion system and method for monitoring vital signs and motion posture, so as to reduce false alarms caused by motion interference and improve the reliability and usability of vital sign monitoring.

[0007] In a first aspect, the present invention provides a multi-sensor fusion system for monitoring vital signs and motion posture, comprising: The photoplethysmography (PPG) acquisition module is used to acquire the PPG signal of the target object. The electrocardiogram (ECG) acquisition module is used to acquire the ECG signal of the target object; An inertial measurement module is used to acquire motion information of a target object, wherein the motion information includes at least one of acceleration information and angular velocity information; The synchronization module is used to synchronously acquire and timestamp the photoplethysmography (PPG) signal, the electrocardiogram (ECG) signal, and the motion information under a unified time reference. The artifact suppression module is used to suppress motion artifacts in the photoplethysmography (PPG) signal and / or electrocardiogram (ECG) signal based on the motion information provided by the inertial measurement module, so as to obtain an artifact-free physiological signal. The contact status recognition module is used to determine whether the current wearing contact status is good contact, loose contact, or detachment based on the amplitude characteristics, baseline characteristics, and / or lead contact impedance characteristics of the artifact-free physiological signal. The data quality assessment module is used to generate data quality indicators related to the reliability of the current measurement based on the artifact removal physiological signal and the contact state. The vital signs extraction module is used to extract heart rate, blood oxygen saturation and / or respiratory information from the artifact-free physiological signal when the data quality indicators meet the preset reliability conditions, so as to generate vital signs parameters. An anomaly detection module is used to identify and mark abnormal heart rate rhythms and / or postures based on the artifact-removed physiological signals, the motion information, and the data quality indicators, and to obtain anomaly labels. The communication module is used to send the vital signs parameters, the data quality indicators, and the abnormality markers to an external device.

[0008] Optionally, the synchronization module is configured to perform millisecond-level time alignment of the photoplethysmography (PPG) signal, the electrocardiogram (ECG) signal, and the motion information using a unified hardware timer interrupt and a global timestamp structure. This ensures that different types of signals can be matched one-to-one in subsequent processing, thereby achieving time alignment of multi-channel signals.

[0009] Optionally, the artifact suppression module uses the motion information output by the inertial measurement module as a reference input to perform adaptive filtering on the photoplethysmography (PPG) signal and / or the electrocardiogram (ECG) signal to suppress artifacts caused by the movement of the target object and obtain an artifact-free physiological signal. By introducing motion information as a reference input, interference components caused by limb movement, tissue compression, and skin micro-slippage can be suppressed when the wearer is walking, swinging their arms, or changing their body position, thereby improving the effectiveness of the physiological signal in motion.

[0010] Optionally, the contact state recognition module distinguishes between good contact, loose contact, and detachment by detecting the ratio between the DC and AC components of the photoplethysmography (PPG) signal and / or by detecting changes in the electrode-skin contact impedance of the ECG acquisition circuit. This allows for the determination of the adhesion quality between the sensors (including the PPG acquisition module and the ECG acquisition module) and the skin.

[0011] Optionally, the data quality assessment module is configured to generate the data quality metric based on at least one of the following: Signal-to-noise ratio of physiological signals after artifact removal; Morphological stability of physiological waveforms; The rhythmic stability of the cardiac cycle or pulse cycle; The contact status is provided by the contact status recognition module. Data quality indicators can comprehensively consider the signal-to-noise ratio of the artifact-free physiological signal, the stability of the waveform morphology, the rhythmic stability of the cardiac cycle or pulse cycle, and the aforementioned wearing contact status, thereby quantifying the reliability of the measurement for the current time period.

[0012] Optionally, the vital signs extraction module includes: A heart rate extraction unit is used to perform R-wave detection on the electrocardiogram signal to obtain heart rate, and / or perform pulse wave peak detection on the photoplethysmography signal to obtain pulse rate; The blood oxygen saturation extraction unit is used to calculate blood oxygen saturation based on the ratio of photoplethysmography pulse wave signals of different wavelengths. And / or a respiratory information extraction unit, used to estimate respiratory-related information based on the envelope changes of the photoplethysmography (PPG) signal and / or based on the baseline fluctuations of the electrocardiogram (ECG) signal. Through linkage with the data quality indicators, the system outputs the corresponding vital sign parameters only when the current time period is deemed reliable.

[0013] Optionally, the anomaly detection module includes a deep learning model. This model uses the artifact-removed physiological signals and / or the motion information as input features to classify suspected abnormal waveforms of abnormal heart rate rhythm and / or posture, and outputs anomaly markers. The system outputs not only the vital signs parameters themselves but also abnormal alerts regarding the wearer's condition.

[0014] Secondly, the multi-sensor fusion method for monitoring vital signs and motion posture according to the present invention employs the multi-sensor fusion system for monitoring vital signs and motion posture as described in the present invention, and the method includes the following steps: Acquire photoplethysmography (PPG) signals, electrocardiogram (ECG) signals, and motion information of the target object, wherein the motion information includes at least one of acceleration information and angular velocity information; The photoplethysmography (PPG) signal, the electrocardiogram (ECG) signal, and the motion information are synchronously acquired under a unified time reference and timestamped. Based on the motion information, artifact suppression processing is performed on the photoplethysmography signal and / or the electrocardiogram signal to obtain artifact-free physiological signals. Based on the amplitude characteristics, baseline characteristics, and / or lead contact impedance characteristics of the artifact-removed physiological signal, the contact state between the system and the target object is identified. Based on the artifact removal physiological signals and the contact state, the current acquisition quality is evaluated, and data quality indicators related to measurement reliability are generated. When the data quality indicators meet the preset reliability conditions, heart rate, blood oxygen saturation and / or respiratory information are extracted from the artifact-free physiological signals, and corresponding vital sign parameters are generated. Based on the artifact-removed physiological signals and / or the motion information, suspected abnormal states are detected and abnormal markers are generated. The vital signs parameters, the data quality indicators, and the abnormality markers are packaged and sent to an external device for display and / or storage.

[0015] Optionally, the artifact suppression process includes: Using the motion information as a reference input, the least mean square adaptive filtering, recursive least square adaptive filtering, or Kalman filter-based estimation is performed on the photoplethysmography (PPG) signal and / or the electrocardiogram (ECG) signal.

[0016] Optionally, the check for the suspected abnormal state includes: The artifact removal physiological signals and / or motion information are analyzed using a deep learning model to identify abnormal conditions such as abnormal heart rate rhythm and / or abnormal posture, and to output corresponding abnormality markers.

[0017] The beneficial effects of this invention are: (1) The present invention synchronously acquires photoplethysmography pulse wave signal, electrocardiogram signal and motion information of inertial measurement unit under a unified time reference, and marks each signal with a timestamp so that different types of signals can correspond on the same time axis, avoiding time sequence misalignment across sensors.

[0018] (2) The present invention uses motion information obtained by inertial measurement module to suppress motion artifacts in photoplethysmography pulse wave signal and / or electrocardiogram signal to obtain artifact-free physiological signal, so that vital signs parameters such as heart rate, blood oxygen saturation and / or respiratory information can still be stably extracted when the wearer is walking, swinging arms and other movement states.

[0019] (3) The present invention identifies the wearing contact status (good contact, loose contact or detachment) based on the physiological signal amplitude characteristics, baseline characteristics and / or lead contact impedance characteristics, and combines the contact status with the data quality indicators related to the reliability of artifact removal physiological signal generation and measurement, thereby providing a quantifiable reliability judgment for each monitoring data segment.

[0020] (4) The present invention outputs heart rate, blood oxygen saturation and / or respiratory information and forms vital signs parameters only when the data quality indicators meet the preset conditions. The vital signs parameters and corresponding quality indicators are sent to external devices through the communication module so that downstream users can display, record or analyze them based on the credibility.

[0021] (5) The present invention performs joint analysis of artifact-free physiological signals and motion information through an anomaly detection module. The anomaly detection module can use a deep learning model to identify and mark abnormal heart rate rhythm and / or posture, so as to realize the synchronous output of vital signs parameters and abnormal status prompts. Attached Figure Description

[0022] Figure 1 This is a block diagram of the overall structure of the multi-sensor fusion vital signs and motion posture monitoring system of this application; Figure 2 This is a flowchart of the multi-sensor fusion method for monitoring vital signs and motion posture in this application; Figure 3 This is a schematic diagram of time synchronization and timestamp marking in this application; Figure 4 This is a functional block diagram of the artifact suppression module of this application; Figure 5 This is a flowchart of the wearing contact state recognition and data quality assessment in this application; Figure 6 This is a schematic diagram of the anomaly detection module of this application; Figure 7 This is a schematic diagram of the system communication and data transmission provided in this application; Figure 8 This is a schematic block diagram of the electronic device provided in this application; Figure labeling: 1-Photoplethysmography (PPG) acquisition module; 2-Electrocardiogram (ECG) acquisition module; 3-Inertial measurement module; 4-Synchronization module; 5-Artifact suppression module; 6-Contact status recognition module; 7-Data quality assessment module; 8-Vital signs extraction module; 9-Anomaly detection module; 10-Communication module; 11-Processor; 12-Memory. Detailed Implementation

[0023] The embodiments of the present invention will be described below with reference to the accompanying drawings and preferred embodiments. Those skilled in the art can understand other advantages and effects of the present invention from the content disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and various details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of the present invention. It should be understood that the preferred embodiments are only for illustrating the present invention and not for limiting the scope of protection of the present invention.

[0024] See Figure 1 In this embodiment of the application, a multi-sensor fusion vital signs and motion posture monitoring system includes a photoplethysmography (PPG) pulse wave acquisition module 1, an electrocardiogram (ECG) acquisition module 2, an inertial measurement module 3, a synchronization module 4, an artifact suppression module 5, a contact state recognition module 6, a data quality assessment module 7, a vital signs extraction module 8, an anomaly detection module 9, and a communication module 10. The above modules can be implemented in the form of hardware circuit units, firmware logic, software functional modules, or combinations thereof.

[0025] The above modules work together to extract vital signs parameters, perform motion / posture analysis, identify contact status, assess data quality, and mark and output abnormal statuses.

[0026] In this embodiment, the artifact-free physiological signal refers to the physiological signal processed by the artifact suppression module 5, based on the photoplethysmography (PPG) signal and / or the electrocardiogram (ECG) signal, where artifacts have been suppressed. Motion information refers to at least one of the acceleration and angular velocity information provided by the inertial measurement module 3. In some examples, it may also include motion / attitude information derived from the acceleration and / or angular velocity information. " / " indicates "or".

[0027] This system simultaneously acquires photoplethysmography (PPG), electrocardiogram (ECG), and motion information under a unified time reference. It utilizes motion information to suppress motion artifacts in physiological signals and combines this with the identification of the wearing contact status to determine the reliability of the current measurement results. Simultaneously, it detects and marks suspected abnormalities. Furthermore, this system possesses anomaly detection capabilities based on a deep learning model. It jointly analyzes the acquired PPG and ECG signals and their corresponding motion information to automatically identify events such as abnormal heart rate rhythms and posture abnormalities, reducing false alarms caused by motion interference. This improves the reliability and usability of vital sign monitoring in everyday scenarios.

[0028] The following is a detailed explanation of each module: In this embodiment, the photoplethysmography (PPG) acquisition module 1 is used to acquire the PPG signal from the wearing area (e.g., wrist) of the target object. The PPG acquisition module 1 may include an emitting light source (e.g., a red or infrared light-emitting diode) and a photodetector to detect the periodic changes in the intensity of reflected or transmitted light from local tissues in response to heartbeats.

[0029] For example, the PPG signal includes: (1) The low-frequency, slowly changing baseline component (i.e., DC component) is related to blood perfusion level and tissue optical path; (2) The pulsatile component (AC component) associated with each heartbeat reflects the pulse wave intensity.

[0030] The PPG signal will be used for blood oxygen saturation estimation, pulse rate / heart rate estimation, and respiratory information estimation.

[0031] In this embodiment, the ECG acquisition module 2 is used to acquire ECG signals from the wearing site or adjacent sites of the target object to characterize cardiac electrical activity. The ECG acquisition module includes an ECG acquisition circuit and an ECG front-end analog circuit, which are connected. The ECG acquisition circuit acquires ECG signals, and the ECG front-end analog circuit amplifies, filters, and adjusts the bias of the weak ECG signals to obtain an ECG waveform suitable for digital processing. Furthermore, the ECG acquisition module 2 is also used to estimate signals related to the contact state, such as electrode-skin contact impedance, for use by the contact state recognition module 6.

[0032] In this embodiment, the inertial measurement module 3 is used to collect motion information of the target object's wearing position. For example, the inertial measurement module 3 is a combination of a three-axis accelerometer and a three-axis gyroscope. The motion information output by the inertial measurement module 3 can be used to characterize the wearer's motion state (such as walking, arm swinging, etc.) and posture state (such as raising an arm, etc.). This motion information will subsequently serve as reference input for the artifact suppression module 5, input features for the anomaly detection module 9, and auxiliary information for interpreting the credibility of vital signs.

[0033] In this embodiment of the application, the synchronization module 4 is used to synchronously acquire the signals output by the photoplethysmography (PPG) acquisition module 1, the electrocardiogram (ECG) acquisition module 2, and the inertial measurement module 3 under a unified time reference, and to assign timestamps to each signal.

[0034] See Figure 3 For example, the system can use a unified clock / timer interrupt to sample or read PPG signals, ECG signals, and motion information. Synchronization module 4 includes a timestamp generation unit, which assigns a unified timestamp to the current batch of data for each channel, resulting in multi-channel synchronized data with a unified timestamp. Through a unified time base, different types of signals can correspond on the same time axis, avoiding situations where the PPG signal originates from the first moment while the motion signal originates from the second moment. This millisecond-level alignment provides a foundation for subsequent artifact suppression, data quality assessment, and anomaly detection.

[0035] In one possible embodiment, the artifact suppression module 5 is used to reduce artifact interference caused by the wearer's movement, thereby obtaining a de-artifacted physiological signal that is closer to the true physiological state. See also Figure 4 The inputs to the artifact suppression module 5 include the original PPG signal and / or the original ECG signal, as well as the motion information output by the inertial measurement module 3; the output is the artifact-free physiological signal.

[0036] In one embodiment, the artifact suppression module 5 may employ an adaptive filtering structure, using the motion information output by the inertial measurement module 3 as a reference input to estimate and cancel the motion-coupled artifact components. Taking least mean square (LMS) adaptive filtering as an example, its weight update can be expressed as: , Where a(k) is the reference input vector at time k, including motion information at the current time and historical time; w(k) is the filter weight at time k; and w(k+1) is the updated filter weight at time K+1. This is the step size coefficient; Let be the error signal at time k.

[0037] The filtered output can be expressed as: .

[0038] Where y(k) is the filtered output at time k; This represents the transpose of w(k) at time k (the superscript T is the transpose symbol).

[0039] The physiological signal for artifact removal can then be expressed as: , in, This is the original PPG signal or the original ECG signal; This is the artifact removal physiological signal output by the artifact suppression module 5.

[0040] In this way, the system does not simply apply fixed bandpass or notch filtering to the signal, but uses motion information as a reference to interpret the perturbation components in the physiological signal, thereby maintaining high signal availability under conditions such as walking, arm swinging, local pressure, and relative skin slippage.

[0041] In one possible embodiment, the contact state recognition module 6 is used to identify the current wearing contact state, distinguishing between good contact, loose contact, or detachment. See also Figure 5 The contact state recognition module 6 can simultaneously use features from the PPG channel and the ECG channel.

[0042] In one embodiment, for a PPG signal, its DC and AC components can be extracted within a preset time window. Let the original PPG signal be... The signal after bandpass filtering (preserving the dominant heart rate component) is: Then we can define: , , in, This represents the number of sampling points within the window, where n is the number of sampling points.

[0043] Define the ratio of the AC component to the DC component: .

[0044] in, This represents the ratio of the AC component to the DC component; the contact state recognition module 6 can, based on... The size of the PPG probe determines the quality of its fit with the skin. For example, when... When the contact is above the first threshold, it is considered a good contact; when When the contact is between the first and second thresholds, it is considered a loose contact; when If the value falls below the second threshold, it is considered a detachment. The first and second thresholds are set according to actual needs, with the first threshold being greater than the second threshold.

[0045] In another embodiment, for ECG signals, the ECG acquisition module 2 can estimate the electrode-skin contact impedance to aid in contact determination. A test current of known amplitude can be applied without affecting normal measurements. The corresponding voltage change was measured. Estimate electrode-skin contact impedance: , in, This is the estimated electrode-skin contact impedance.

[0046] when When the threshold exceeds the third threshold (set according to actual needs), it can be directly determined as an electrode detachment or severe loosening.

[0047] The contact status identification module 6 ultimately outputs a contact status label (e.g., good contact, loose, or detached), which will be used by the data quality assessment module 7.

[0048] In one possible embodiment, the data quality assessment module 7 is used to generate data quality metrics related to the current measurement reliability, describing whether the signal is reliable for that time period. See also Figure 5 The data quality assessment module 7 can take the following information as input: (1) Physiological signals for artifact removal (i.e., the output of the artifact suppression module 5) ); (2) The contact status tag output by the contact status recognition module 6; (3) Statistical characteristics such as the morphological stability and rhythm stability of physiological waveforms.

[0049] In one embodiment, the data quality assessment module 7 can calculate data quality indicators. For example: , in, The signal-to-noise ratio of the artifact-free physiological signal can be estimated using the following formula: , in, The effective power near the heart rate can be taken. The power of high-frequency residual and non-physiological components can be taken. H represents waveform morphological stability (e.g., waveform similarity between adjacent cardiac / pulse cycles), R1 represents rhythm stability (e.g., consistency between adjacent RR intervals or pulse intervals), and C represents the quantitative coding of contact status (e.g., good = 1, loose = 0.5, detached = 0). α is the first weighting coefficient, β is the second weighting coefficient, γ is the third weighting coefficient, and δ is the fourth weighting coefficient. α, β, γ, and δ can be set empirically or by calibration.

[0050] In one possible embodiment, the data quality assessment module 7 outputs data quality indicators. This will be used as a gating condition for the subsequent vital signs extraction module 8, that is, only when... The system considers the current measurement reliable only when the value exceeds the fourth threshold (set according to actual needs). Data quality indicators output by data quality assessment module 7. It can be transmitted to an external device along with vital sign parameters via the communication module 10. The external device can then base its data transmission on data quality indicators. The system filters, marks, or adds prompts to the received vital sign parameters, thereby alerting users that the data in certain time periods has lower reliability (e.g., loose contact, strong movement interference, etc.).

[0051] In one possible embodiment, the vital signs extraction module 8 is used for data quality indicators When preset reliability conditions are met, vital signs parameters are extracted from the artifact-free physiological signals. Vital signs parameters may include, but are not limited to, heart rate, blood oxygen saturation, and / or respiratory information.

[0052] Heart rate extraction: The time interval between two adjacent R waves can be obtained by performing R-wave detection on the artifact-free ECG signal. (Unit: seconds), heart rate can be expressed as: ; HR stands for heart rate.

[0053] Alternatively, pulse peak detection can be performed on the artifact-free PPG signal to obtain the interval between adjacent pulse peaks. It can calculate pulse rate: .

[0054] PR stands for pulse rate.

[0055] Blood oxygen saturation (SpO2) extraction: In the embodiment of the dual-wavelength PPG signal, the AC component of the red light channel can be obtained separately. With DC and the AC component of the infrared channel. With DC component .

[0056] Define the ratio (R²): .

[0057] Using empirical calibration functions Estimate blood oxygen saturation: .

[0058] in, This refers to blood oxygen saturation.

[0059] Respiratory information extraction: The envelope of the PPG signal can be calculated and low-pass filtered to obtain a slowly fluctuating respiratory-related waveform r(t). Peak detection or frequency analysis of r(t) can then yield the respiratory rate or respiratory activity intensity.

[0060] The vital signs extraction module 8 outputs vital sign parameters such as heart rate, blood oxygen saturation, and respiratory information, which will be transmitted to the communication module 10 and can be compared with the data quality indicators at the same time. Related.

[0061] In one possible embodiment, the anomaly detection module 9 is used to identify and mark suspected abnormal states. See also Figure 6 The anomaly detection module 9 can receive: (1) Artifact removal physiological signals, i.e., the output of artifact suppression module 5 ; (2) Motion information, namely the motion information output by the inertial measurement module 3; (3) Data quality indicators, i.e., the output of data quality assessment module 7. .

[0062] In one embodiment, the anomaly detection module 9 includes a deep learning model or other classification model to perform joint analysis on the input. This can be formalized as: , Where s represents a segment of artifact-free physiological signal (e.g., PPG waveform and / or ECG waveform) within a time window, and m represents a segment of motion information (e.g., acceleration information, angular velocity information, or attitude estimation value) within the same time window. This indicates the data quality index corresponding to this time window. This is the corresponding exception marker.

[0063] Anomaly markers may include, but are not limited to: (1) Abnormal heart rhythm (e.g., suspected arrhythmia segment); (2) Abnormal posture (e.g., abnormal body position or abnormal activity pattern); (3) Other abnormal status prompts.

[0064] The anomaly detection module 9 can associate the anomaly marker with vital sign parameters for the corresponding time period, so that the indicator value at that moment and whether there is a suspicious anomaly can be displayed on the external device's interface simultaneously. This helps to distinguish between genuine physiological abnormalities and false anomalies caused only by strenuous exercise.

[0065] In one possible embodiment, the communication module 10 is used to transmit the vital sign parameters output by the vital sign extraction module 8 and the data quality indicators output by the data quality assessment module 7. The anomaly flags output by the anomaly detection module 9 are sent to external devices for display, recording, and / or further analysis. See also Figure 7 The communication module 10 can establish a communication connection with external devices via wireless serial port protocol and / or Bluetooth Low Energy, and transmit vital signs parameters, motion information, and data quality indicators. And abnormal markers are packaged and transmitted. External devices can display, record, alarm, or perform further analysis on this information.

[0066] External devices can be mobile terminals, host computers, medical monitoring terminals, or rehabilitation training terminals, etc. External devices can simultaneously display parameter values ​​and their corresponding quality indicators. This indicates which data segments are trustworthy and which are of low trustworthiness or are abnormal.

[0067] It should be understood that Figures 1 to 7 The module division shown is a logical division method; in the specific hardware implementation, several of these modules can be implemented by the same processor's software / firmware program; that is, different modules do not necessarily correspond to independent physical chips.

[0068] This system can be implemented as a wristband device, forearm strap device, upper arm strap device, finger clip device, chest patch device, electrode patch device, glove device, ankle strap device, or even a cushion or seat contact module. As long as the device can simultaneously acquire PPG, ECG, and IMU signals while in contact with human skin, it falls under the embodiments of this invention. Different wearing positions can affect specific thresholds (e.g., R...). ppg The judgment threshold, Z skin (The threshold for judgment), but it does not affect the overall technical solution of this application.

[0069] See Figure 2 In this embodiment of the application, a multi-sensor fusion method for monitoring vital signs and motion posture is provided, employing the multi-sensor fusion vital signs and motion posture monitoring system described in this embodiment of the application. The method includes the following steps: S1: Data Acquisition: Under conditions of contact with the human body, PPG signals, ECG signals, and motion information output by the inertial measurement module 3 (including at least one of acceleration information and angular velocity information) are acquired to form multi-channel raw physiological data and motion data.

[0070] S2: Synchronization and Timestamp: Using synchronization module 4, PPG signals, ECG signals and motion information are synchronously acquired under a unified time reference, and a unified timestamp is added to each type of signal to obtain multi-channel synchronized data with a unified time reference.

[0071] S3: Artifact Suppression: Based on the motion information provided by the inertial measurement module 3, artifact suppression processing is performed on the PPG signal and / or ECG signal to obtain artifact-free physiological signals.

[0072] S4: Contact Status Identification: Based on physiological signal amplitude characteristics, baseline characteristics (mainly referring to the zero-point drift and low-frequency components of the signal, such as the average value or zero point of the signal, specifically referring to the DC component of the signal here) and / or lead contact impedance characteristics, identify the current wearing contact status (good contact, loose contact, or detachment), and / or estimate the electrode-skin contact impedance.

[0073] S5: Data Quality Assessment: Combining artifact removal physiological signals, contact status, and other information, generate data quality indicators related to measurement reliability. .

[0074] S6: Vital Signs Extraction: When the data quality indicators When the preset reliability conditions are met, vital signs parameters, including heart rate, blood oxygen saturation and / or respiratory information, are extracted from the artifact-free physiological signals.

[0075] S7: Anomaly Detection and Labeling: Based on the artifact-free physiological signal and the motion information output by the inertial measurement module 3, suspected abnormal states are detected, and abnormal heart rate rhythm and / or posture are identified and labeled.

[0076] S8: Data Packaging and Transmission: Packaging vital signs parameters and data quality indicators... The exception markers are packaged and sent to an external device via communication module 1 for display, recording, and / or storage.

[0077] It should be understood that S1 to S8 are given in logical order, but in actual implementation, some steps can be executed periodically in parallel or run in a pipeline manner. For example, contact status recognition S4 and data quality assessment S5 can be continuously refreshed, and anomaly detection S7 can be performed in real time along with vital sign extraction S6.

[0078] This method can simultaneously acquire photoplethysmography (PPG) signals, electrocardiogram (ECG) signals, and motion information from the inertial measurement module under a unified time reference. It then uses this motion information to suppress motion artifacts in the PPG and ECG signals, resulting in artifact-free physiological signals. The current wearing contact state is identified based on physiological signal characteristics and / or lead contact impedance. Data quality indicators related to measurement reliability are generated by combining the contact state with the artifact-free physiological signals. When the data quality indicators meet preset conditions, heart rate, blood oxygen saturation, and / or respiratory information are extracted from the artifact-free physiological signals to form vital sign parameters. Furthermore, based on the artifact-free physiological signals and motion information, an anomaly detection module (which may include a deep learning model) identifies and marks abnormal heart rate rhythms and / or posture abnormalities, thereby improving the continuity and reliability of vital sign monitoring under daily activity conditions.

[0079] join Figure 8 This application also provides an electronic device, which includes a processor 11 and a memory 12. The memory 12 stores instructions, which, when executed by the processor 11, are used to perform at least a portion of the steps in the method described in this application, including: multi-source signal acquisition, time synchronization, artifact suppression, contact state recognition, data quality assessment, vital sign extraction, anomaly detection, and external communication output. This electronic device can be a wearable / attached monitoring device or an external terminal device (such as a host computer, mobile terminal, or medical monitoring terminal).

[0080] It should be understood that the instructions can be stored on the storage medium in the form of firmware, software programs, driver modules, or algorithm libraries, and loaded and executed by the processor 11 at runtime. Different implementations do not constitute a limitation on the scope of protection of this invention; as long as their functions implement at least one or more of the above steps, they belong to the embodiments of this invention.

[0081] In this embodiment, the processor 11 or memory 12 can be a single-chip microcontroller unit (MCU) or a combination of a main processor and a coprocessor. The artifact suppression module 5, contact state recognition module 6, data quality assessment module 7, vital sign extraction module 8, and anomaly detection module 9 can be implemented, in whole or in part, through firmware or software programs executed by the processor, rather than necessarily being separate dedicated hardware modules. The memory 12 can be used to store multi-channel data with a unified timestamp, artifact-free physiological signals for corresponding time periods, contact state tags, and data quality indicators. Vital signs parameters and abnormal markers.

[0082] The above embodiments are preferred embodiments of the present invention, but the embodiments of the present invention are not limited to the above embodiments. Any changes, modifications, substitutions, combinations, or simplifications made without departing from the spirit and principle of the present invention shall be considered equivalent substitutions and shall be included within the protection scope of the present invention.

Claims

1. A multi-sensor fusion system for monitoring vital signs and motion posture, characterized in that, include: Photoplethysmography (PPG) acquisition module (1) is used to acquire the PPG signal of the target object; ECG acquisition module (2) is used to acquire the ECG signal of the target object; An inertial measurement module (3) is used to acquire motion information of a target object, wherein the motion information includes at least one of acceleration information and angular velocity information; The synchronization module (4) is used to synchronously acquire and mark the timestamps of the photoplethysmography pulse wave signal, the electrocardiogram signal and the motion information under a unified time reference. The artifact suppression module (5) is used to suppress motion artifacts in the photoplethysmography (PPG) signal and / or electrocardiogram (ECG) signal based on the motion information provided by the inertial measurement module (3) to obtain an artifact-free physiological signal. The contact status identification module (6) is used to determine whether the current wearing contact status is good contact, loose contact or detachment based on the amplitude characteristics, baseline characteristics and / or lead contact impedance characteristics of the artifact removal physiological signal. The data quality assessment module (7) is used to generate data quality indicators related to the reliability of the current measurement based on the artifact removal physiological signal and the contact state. The vital signs extraction module (8) is used to extract heart rate, blood oxygen saturation and / or respiratory information from the artifact-free physiological signal when the data quality indicators meet the preset reliability conditions, so as to generate vital signs parameters. The anomaly detection module (9) is used to identify and mark abnormal heart rate rhythm and / or abnormal posture based on the artifact removal physiological signal, the motion information and the data quality index, and obtain anomaly labels; The communication module (10) is used to send the vital signs parameters, the data quality indicators and the abnormality markers to an external device.

2. The multi-sensor fusion vital signs and motion posture monitoring system according to claim 1, characterized in that, The synchronization module (4) is configured to perform millisecond-level time alignment of the photoplethysmography signal, the electrocardiogram signal and the motion information through a unified hardware timer interrupt and a global timestamp structure.

3. The multi-sensor fusion vital signs and motion posture monitoring system according to claim 1, characterized in that, The artifact suppression module (5) takes the motion information output by the inertial measurement module (3) as a reference input and performs adaptive filtering on the photoplethysmography pulse wave signal and / or the electrocardiogram signal to suppress artifacts caused by the motion of the target object and obtain an artifact-free physiological signal.

4. The multi-sensor fusion vital signs and motion posture monitoring system according to claim 1, characterized in that, The contact state recognition module (6) distinguishes between good contact, loose contact and detachment by detecting the ratio between the DC component and the AC component of the photoplethysmography signal and / or detecting the change in the electrode-skin contact impedance of the ECG acquisition circuit.

5. The multi-sensor fusion vital signs and motion posture monitoring system according to claim 1, characterized in that, The data quality assessment module (7) is configured to generate the data quality metric based on at least one of the following: Signal-to-noise ratio of physiological signals after artifact removal; Morphological stability of physiological waveforms; The rhythmic stability of the cardiac cycle or pulse cycle; And the contact state given by the contact state recognition module (6).

6. The multi-sensor fusion vital signs and motion posture monitoring system according to claim 1, characterized in that, The vital signs extraction module (8) includes: A heart rate extraction unit is used to perform R-wave detection on the electrocardiogram signal to obtain heart rate, and / or perform pulse wave peak detection on the photoplethysmography signal to obtain pulse rate; The blood oxygen saturation extraction unit is used to calculate blood oxygen saturation based on the ratio of photoplethysmography pulse wave signals of different wavelengths. And / or a respiratory information extraction unit, used to estimate respiratory-related information based on the envelope change of the photoplethysmography pulse wave signal and / or based on the baseline fluctuation of the electrocardiogram signal.

7. The multi-sensor fusion vital signs and motion posture monitoring system according to claim 1, characterized in that, The anomaly detection module (9) includes a deep learning model, which uses the artifact-removed physiological signal and / or the motion information as input features to classify suspected abnormal waveforms of abnormal heart rate rhythm and / or abnormal posture and output anomaly labels.

8. A method for monitoring vital signs and motion posture using multi-sensor fusion, characterized in that, The method of using the multi-sensor fusion vital signs and motion posture monitoring system as described in any one of claims 1 to 7 includes the following steps: Acquire photoplethysmography (PPG) signals, electrocardiogram (ECG) signals, and motion information of the target object, wherein the motion information includes at least one of acceleration information and angular velocity information; The photoplethysmography (PPG) signal, the electrocardiogram (ECG) signal, and the motion information are synchronously acquired under a unified time reference and timestamped. Based on the motion information, artifact suppression processing is performed on the photoplethysmography signal and / or the electrocardiogram signal to obtain artifact-free physiological signals. Based on the amplitude characteristics, baseline characteristics, and / or lead contact impedance characteristics of the artifact-removed physiological signal, the contact state between the system and the target object is identified. Based on the artifact removal physiological signals and the contact state, the current acquisition quality is evaluated, and data quality indicators related to measurement reliability are generated. When the data quality indicators meet the preset reliability conditions, heart rate, blood oxygen saturation and / or respiratory information are extracted from the artifact-free physiological signals, and corresponding vital sign parameters are generated. Based on the artifact removal physiological signals and / or the motion information, suspected abnormal states are detected and abnormal markers are generated; The vital signs parameters, the data quality indicators, and the abnormality markers are packaged and sent to an external device for display and / or storage.

9. The multi-sensor fusion method for monitoring vital signs and motion posture according to claim 8, characterized in that, The artifact suppression process includes: Using the motion information as a reference input, the photoplethysmography (PPG) signal and / or the electrocardiogram (ECG) signal are subjected to least mean square adaptive filtering, recursive least squares adaptive filtering, or Kalman filtering-based estimation.

10. The multi-sensor fusion method for monitoring vital signs and motion posture according to claim 8, characterized in that, The detection of suspected abnormal states includes: The artifact removal physiological signals and / or motion information are analyzed using a deep learning model to identify abnormal conditions such as abnormal heart rate rhythm and / or abnormal posture, and to output corresponding abnormality markers.

Images