Earphone-based ear health monitoring method
Through the integrated photoplethysmography sensor and cloud analysis of headphones, the problems of complex operation and motion artifact interference of traditional health monitoring equipment are solved, portable non-invasive physiological parameter monitoring and personalized health management are realized, and the accuracy of health assessment is improved. Sex and comprehensiveness.
Patent Information
- Application Number
- CN202411076305.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-07
- Publication Date
- 2024-11-22
AI Technical Summary
Traditional health monitoring equipment is complex to operate, making it difficult for users to use on their own, and sports artifacts interfere with health assessment accuracy, and existing headphone health monitoring equipment is difficult to achieve portable and multi-functional non-invasive physiological parameter monitoring.
The integrated photoplethysmography sensor of headphones collects photoelectric signal data, and obtains physiological parameters such as heart rate, blood oxygen saturation, body temperature and blood pressure through signal enhancement and physiological characteristic analysis. Combined with flexible circuit board and cloud analysis, multimodal fusion analysis and Movement interference is eliminated, personalized health reports are generated and voice prompts are used.
Achieve non-invasive and portable physiological parameter monitoring, improve data quality and accuracy, provide quantitative assessment of user health status and personalized health management suggestions, reduce exercise interference, and improve the comprehensiveness and reliability of health assessment. sex.
Smart Images

Figure CN118986302A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of earphone control technology, and in particular to an ear health monitoring method based on earphones. Background Art
[0002] With the widespread use of wearable devices such as smart watches and smart bracelets, people's demand for real-time and portable health monitoring is growing. As an important sensory organ of the human body, the ear has the characteristics of rich blood flow and stable position, which provides an ideal location for continuous and stable monitoring of physiological parameters. In addition, the relatively closed environment of the ear canal can reduce external interference and improve monitoring accuracy. Modern people pursue portability and versatility, and hope that a single device can meet multiple needs at the same time. As an electronic device with a high frequency of daily use, headphones have the potential to integrate health monitoring functions, so headphones with ear health monitoring functions have come into being. However, traditional health monitoring equipment usually requires professional equipment and complex operations, which are difficult for ordinary users to use by themselves; motion artifact interference during continuous monitoring will also greatly affect the accuracy of health assessment. Summary of the invention
[0003] Based on this, it is necessary for the present invention to provide an earphone-based ear health monitoring method to solve at least one of the above technical problems.
[0004] To achieve the above object, an ear health monitoring method based on earphones comprises the following steps:
[0005] Step S1: collecting raw photoelectric signal data through a photoplethysmography sensor integrated in the headset, wherein the photoplethysmography sensor includes a plurality of micro-LEDs of different wavelengths and a highly sensitive micro-photodiode; performing signal enhancement on the raw photoelectric signal data to obtain high-quality photoelectric signal data;
[0006] Step S2: Performing physiological characteristic analysis on the high-quality photoelectric signal data to obtain physiological characteristic data including heart rate, heart rate variability, blood oxygen saturation, body temperature and blood pressure estimation; transmitting the physiological characteristic data to the signal processing module through the flexible circuit board, and performing preliminary filtering and feature extraction to obtain physiological parameter data;
[0007] Step S3: Perform multimodal fusion analysis based on the physiological parameter data, and eliminate motion interference based on the pre-acquired motion data to obtain purified physiological data; send the purified physiological data to the cloud for in-depth analysis and long-term trend tracking, thereby obtaining in-depth analysis data;
[0008] Step S4: Evaluate the health status based on the in-depth analysis data and generate personalized health report data; transmit the personalized health report data back to the headset, and display the user's current health status through voice to obtain health prompt data.
[0009] The embodiment of the present invention collects photoelectric signal data through headphones, which can realize non-invasive physiological parameter monitoring without wearing additional sensor equipment; the multi-wavelength LED and high-sensitivity photodiode of the photoplethysmography sensor can provide more detailed and accurate heart rate and hemodynamic information; signal enhancement processing can improve the quality and accuracy of the photoelectric signal, reduce possible noise and interference, and thus improve the accuracy of subsequent physiological characteristic analysis. Through the physiological characteristic analysis of photoelectric signal data, multiple key physiological parameters can be obtained to provide a quantitative assessment of the user's health status; monitoring of physiological indicators such as heart rate, heart rate variability, blood oxygen saturation, body temperature, and blood pressure estimation can provide an assessment of cardiovascular health and overall health status; using flexible circuit boards to transmit physiological characteristic data and perform preliminary filtering and feature extraction can ensure the reliability and accuracy of the data, providing a better data basis for subsequent analysis. Multimodal fusion analysis can comprehensively consider the correlation between different physiological parameters and provide more comprehensive and accurate health assessment results; motion interference elimination can reduce the interference of motion on physiological parameters and improve the accuracy and reliability of data; through purified physiological data, the user's physiological state can be more accurately understood, and more targeted suggestions can be provided for health management and prevention; cloud-based deep analysis and long-term trend tracking can use big data and powerful algorithms to provide more comprehensive, accurate and personalized health analysis and suggestions. Health status assessment based on deep analysis data can provide more accurate and comprehensive health status judgments and help users understand their own health levels; the generated personalized health report data can be customized according to the user's specific situation and needs, and provide targeted health advice and guidance; by displaying health reports by voice, users can easily obtain health prompt information without viewing the screen or other devices. Health prompt data can help users pay attention to health problems in a timely manner and take positive measures to improve and manage them.
[0010] Preferably, step S1 comprises the following steps:
[0011] Step S11: Initializing the photoplethysmography sensor in the headset to obtain sensor status data, wherein the initialization includes activating micro LEDs of different wavelengths and a high-sensitivity photodiode;
[0012] Step S12: performing self-calibration on the initialized photoplethysmography sensor according to the sensor state data, thereby obtaining calibration parameter data;
[0013] Step S13: using the micro LED to emit light of a preset wavelength, and receiving reflected light through the micro photodiode, thereby obtaining original photoelectric signal data;
[0014] Step S14: amplifying and filtering the original photoelectric signal data according to the calibration parameter data, so as to obtain preliminary photoelectric signal data;
[0015] Step S15: performing wavelength-based separation on the preliminary photoelectric signal data to obtain multi-wavelength photoelectric signal data;
[0016] Step S16: Calculating the signal-to-noise ratio and signal quality index of the multi-wavelength photoelectric signal data to obtain signal quality evaluation data;
[0017] Step S17: Perform quality judgment on the signal quality assessment data according to a preset quality threshold. If the quality is lower than the preset threshold, re-collection is triggered to obtain high-quality photoelectric signal data.
[0018] The present invention can ensure that the sensor is in normal working condition and ready to receive light signals by initializing the photoplethysmography sensor, and activating micro-LEDs and high-sensitivity photodiodes of different wavelengths helps to provide a variety of light source and receiver options for subsequent steps for more accurate measurement and analysis. The self-calibration process can correct the internal deviations and inaccuracies of the sensor, thereby improving the accuracy and reliability of the measurement. By obtaining calibration parameter data, the subsequent steps can use these data to accurately process the original photoelectric signals. By using micro-LEDs to emit light of a specific wavelength and receiving reflected light through micro-photodiodes, the original photoelectric signals related to the tissue and blood in the headset can be obtained, and these original signals can provide information about biological parameters (such as heart rate, blood oxygen saturation, etc.). By applying the calibration parameter data, the original photoelectric signals can be amplified and filtered to enhance the biological signals of interest and reduce the interference of noise, which can improve the accuracy of subsequent signal processing and analysis. Through wavelength-based separation, the preliminary photoelectric signals can be grouped according to different wavelengths, and this separation can provide more detailed and targeted biological parameter information because different wavelengths of light have different absorption and scattering characteristics in tissues and blood. By calculating the signal-to-noise ratio and signal quality index, the quality of multi-wavelength photoelectric signals can be evaluated. These evaluation data can help determine whether the signal is interfered by noise, whether it has sufficient strength and clarity, and whether it is suitable for subsequent biological parameter calculation and analysis. By setting a quality threshold and judging the signal quality evaluation data, it can be determined whether the quality of the photoelectric signal meets the preset standard. If the signal quality is lower than the preset threshold, triggering re-acquisition can ensure the acquisition of high-quality photoelectric signal data to improve the accuracy and reliability of subsequent biological parameter calculation and analysis.
[0019] Preferably, step S15 comprises the following steps:
[0020] Step S151: performing multi-scale decomposition on the preliminary photoelectric signal data using discrete wavelet transform, thereby obtaining signal component data of different frequency bands;
[0021] Step S152: performing independent high-frequency noise removal processing on the signal component of each wavelength according to the signal component data by using a soft threshold method, thereby obtaining preliminarily denoised multi-wavelength signal data;
[0022] Step S153: using independent component analysis technology to separate the signal components corresponding to different physiological processes on the multi-wavelength signal data, thereby obtaining physiological signal component data;
[0023] Step S154: Perform short-time Fourier transform on the physiological signal component data to obtain multi-wavelength photoelectric signal data.
[0024] The present invention can decompose preliminary photoelectric signal data into signal components of different frequency bands through discrete wavelet transform. This multi-scale decomposition can provide information about biological signals in different frequency ranges, thereby helping to further analyze and interpret the characteristics of photoelectric signals. By applying the soft threshold method, the signal component of each wavelength can be independently denoised. This high-frequency noise removal process helps to reduce the noise introduced by factors such as environmental interference and electronic noise, and improve the quality and accuracy of multi-wavelength signal data. Through independent component analysis technology, multi-wavelength signal data can be separated into signal components corresponding to different physiological processes. This separation can identify and extract signal components related to specific physiological parameters, such as heartbeat signals, pulse signals, etc., thereby achieving quantitative analysis and evaluation of physiological processes. By applying short-time Fourier transform, physiological signal component data can be converted into multi-wavelength photoelectric signal data. This transformation can provide the changing characteristics of photoelectric signals of different wavelengths in time and frequency, helping to further analyze and interpret information related to physiological processes.
[0025] Preferably, step S2 comprises the following steps:
[0026] Step S21: segmenting the high-quality photoelectric signal data into time windows, and dynamically adjusting the window size using an adaptive window algorithm, thereby obtaining photoelectric signal segmented data;
[0027] Step S22: performing physiological characteristic analysis on the high-quality photoelectric signal segmentation data to obtain physiological characteristic data including heart rate, heart rate variability, blood oxygen saturation, body temperature and blood pressure estimation;
[0028] Step S23: performing abnormality detection and preliminary classification on the physiological characteristic data by machine learning technology, thereby obtaining abnormal labeling data;
[0029] Step S24: performing lossless compression encoding on the physiological characteristic data to obtain physiological characteristic compressed data;
[0030] Step S25: Packing the physiological characteristic compressed data and the abnormal mark data, and adding a timestamp and a user ID, thereby obtaining a physiological characteristic data packet;
[0031] Step S26: The physiological characteristic data packet is transmitted to the signal processing module through the flexible printed circuit board, and preliminary filtering and feature extraction are performed to obtain physiological parameter data.
[0032] The present invention can divide the continuous signal into smaller segments by segmenting the photoelectric signal data into time windows; the adaptive window algorithm can adjust the window size according to the dynamic characteristics of the signal, so that the signal contained in each segment has better local characteristics, thereby improving the accuracy and reliability of subsequent analysis. By performing physiological characteristic analysis on the photoelectric signal segmentation data, indicators related to the physiological state can be extracted; physiological characteristic data such as heart rate, heart rate variability, blood oxygen saturation, body temperature, and blood pressure estimation can provide important indications of the user's health and physiological state, supporting health monitoring and disease diagnosis. By using machine learning technology to perform abnormal detection and classification on physiological characteristic data, potential abnormal conditions can be automatically identified, and abnormal marking data can help medical staff quickly discover and handle abnormal conditions, take necessary medical measures in time, and improve the safety and effectiveness of health monitoring. Through lossless compression coding, the storage space occupied and transmission cost of physiological characteristic data can be reduced, while maintaining the integrity and accuracy of the data. The physiological characteristic compressed data occupies a small storage space, which is convenient for data transmission and storage, and saves resources. Packing physiological characteristic compressed data, abnormal marking data, timestamp and user ID into a data packet helps to integrate and organize data and facilitate subsequent data management and processing. The addition of timestamp and user ID can provide data timing information and identity recognition, and support data traceability and personalized analysis. Transmitting physiological characteristic data packets through flexible circuit boards can achieve fast and reliable data transmission. Preliminary filtering and feature extraction in the signal processing module helps to remove noise and extract information related to specific physiological parameters, thereby obtaining accurate physiological parameter data and providing a reliable data basis for subsequent analysis and application.
[0033] Preferably, step S22 comprises the following steps:
[0034] Step S221: performing peak detection and heart rate calculation on the photoelectric signal segmented data, and removing outliers based on median filtering, so as to obtain preliminary heart rate data;
[0035] Step S222: Calculating the continuous heartbeat intervals according to the preliminary heart rate data, and performing time domain and frequency domain analysis to obtain heart rate variability characteristic data;
[0036] Step S223: performing ratio calculation based on different wavelengths according to the heart rate variability characteristic data, thereby obtaining blood oxygen saturation data;
[0037] Step S224: analyzing the infrared light signal intensity variation of the photoelectric signal segmentation data, and estimating the body temperature according to a pre-established temperature-signal intensity calibration curve, thereby obtaining body temperature estimation data;
[0038] Step S225: performing pulse wave propagation time analysis on the photoelectric signal segment data to obtain blood pressure estimation data;
[0039] Step S226: combining the preliminary heart rate data, heart rate variability characteristic data, blood oxygen saturation data, body temperature estimation data and blood pressure estimation data into physiological characteristic data.
[0040] The present invention can obtain preliminary heart rate data by performing peak detection and heart rate calculation on the segmented data of the photoelectric signal; the outlier removal based on median filtering can effectively remove outliers caused by signal noise or interference, thereby improving the accuracy and reliability of the heart rate data. By calculating the continuous heartbeat interval and analyzing the time domain and frequency domain, the characteristic data of heart rate variability can be obtained. Heart rate variability is the variation pattern of the heartbeat interval, which provides important information on the functional state and overall health status of the cardiac autonomic nervous system. The characteristic data of heart rate variability can be used to evaluate the heart health, stress level and regulatory ability of the autonomic nervous system. According to the characteristic data of heart rate variability, the ratio calculation based on different wavelengths can be performed to infer the blood oxygen saturation. The absorption characteristics of light of different wavelengths in the blood are related to the blood oxygen saturation. The level of blood oxygen saturation can be estimated by comparing the light signals of each wavelength. Blood oxygen saturation is an important indicator for measuring the oxygen content in the blood, which is crucial for evaluating the functions of the respiratory and circulatory systems. By analyzing the changes in the intensity of infrared light signals, characteristics related to body temperature can be captured. By using the pre-established temperature-signal intensity calibration curve, the changes in the intensity of the light signal can be converted into an estimate of body temperature. The estimated body temperature data can provide important physiological indicators for health monitoring and disease diagnosis. By analyzing the pulse wave propagation time, changes in blood pressure can be inferred. The pulse wave propagation time reflects the time it takes for the heartbeat to be transmitted to a specific location and can be used to estimate blood pressure. The estimated blood pressure data is of great significance for monitoring blood circulation and cardiovascular health. Combining the above-extracted preliminary heart rate data, heart rate variability characteristic data, blood oxygen saturation data, body temperature estimation data, and blood pressure estimation data into physiological characteristic data can provide a comprehensive assessment of the overall health status. Taking these different aspects of physiological indicators into consideration can provide more comprehensive and accurate health status information, which is helpful for monitoring the individual's physiological health, assessing disease risks, and guiding medical intervention and health management.
[0041] Preferably, step S226 includes the following steps:
[0042] Step S2261: Time-aligning the preliminary heart rate data, the heart rate variability characteristic data, the blood oxygen saturation data, the body temperature estimation data, and the blood pressure estimation data, thereby obtaining time-aligned time data;
[0043] Step S2262: normalizing the time-aligned physiological data to obtain normalized physiological characteristic data;
[0044] Step S2263: performing credibility scoring on the normalized physiological characteristic data and marking low-credibility data, thereby obtaining physiological data quality score data;
[0045] Step S2264: Filter the normalized physiological characteristic data according to the physiological data quality score data to obtain the physiological characteristic data.
[0046] The present invention can ensure that different physiological data correspond to the same time point by time alignment; time alignment helps to integrate physiological data from different sources so that they have a consistent time scale, thereby facilitating subsequent data analysis and comprehensive evaluation. By normalizing the time-aligned physiological data, the dimensional differences between different physiological indicators can be eliminated, so that they can be compared and analyzed within the same numerical range. Normalization helps to reduce the deviation introduced by different data scales and improve the comparability and consistency of the data. By scoring the credibility of the normalized physiological characteristic data, the quality and reliability of the data can be evaluated. For each data point, the credibility score can be calculated based on factors such as data stability, noise level, and acquisition quality. The marking of low-credibility data can help subsequent data screening and analysis, ensuring that only high-quality physiological data is used for further processing. By screening according to the physiological data quality score data, low-quality or unreliable data points can be excluded, thereby obtaining high-quality physiological characteristic data. The screened data is more reliable and accurate, which can improve the credibility and effectiveness of subsequent analysis and research.
[0047] Preferably, step S26 comprises the following steps:
[0048] Step S261: transmitting the physiological characteristic data to the signal processing module through the flexible circuit board, and performing integrity check and decoding, so as to restore the physiological characteristic compressed data and abnormal mark data;
[0049] Step S262: performing Gaussian noise filtering on the physiological characteristic compressed data through a particle filter, thereby obtaining preliminary filtered physiological data;
[0050] Step S263: performing bandpass filtering on the preliminary filtered physiological data to retain signal components within a preset frequency range, thereby obtaining filtered physiological data;
[0051] Step S264: repairing the abnormal value of the filtered physiological data according to the abnormal marking data, thereby obtaining physiological abnormality repair data;
[0052] Step S265: extracting features of the physiological abnormality repair data to obtain preliminary physiological parameter data, wherein the features include average heart rate, maximum heart rate, minimum heart rate, heart rate variability index, average blood oxygen saturation and body temperature change trend;
[0053] Step S265: abnormally marking the preliminary physiological parameter data according to the abnormal marking data, identifying and marking abnormal events, thereby obtaining marked physiological parameter data.
[0054] The present invention can effectively transmit physiological characteristic data through the operation of flexible circuit board transmission and signal processing module, and perform integrity check and decoding to ensure the integrity and accuracy of the data. The restored physiological characteristic compressed data and abnormal marking data provide a basis for subsequent data processing and analysis. Through Gaussian noise filtering, the noise components in the physiological characteristic compressed data can be effectively removed, the quality and signal-to-noise ratio of the data can be improved, and the physiological data obtained by preliminary filtering provide a more reliable basis for subsequent signal processing and analysis. Through bandpass filtering, the physiological signal components within the preset frequency range can be selectively retained, and the interference signals in other frequency ranges can be removed. The physiological data obtained after filtering is more focused on the frequency range of interest, which is helpful for subsequent feature extraction and analysis. By repairing the outliers of the filtered physiological data according to the abnormal marking data, the outliers or noise points in the physiological data can be removed, and the accuracy and reliability of the data can be improved. The physiological abnormal repair data provides a more reliable data basis for subsequent feature extraction and parameter calculation. By extracting features from the physiological abnormal repair data, some important physiological parameters can be extracted from the data. These parameters include average heart rate, maximum heart rate, minimum heart rate, heart rate variability index, average blood oxygen saturation, and temperature change trend, etc., providing a quantitative description and evaluation of the individual's physiological state. By marking the initial physiological parameter data for abnormalities based on the abnormal marking data, abnormal events in the physiological parameters can be identified and annotated. These abnormal events include abnormal heart rate, abnormal blood oxygen saturation, abnormal body temperature, etc. The marked physiological parameter data facilitates further abnormality detection and health status monitoring.
[0055] Preferably, step S3 comprises the following steps:
[0056] Step S31: performing time alignment and normalization processing on the physiological parameter data to obtain standardized physiological parameter data;
[0057] Step S32: using an acceleration sensor to collect user motion data, and pre-processing to obtain motion data;
[0058] Step S33: Time-synchronizing the standardized physiological parameter data with the motion data to obtain multi-modal time series data;
[0059] Step S34: using a Kalman filter algorithm to eliminate motion interference in the multimodal time series data, thereby obtaining preliminary purified physiological data;
[0060] Step S35: Using independent component analysis technology to separate physiological signals and residual noise from the initially purified physiological data, thereby obtaining purified physiological data
[0061] Step S36: Send the purified physiological data to the cloud for in-depth analysis and long-term trend tracking, thereby obtaining in-depth analysis data.
[0062] The present invention can unify the physiological parameter data of different acquisition time points to the same time scale and perform data standardization through time alignment and normalization processing; standardized physiological parameter data helps to eliminate time differences and dimensional differences, provide a consistent data benchmark, and facilitate subsequent data analysis and comparison. By collecting acceleration sensor data, the user's motion data can be obtained, and the preprocessing operation can include filtering, noise removal, etc. to improve the quality and accuracy of the motion data; the motion data provides information about the user's activity level, posture changes, etc., and provides a basis for subsequent multimodal data fusion. By synchronizing the standardized physiological parameter data and motion data, the time axes of the two data sources can be aligned to ensure that they correspond at the same time point; the multimodal time series data obtained in this way combines physiological parameters and motion information, provides a more comprehensive data perspective, and helps to reveal the correlation and influence between physiology and motion. The Kalman filter algorithm can eliminate the interference of motion on physiological parameters by modeling and estimating the motion data in the multimodal time series data; by removing motion noise, preliminary purified physiological data is obtained, and the accuracy and reliability of physiological data are improved. Independent component analysis technology can decompose the initially purified physiological data into independent physiological signals and residual noise components. By separating physiological signals and noise, the characteristics and change trends of physiological signals can be further extracted to obtain cleaner and more interpretable purified physiological data. By sending the purified physiological data to the cloud for in-depth analysis, cloud computing power and algorithm models can be used to perform more complex and comprehensive analysis of the data, including pattern recognition, anomaly detection, and health status assessment of physiological data; at the same time, long-term trend tracking can help monitor changes and trends in an individual's physiological state, providing more in-depth health management and early warning functions. BRIEF DESCRIPTION OF THE DRAWINGS
[0063] Other features, objects and advantages of the present invention will become more apparent from the detailed description of non-limiting embodiments thereof made with reference to the following drawings:
[0064] Figure 1 The present invention is a flowchart of the steps of the ear health monitoring method based on earphones;
[0065] Figure 2 for Figure 1 Detailed step flow diagram of step S1;
[0066] Figure 3 for Figure 1 Detailed step flow chart of step S2 in FIG. DETAILED DESCRIPTION
[0067] The technical method of the present invention is described clearly and completely below in conjunction with the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by technicians in this field without creative work are within the scope of protection of the present invention.
[0068] In addition, the accompanying drawings are only schematic illustrations of the present invention and are not necessarily drawn to scale. The same reference numerals in the figures represent the same or similar parts, and their repeated description will be omitted. Some of the block diagrams shown in the accompanying drawings are functional entities and do not necessarily correspond to physically or logically independent entities. The functional entities can be implemented in software form, or implemented in one or more hardware modules or integrated circuits, or implemented in different networks and / or processor methods and / or microcontroller methods.
[0069] It should be understood that, although the terms "first", "second", etc. may be used herein to describe various units, these units should not be limited by these terms. These terms are used only to distinguish one unit from another unit. For example, without departing from the scope of the exemplary embodiments, the first unit may be referred to as the second unit, and similarly the second unit may be referred to as the first unit. The term "and / or" used herein includes any and all combinations of one or more of the listed associated items.
[0070] To achieve this, please refer to Figures 1 to 3 The present invention provides an ear health monitoring method based on earphones, the method comprising the following steps:
[0071] Step S1: collecting raw photoelectric signal data through a photoplethysmography sensor integrated in the headset, wherein the photoplethysmography sensor includes a plurality of micro-LEDs of different wavelengths and a highly sensitive micro-photodiode; performing signal enhancement on the raw photoelectric signal data to obtain high-quality photoelectric signal data;
[0072] Step S2: Performing physiological characteristic analysis on the high-quality photoelectric signal data to obtain physiological characteristic data including heart rate, heart rate variability, blood oxygen saturation, body temperature and blood pressure estimation; transmitting the physiological characteristic data to the signal processing module through the flexible circuit board, and performing preliminary filtering and feature extraction to obtain physiological parameter data;
[0073] Step S3: Perform multimodal fusion analysis based on the physiological parameter data, and eliminate motion interference based on the pre-acquired motion data to obtain purified physiological data; send the purified physiological data to the cloud for in-depth analysis and long-term trend tracking, thereby obtaining in-depth analysis data;
[0074] Step S4: Evaluate the health status based on the in-depth analysis data and generate personalized health report data; transmit the personalized health report data back to the headset, and display the user's current health status through voice to obtain health prompt data.
[0075] In the embodiment of the present invention, reference Figure 1 The above is a schematic flow chart of the steps of an earphone-based ear health monitoring method of the present invention. In this example, the earphone-based ear health monitoring method includes the following steps:
[0076] Step S1: collecting raw photoelectric signal data through a photoplethysmography sensor integrated in the headset, wherein the photoplethysmography sensor includes a plurality of micro-LEDs of different wavelengths and a highly sensitive micro-photodiode; performing signal enhancement on the raw photoelectric signal data to obtain high-quality photoelectric signal data;
[0077] In the embodiment of the present invention, a plurality of micro LEDs with different wavelengths and a highly sensitive micro photodiode are installed in the earphone to collect photoelectric signals, and the micro LEDs are used to emit light of different wavelengths to irradiate the ear skin. The micro photodiode receives the light reflected and scattered by the skin tissue, generates the original photoelectric signal data, and performs signal enhancement processing on the original photoelectric signal data, and can use filtering, noise elimination and other technologies to improve the signal quality.
[0078] Step S2: Performing physiological characteristic analysis on the high-quality photoelectric signal data to obtain physiological characteristic data including heart rate, heart rate variability, blood oxygen saturation, body temperature and blood pressure estimation; transmitting the physiological characteristic data to the signal processing module through the flexible circuit board, and performing preliminary filtering and feature extraction to obtain physiological parameter data;
[0079] The embodiment of the present invention performs physiological characteristic analysis on high-quality photoelectric signal data, and can use signal processing and feature extraction algorithms to extract physiological characteristic data such as heart rate, heart rate variability, blood oxygen saturation, body temperature, etc.; the physiological characteristic data is transmitted to the signal processing module, and a flexible circuit board can be used to transmit the data to a suitable location; in the signal processing module, the physiological characteristic data is preliminarily filtered and feature extracted to improve the accuracy and availability of the data, and physiological parameter data such as heart rate, heart rate variability, blood oxygen saturation, body temperature and blood pressure estimation are extracted from the data obtained after the preliminary filtering and feature extraction.
[0080] Step S3: Perform multimodal fusion analysis based on the physiological parameter data, and eliminate motion interference based on the pre-acquired motion data to obtain purified physiological data; send the purified physiological data to the cloud for in-depth analysis and long-term trend tracking, thereby obtaining in-depth analysis data;
[0081] The embodiment of the present invention utilizes a multimodal fusion analysis method, combines different physiological parameter data, comprehensively considers multiple indicators such as heart rate, blood oxygen saturation, body temperature, etc., and conducts comprehensive evaluation and analysis; based on the pre-acquired motion data, the physiological parameter data is subjected to motion interference elimination to eliminate the influence of motion on the physiological parameter data, and purified physiological data is obtained. The purified physiological data is sent to the cloud for in-depth analysis and long-term trend tracking, and in-depth analysis can be performed using cloud computing, machine learning, data mining and other technologies. These technologies can be used to detect abnormal conditions, identify potential disease risks, monitor changing trends in physical health status, etc.
[0082] Step S4: Evaluate the health status based on the in-depth analysis data and generate personalized health report data; transmit the personalized health report data back to the headset, and display the user's current health status through voice to obtain health prompt data.
[0083] The embodiment of the present invention uses health assessment algorithms and models to assess and analyze the user's health status based on the deeply analyzed data; generates personalized health report data based on the assessment results, including information such as a summary of the health status, suggestions and preventive measures. The personalized health report data is transmitted back to the headset, and the current health status can be displayed to the user in the form of voice; health reminder data is provided, including recommended measures, precautions or questions that need to be consulted with a doctor, etc., to help the user improve and maintain health.
[0084] The embodiment of the present invention collects photoelectric signal data through headphones, which can realize non-invasive physiological parameter monitoring without wearing additional sensor equipment; the multi-wavelength LED and high-sensitivity photodiode of the photoplethysmography sensor can provide more detailed and accurate heart rate and hemodynamic information; signal enhancement processing can improve the quality and accuracy of the photoelectric signal, reduce possible noise and interference, and thus improve the accuracy of subsequent physiological characteristic analysis. Through the physiological characteristic analysis of photoelectric signal data, multiple key physiological parameters can be obtained to provide a quantitative assessment of the user's health status; monitoring of physiological indicators such as heart rate, heart rate variability, blood oxygen saturation, body temperature, and blood pressure estimation can provide an assessment of cardiovascular health and overall health status; using flexible circuit boards to transmit physiological characteristic data and perform preliminary filtering and feature extraction can ensure the reliability and accuracy of the data, providing a better data basis for subsequent analysis. Multimodal fusion analysis can comprehensively consider the correlation between different physiological parameters and provide more comprehensive and accurate health assessment results; motion interference elimination can reduce the interference of motion on physiological parameters and improve the accuracy and reliability of data; through purified physiological data, the user's physiological state can be more accurately understood, and more targeted suggestions can be provided for health management and prevention; cloud-based deep analysis and long-term trend tracking can use big data and powerful algorithms to provide more comprehensive, accurate and personalized health analysis and suggestions. Health status assessment based on deep analysis data can provide more accurate and comprehensive health status judgments and help users understand their own health levels; the generated personalized health report data can be customized according to the user's specific situation and needs, and provide targeted health advice and guidance; by displaying health reports by voice, users can easily obtain health prompt information without viewing the screen or other devices. Health prompt data can help users pay attention to health problems in a timely manner and take positive measures to improve and manage them.
[0085] Preferably, step S1 comprises the following steps:
[0086] Step S11: Initializing the photoplethysmography sensor in the headset to obtain sensor status data, wherein the initialization includes activating micro LEDs of different wavelengths and a high-sensitivity photodiode;
[0087] Step S12: performing self-calibration on the initialized photoplethysmography sensor according to the sensor state data, thereby obtaining calibration parameter data;
[0088] Step S13: using the micro LED to emit light of a preset wavelength, and receiving reflected light through the micro photodiode, thereby obtaining original photoelectric signal data;
[0089] Step S14: amplifying and filtering the original photoelectric signal data according to the calibration parameter data, so as to obtain preliminary photoelectric signal data;
[0090] Step S15: performing wavelength-based separation on the preliminary photoelectric signal data to obtain multi-wavelength photoelectric signal data;
[0091] Step S16: Calculating the signal-to-noise ratio and signal quality index of the multi-wavelength photoelectric signal data to obtain signal quality evaluation data;
[0092] Step S17: Perform quality judgment on the signal quality assessment data according to a preset quality threshold. If the quality is lower than the preset threshold, re-collection is triggered to obtain high-quality photoelectric signal data.
[0093] As an embodiment of the present invention, refer to Figure 2 As shown, Figure 1 Detailed step flow diagram of step S1 in the embodiment of the present invention, step S1 includes the following steps:
[0094] Step S11: Initializing the photoplethysmography sensor in the headset to obtain sensor status data, wherein the initialization includes activating micro LEDs of different wavelengths and a high-sensitivity photodiode;
[0095] The embodiment of the present invention activates micro LEDs of different wavelengths in the earphones in sequence through a control circuit so as to subsequently collect light signals of different wavelengths; and activates a high-sensitivity photodiode in the earphones through a control circuit so as to receive reflected light signals.
[0096] Step S12: performing self-calibration on the initialized photoplethysmography sensor according to the sensor state data, thereby obtaining calibration parameter data;
[0097] The embodiment of the present invention utilizes the sensor state data to perform self-calibration on the initialized photoplethysmography sensor; the self-calibration process may involve a calibration algorithm and a model, and parameter adjustment is performed based on the feedback of the sensor state data to optimize the performance of the sensor. The result of the self-calibration is calibration parameter data for subsequent signal processing and analysis.
[0098] Step S13: using the micro LED to emit light of a preset wavelength, and receiving reflected light through the micro photodiode, thereby obtaining original photoelectric signal data;
[0099] The embodiment of the present invention controls the circuit to make the micro LED of a specific wavelength emit light to the ear skin, and the micro photodiode receives the light reflected and scattered by the skin tissue to generate original photoelectric signal data.
[0100] Step S14: amplifying and filtering the original photoelectric signal data according to the calibration parameter data, so as to obtain preliminary photoelectric signal data;
[0101] The embodiment of the present invention uses calibration parameter data to perform signal amplification processing on the original photoelectric signal data to enhance the amplitude of the signal; the amplified signal is filtered, and technologies such as digital filters can be used to remove noise and unnecessary frequency components. After signal amplification and filtering processing, preliminary photoelectric signal data is obtained for subsequent analysis and processing.
[0102] Step S15: performing wavelength-based separation on the preliminary photoelectric signal data to obtain multi-wavelength photoelectric signal data;
[0103] The embodiment of the present invention performs wavelength separation according to optical signals of different wavelengths in the preliminary optical signal data; optical signals of different wavelengths can be separated by using filters or spectrum analysis techniques to obtain multi-wavelength optical signal data.
[0104] Step S16: Calculating the signal-to-noise ratio and signal quality index of the multi-wavelength photoelectric signal data to obtain signal quality evaluation data;
[0105] The embodiment of the present invention calculates the signal-to-noise ratio of multi-wavelength photoelectric signal data, evaluates the quality of the signal by comparing the ratio of the signal intensity to the noise level, and calculates the signal quality index. According to the characteristics of the signal and prior knowledge, a corresponding algorithm or model can be designed to evaluate the quality of the signal. The signal-to-noise ratio and the signal quality index calculation results are combined to obtain signal quality evaluation data for subsequent quality judgment.
[0106] Step S17: Perform quality judgment on the signal quality assessment data according to a preset quality threshold. If the quality is lower than the preset threshold, re-collection is triggered to obtain high-quality photoelectric signal data.
[0107] The embodiment of the present invention sets a preset quality threshold to determine whether the quality of the signal meets the requirements, and compares the signal quality evaluation data. If it is lower than the preset quality threshold, it is determined that the signal quality does not meet the requirements; when the signal quality does not meet the requirements, the re-collection step is triggered, and the photoelectric signal is collected and processed again to obtain high-quality photoelectric signal data.
[0108] The present invention can ensure that the sensor is in normal working condition and ready to receive light signals by initializing the photoplethysmography sensor, and activating micro-LEDs and high-sensitivity photodiodes of different wavelengths helps to provide a variety of light source and receiver options for subsequent steps for more accurate measurement and analysis. The self-calibration process can correct the internal deviations and inaccuracies of the sensor, thereby improving the accuracy and reliability of the measurement. By obtaining calibration parameter data, the subsequent steps can use these data to accurately process the original photoelectric signals. By using micro-LEDs to emit light of a specific wavelength and receiving reflected light through micro-photodiodes, the original photoelectric signals related to the tissue and blood in the headset can be obtained, and these original signals can provide information about biological parameters (such as heart rate, blood oxygen saturation, etc.). By applying the calibration parameter data, the original photoelectric signals can be amplified and filtered to enhance the biological signals of interest and reduce the interference of noise, which can improve the accuracy of subsequent signal processing and analysis. Through wavelength-based separation, the preliminary photoelectric signals can be grouped according to different wavelengths, and this separation can provide more detailed and targeted biological parameter information because different wavelengths of light have different absorption and scattering characteristics in tissues and blood. By calculating the signal-to-noise ratio and signal quality index, the quality of multi-wavelength photoelectric signals can be evaluated. These evaluation data can help determine whether the signal is interfered by noise, whether it has sufficient strength and clarity, and whether it is suitable for subsequent biological parameter calculation and analysis. By setting a quality threshold and judging the signal quality evaluation data, it can be determined whether the quality of the photoelectric signal meets the preset standard. If the signal quality is lower than the preset threshold, triggering re-acquisition can ensure the acquisition of high-quality photoelectric signal data to improve the accuracy and reliability of subsequent biological parameter calculation and analysis.
[0109] Preferably, step S15 comprises the following steps:
[0110] Step S151: performing multi-scale decomposition on the preliminary photoelectric signal data using discrete wavelet transform, thereby obtaining signal component data of different frequency bands;
[0111] The embodiment of the present invention uses discrete wavelet transform to perform multi-scale decomposition on preliminary photoelectric signal data. Discrete wavelet transform decomposes the signal into sub-signals of different frequency bands, which include low-frequency components and high-frequency components. By decomposing the signal using wavelet basis functions, signal component data of multiple scales can be obtained.
[0112] Step S152: performing independent high-frequency noise removal processing on the signal component of each wavelength according to the signal component data by using a soft threshold method, thereby obtaining preliminarily denoised multi-wavelength signal data;
[0113] The embodiment of the present invention applies a soft threshold method to perform denoising on the signal component data of each wavelength. The soft threshold method is a commonly used wavelet denoising method, which suppresses high-frequency noise by performing threshold processing on the wavelet coefficients of the signal. The signal component data of each wavelength is compared with the set threshold value according to its wavelet coefficient. If the wavelet coefficient is less than the threshold value, it is set to zero, otherwise the original value is retained. By performing independent soft threshold processing on the signal component of each wavelength, high-frequency noise can be removed to obtain preliminary denoised multi-wavelength signal data.
[0114] Step S153: using independent component analysis technology to separate the signal components corresponding to different physiological processes on the multi-wavelength signal data, thereby obtaining physiological signal component data;
[0115] The embodiment of the present invention uses independent component analysis technology to separate the components of multi-wavelength signal data. Independent component analysis is a blind source separation method that can separate mixed signals into independent component signals. Assuming that the multi-wavelength signal data is a mixture of signal components corresponding to different physiological processes, these independent component signals can be estimated through ICA; the separated physiological signal component data is subsequently analyzed and processed, such as the extraction of physiological parameters such as heart rate and pulse.
[0116] Step S154: Perform short-time Fourier transform on the physiological signal component data to obtain multi-wavelength photoelectric signal data.
[0117] The embodiment of the present invention divides the physiological signal component data into small time windows, applies short-time Fourier transform to the data of each time window, and converts the signal from the time domain to the frequency domain; thus, multi-wavelength photoelectric signal data is obtained, wherein the spectrum information of each wavelength can be used for further analysis and processing.
[0118] The present invention can decompose preliminary photoelectric signal data into signal components of different frequency bands through discrete wavelet transform. This multi-scale decomposition can provide information about biological signals in different frequency ranges, thereby helping to further analyze and interpret the characteristics of photoelectric signals. By applying the soft threshold method, the signal component of each wavelength can be independently denoised. This high-frequency noise removal process helps to reduce the noise introduced by factors such as environmental interference and electronic noise, and improve the quality and accuracy of multi-wavelength signal data. Through independent component analysis technology, multi-wavelength signal data can be separated into signal components corresponding to different physiological processes. This separation can identify and extract signal components related to specific physiological parameters, such as heartbeat signals, pulse signals, etc., thereby achieving quantitative analysis and evaluation of physiological processes. By applying short-time Fourier transform, physiological signal component data can be converted into multi-wavelength photoelectric signal data. This transformation can provide the changing characteristics of photoelectric signals of different wavelengths in time and frequency, helping to further analyze and interpret information related to physiological processes.
[0119] Preferably, step S2 comprises the following steps:
[0120] Step S21: segmenting the high-quality photoelectric signal data into time windows, and dynamically adjusting the window size using an adaptive window algorithm, thereby obtaining photoelectric signal segmented data;
[0121] Step S22: performing physiological characteristic analysis on the high-quality photoelectric signal segmentation data to obtain physiological characteristic data including heart rate, heart rate variability, blood oxygen saturation, body temperature and blood pressure estimation;
[0122] Step S23: performing abnormality detection and preliminary classification on the physiological characteristic data by machine learning technology, thereby obtaining abnormal labeling data;
[0123] Step S24: performing lossless compression encoding on the physiological characteristic data to obtain physiological characteristic compressed data;
[0124] Step S25: Packing the physiological characteristic compressed data and the abnormal mark data, and adding a timestamp and a user ID, thereby obtaining a physiological characteristic data packet;
[0125] Step S26: The physiological characteristic data packet is transmitted to the signal processing module through the flexible printed circuit board, and preliminary filtering and feature extraction are performed to obtain physiological parameter data.
[0126] As an embodiment of the present invention, refer to Figure 3 As shown, Figure 1 Detailed step flow diagram of step S2 in the embodiment of the present invention, step S2 includes the following steps:
[0127] Step S21: segmenting the high-quality photoelectric signal data into time windows, and dynamically adjusting the window size using an adaptive window algorithm, thereby obtaining photoelectric signal segmented data;
[0128] The embodiment of the present invention uses a photoelectric signal acquisition device to obtain high-quality photoelectric signal data, divides the photoelectric signal data into time windows, selects appropriate window lengths and overlap rates, and divides the photoelectric signal data into multiple continuous time windows. The window size is dynamically adjusted using an adaptive window algorithm, and the window size is adaptively adjusted according to the characteristics and changes of the signal to adapt to the dynamic changes of the signal, thereby obtaining photoelectric signal segmented data, each segment corresponding to a time window.
[0129] Step S22: performing physiological characteristic analysis on the high-quality photoelectric signal segmentation data to obtain physiological characteristic data including heart rate, heart rate variability, blood oxygen saturation, body temperature and blood pressure estimation;
[0130] The embodiment of the present invention uses a physiological feature analysis algorithm to process and analyze the photoelectric signal segmentation data, and extracts the physiological features of interest based on the characteristics of the signal and physiological knowledge. Common physiological features include heart rate, heart rate variability, blood oxygen saturation, body temperature, and blood pressure estimation. According to the needs, appropriate algorithms and methods are selected for feature extraction and calculation; physiological features are extracted from each photoelectric signal segmentation data to obtain corresponding feature data.
[0131] Step S23: performing abnormality detection and preliminary classification on the physiological characteristic data by machine learning technology, thereby obtaining abnormal labeling data;
[0132] The embodiment of the present invention uses machine learning technology to detect and classify abnormalities in physiological characteristic data, and builds appropriate machine learning models, such as support vector machine (SVM), random forest, etc. The trained model is used to predict and judge the physiological characteristic data and identify abnormal samples. The abnormal detection model can be trained based on known normal and abnormal data to obtain abnormal labeling data, marking whether each sample is abnormal.
[0133] Step S24: performing lossless compression encoding on the physiological characteristic data to obtain physiological characteristic compressed data;
[0134] The embodiment of the present invention uses a lossless compression coding algorithm to compress the physiological characteristic data. Common lossless compression algorithms include Huffman coding, arithmetic coding, etc. The physiological characteristic data is encoded and converted into smaller compressed data to ensure that the encoded data can be restored to the original physiological characteristic data.
[0135] Step S25: Packing the physiological characteristic compressed data and the abnormal mark data, and adding a timestamp and a user ID, thereby obtaining a physiological characteristic data packet;
[0136] The embodiment of the present invention combines the physiological feature compressed data and the abnormal mark data, and pairs the physiological feature compressed data of each sample with the corresponding abnormal mark data. Use a data packaging tool to package the paired data into a data packet, ensuring that the format and structure of the data packet can meet the requirements of subsequent processing and transmission. Add a timestamp and user ID, add a timestamp to each data packet, and record the generation time of the data packet; at the same time, associate the user ID of each data packet for subsequent data management and analysis.
[0137] Step S26: The physiological characteristic data packet is transmitted to the signal processing module through the flexible printed circuit board, and preliminary filtering and feature extraction are performed to obtain physiological parameter data.
[0138] The embodiment of the present invention uses a flexible circuit board to transmit the physiological characteristic data packet to the signal processing module. The flexible circuit board is flexible and can adapt to different transmission environments and requirements. Preliminary filtering and feature extraction are performed in the signal processing module, and the physiological characteristic data are filtered using an appropriate filter to remove noise and interference. The physiological parameter data of interest is extracted, and the required physiological parameters, such as heart rate, blood oxygen, etc., are extracted from the filtered signal according to actual needs.
[0139] The present invention can divide the continuous signal into smaller segments by segmenting the photoelectric signal data into time windows; the adaptive window algorithm can adjust the window size according to the dynamic characteristics of the signal, so that the signal contained in each segment has better local characteristics, thereby improving the accuracy and reliability of subsequent analysis. By performing physiological characteristic analysis on the photoelectric signal segmentation data, indicators related to the physiological state can be extracted; physiological characteristic data such as heart rate, heart rate variability, blood oxygen saturation, body temperature, and blood pressure estimation can provide important indications of the user's health and physiological state, supporting health monitoring and disease diagnosis. By using machine learning technology to perform abnormal detection and classification on physiological characteristic data, potential abnormal conditions can be automatically identified, and abnormal marking data can help medical staff quickly discover and handle abnormal conditions, take necessary medical measures in time, and improve the safety and effectiveness of health monitoring. Through lossless compression coding, the storage space occupied and transmission cost of physiological characteristic data can be reduced, while maintaining the integrity and accuracy of the data. The physiological characteristic compressed data occupies a small storage space, which is convenient for data transmission and storage, and saves resources. Packing physiological characteristic compressed data, abnormal marking data, timestamp and user ID into a data packet helps to integrate and organize data and facilitate subsequent data management and processing. The addition of timestamp and user ID can provide data timing information and identity recognition, and support data traceability and personalized analysis. Transmitting physiological characteristic data packets through flexible circuit boards can achieve fast and reliable data transmission. Preliminary filtering and feature extraction in the signal processing module helps to remove noise and extract information related to specific physiological parameters, thereby obtaining accurate physiological parameter data and providing a reliable data basis for subsequent analysis and application.
[0140] Preferably, step S22 comprises the following steps:
[0141] Step S221: performing peak detection and heart rate calculation on the photoelectric signal segmented data, and removing outliers based on median filtering, so as to obtain preliminary heart rate data;
[0142] The embodiment of the present invention performs peak detection on each photoelectric signal segment data, and uses an appropriate peak detection algorithm, such as a method based on threshold or waveform morphology, to detect the peak point in the photoelectric signal. The preliminary heart rate is calculated based on the time interval of the peak points, and the time interval between the peak points is calculated and converted into a heart rate value to obtain preliminary heart rate data. The preliminary heart rate data is subjected to outlier removal based on median filtering, and the preliminary heart rate data is smoothed using a median filter to remove outliers and noise.
[0143] Step S222: Calculating the continuous heartbeat intervals according to the preliminary heart rate data, and performing time domain and frequency domain analysis to obtain heart rate variability characteristic data;
[0144] The embodiment of the present invention uses the preliminary heart rate data to calculate the continuous heartbeat interval, and calculates the time interval between adjacent heartbeats according to the timestamp in the preliminary heart rate data to obtain the continuous heartbeat interval data. The continuous heartbeat interval data is analyzed in the time domain and frequency domain, and appropriate time domain and frequency domain analysis methods, such as standard deviation, mean, power spectrum density, etc., are applied to extract the heart rate variability features from the continuous heartbeat interval data.
[0145] Step S223: performing ratio calculation based on different wavelengths according to the heart rate variability characteristic data, thereby obtaining blood oxygen saturation data;
[0146] The embodiment of the present invention utilizes relevant parameters in the heart rate variability characteristic data, such as RR interval, frequency component, etc., to calculate the ratio of different wavelengths; and converts the ratio of different wavelengths into blood oxygen saturation data according to a pre-established blood oxygen saturation calculation algorithm.
[0147] Step S224: analyzing the infrared light signal intensity variation of the photoelectric signal segmentation data, and estimating the body temperature according to a pre-established temperature-signal intensity calibration curve, thereby obtaining body temperature estimation data;
[0148] The embodiment of the present invention performs intensity change analysis on the infrared light signal in the photoelectric signal segmentation data, and uses appropriate analysis algorithms, such as peak detection, filtering, time domain analysis, etc., to extract the intensity change information of the infrared light signal; and uses a pre-established temperature-signal intensity calibration curve to convert the intensity change of the infrared light signal into body temperature estimation data.
[0149] Step S225: performing pulse wave propagation time analysis on the photoelectric signal segment data to obtain blood pressure estimation data;
[0150] The embodiment of the present invention performs pulse wave detection on the segmented data of the photoelectric signal, and uses appropriate algorithms, such as peak detection, cross-correlation analysis, etc., to detect the pulse wave in the photoelectric signal; analyzes the propagation time of the pulse wave, and calculates the blood pressure estimation data based on the propagation time of the pulse wave in the photoelectric signal.
[0151] Step S226: combining the preliminary heart rate data, heart rate variability characteristic data, blood oxygen saturation data, body temperature estimation data and blood pressure estimation data into physiological characteristic data.
[0152] The embodiment of the present invention integrates various physiological characteristic data into one data set to ensure the consistency of the timestamps corresponding to the data. As needed, the data is standardized, normalized or otherwise preprocessed for subsequent analysis and application.
[0153] The present invention can obtain preliminary heart rate data by performing peak detection and heart rate calculation on the segmented data of the photoelectric signal; the outlier removal based on median filtering can effectively remove outliers caused by signal noise or interference, thereby improving the accuracy and reliability of the heart rate data. By calculating the continuous heartbeat interval and analyzing the time domain and frequency domain, the characteristic data of heart rate variability can be obtained. Heart rate variability is the variation pattern of the heartbeat interval, which provides important information on the functional state and overall health status of the cardiac autonomic nervous system. The characteristic data of heart rate variability can be used to evaluate the heart health, stress level and regulatory ability of the autonomic nervous system. According to the characteristic data of heart rate variability, the ratio calculation based on different wavelengths can be performed to infer the blood oxygen saturation. The absorption characteristics of light of different wavelengths in the blood are related to the blood oxygen saturation. The level of blood oxygen saturation can be estimated by comparing the light signals of each wavelength. Blood oxygen saturation is an important indicator for measuring the oxygen content in the blood, which is crucial for evaluating the functions of the respiratory and circulatory systems. By analyzing the changes in the intensity of infrared light signals, characteristics related to body temperature can be captured. By using the pre-established temperature-signal intensity calibration curve, the changes in the intensity of the light signal can be converted into an estimate of body temperature. The estimated body temperature data can provide important physiological indicators for health monitoring and disease diagnosis. By analyzing the pulse wave propagation time, changes in blood pressure can be inferred. The pulse wave propagation time reflects the time it takes for the heartbeat to be transmitted to a specific location and can be used to estimate blood pressure. The estimated blood pressure data is of great significance for monitoring blood circulation and cardiovascular health. Combining the above-extracted preliminary heart rate data, heart rate variability characteristic data, blood oxygen saturation data, body temperature estimation data, and blood pressure estimation data into physiological characteristic data can provide a comprehensive assessment of the overall health status. Taking these different aspects of physiological indicators into consideration can provide more comprehensive and accurate health status information, which is helpful for monitoring the individual's physiological health, assessing disease risks, and guiding medical intervention and health management.
[0154] Preferably, step S226 includes the following steps:
[0155] Step S2261: Time-aligning the preliminary heart rate data, the heart rate variability characteristic data, the blood oxygen saturation data, the body temperature estimation data, and the blood pressure estimation data, thereby obtaining time-aligned time data;
[0156] The embodiment of the present invention determines a reference time axis, selects a time axis as a reference, which can be a time axis of one of the physiological data, or creates another unified time axis. Each physiological data is aligned with the reference time axis to ensure that their timestamps are consistent, which can be achieved by using an interpolation or alignment algorithm.
[0157] Step S2262: normalizing the time-aligned physiological data to obtain normalized physiological characteristic data;
[0158] The embodiment of the present invention determines a normalization method and selects an appropriate normalization method, such as linear scaling, maximum and minimum value normalization or standardization, etc., according to the characteristics and requirements of the data. The time-aligned physiological data is normalized, and the data of each physiological feature is normalized and mapped into a unified numerical range to eliminate the dimensional differences between different feature data.
[0159] Step S2263: performing credibility scoring on the normalized physiological characteristic data and marking low-credibility data, thereby obtaining physiological data quality score data;
[0160] The embodiment of the present invention determines the standard or index for credibility scoring, and determines the standard or index for evaluating the credibility of physiological data according to specific needs, such as data stability, noise level, outliers, etc. The credibility of normalized physiological characteristic data is scored, and the normalized physiological characteristic data is evaluated according to the determined standard or index, and a credibility score is assigned to each data point to reflect its data quality; low-credibility data is marked, and low-credibility data points are marked according to the credibility score for subsequent data screening or analysis.
[0161] Step S2264: Filter the normalized physiological characteristic data according to the physiological data quality score data to obtain the physiological characteristic data.
[0162] The embodiment of the present invention sets a threshold value, and sets a threshold value according to the credibility score data and requirements to determine which data points are considered to be high-quality data. Data is screened according to the threshold value, and data points with quality scores higher than the threshold value in the physiological characteristic data are selected according to the set threshold value to form a final physiological characteristic data set.
[0163] The present invention can ensure that different physiological data correspond to the same time point by time alignment; time alignment helps to integrate physiological data from different sources so that they have a consistent time scale, thereby facilitating subsequent data analysis and comprehensive evaluation. By normalizing the time-aligned physiological data, the dimensional differences between different physiological indicators can be eliminated, so that they can be compared and analyzed within the same numerical range. Normalization helps to reduce the deviation introduced by different data scales and improve the comparability and consistency of the data. By scoring the credibility of the normalized physiological characteristic data, the quality and reliability of the data can be evaluated. For each data point, the credibility score can be calculated based on factors such as data stability, noise level, and acquisition quality. The marking of low-credibility data can help subsequent data screening and analysis, ensuring that only high-quality physiological data is used for further processing. By screening according to the physiological data quality score data, low-quality or unreliable data points can be excluded, thereby obtaining high-quality physiological characteristic data. The screened data is more reliable and accurate, which can improve the credibility and effectiveness of subsequent analysis and research.
[0164] Preferably, step S26 comprises the following steps:
[0165] Step S261: transmitting the physiological characteristic data to the signal processing module through the flexible circuit board, and performing integrity check and decoding, so as to restore the physiological characteristic compressed data and abnormal mark data;
[0166] The embodiment of the present invention transmits physiological characteristic data to a flexible circuit board, and uses the flexible circuit board as a transmission medium to transmit the physiological characteristic data from a sensor or an acquisition device to a signal processing module. Integrity check and decoding are performed, and integrity check is performed on the transmitted data in the signal processing module to ensure the integrity and accuracy of the data; then decoding is performed to restore the compressed physiological characteristic data and abnormal mark data.
[0167] Step S262: performing Gaussian noise filtering on the physiological characteristic compressed data through a particle filter, thereby obtaining preliminary filtered physiological data;
[0168] The embodiment of the present invention uses a particle filter for filtering, inputs the physiological characteristic compressed data into the particle filter, and filters the data through the algorithm and model of the particle filter to filter out Gaussian noise and obtain preliminary filtered physiological data.
[0169] Step S263: performing bandpass filtering on the preliminary filtered physiological data to retain signal components within a preset frequency range, thereby obtaining filtered physiological data;
[0170] The embodiment of the present invention sets the parameters of the bandpass filter, and sets the cutoff frequency or frequency bandwidth of the bandpass filter according to the frequency range of the signal components to be retained. The preliminary filtered physiological data is bandpass filtered, and the preliminary filtered physiological data is input into the bandpass filter, and the filter is used to process the data, only the signal components within the preset frequency range are retained, and the interference components of other frequencies are filtered out to obtain the filtered physiological data.
[0171] Step S264: repairing the abnormal value of the filtered physiological data according to the abnormal marking data, thereby obtaining physiological abnormality repair data;
[0172] The embodiment of the present invention identifies abnormal values based on abnormal marking data, and identifies data points marked as abnormal in filtered physiological data based on the abnormal marking data; repairs the abnormal values, and uses appropriate repair methods, such as interpolation, replacement, etc., to repair the data points marked as abnormal to obtain physiological abnormality repair data.
[0173] Step S265: extracting features of the physiological abnormality repair data to obtain preliminary physiological parameter data, wherein the features include average heart rate, maximum heart rate, minimum heart rate, heart rate variability index, average blood oxygen saturation and body temperature change trend;
[0174] The embodiment of the present invention selects a feature extraction method according to the needs, and selects an appropriate method to extract the required physiological parameter features according to specific needs, such as average heart rate, maximum heart rate, minimum heart rate, heart rate variability index, average blood oxygen saturation and temperature change trend, etc. The corresponding algorithm is applied to extract features, and the physiological abnormality repair data is processed and calculated according to the selected feature extraction method to obtain preliminary physiological parameter data.
[0175] Step S265: abnormally marking the preliminary physiological parameter data according to the abnormal marking data, identifying and marking abnormal events, thereby obtaining marked physiological parameter data.
[0176] The embodiment of the present invention identifies abnormal events based on abnormal marking data, and identifies events or data points marked as abnormal in the preliminary physiological parameter data based on the abnormal marking data. The abnormal events are marked, and the abnormal events or data points are marked in the preliminary physiological parameter data for further analysis and processing to obtain marked physiological parameter data.
[0177] The present invention can effectively transmit physiological characteristic data through the operation of flexible circuit board transmission and signal processing module, and perform integrity check and decoding to ensure the integrity and accuracy of the data. The restored physiological characteristic compressed data and abnormal marking data provide a basis for subsequent data processing and analysis. Through Gaussian noise filtering, the noise components in the physiological characteristic compressed data can be effectively removed, the quality and signal-to-noise ratio of the data can be improved, and the physiological data obtained by preliminary filtering provide a more reliable basis for subsequent signal processing and analysis. Through bandpass filtering, the physiological signal components within the preset frequency range can be selectively retained, and the interference signals in other frequency ranges can be removed. The physiological data obtained after filtering is more focused on the frequency range of interest, which is helpful for subsequent feature extraction and analysis. By repairing the outliers of the filtered physiological data according to the abnormal marking data, the outliers or noise points in the physiological data can be removed, and the accuracy and reliability of the data can be improved. The physiological abnormal repair data provides a more reliable data basis for subsequent feature extraction and parameter calculation. By extracting features from the physiological abnormal repair data, some important physiological parameters can be extracted from the data. These parameters include average heart rate, maximum heart rate, minimum heart rate, heart rate variability index, average blood oxygen saturation, and temperature change trend, etc., providing a quantitative description and evaluation of the individual's physiological state. By marking the initial physiological parameter data for abnormalities based on the abnormal marking data, abnormal events in the physiological parameters can be identified and annotated. These abnormal events include abnormal heart rate, abnormal blood oxygen saturation, abnormal body temperature, etc. The marked physiological parameter data facilitates further abnormality detection and health status monitoring.
[0178] Preferably, step S3 comprises the following steps:
[0179] Step S31: performing time alignment and normalization processing on the physiological parameter data to obtain standardized physiological parameter data;
[0180] The embodiment of the present invention time-aligns physiological parameter data from different sources or with different sampling rates so that they correspond at the same time point; normalizes the time-aligned physiological parameter data and maps the data to a uniform numerical range, for example, scaling the data to between 0 and 1.
[0181] Step S32: using an acceleration sensor to collect user motion data, and pre-processing to obtain motion data;
[0182] The embodiment of the present invention uses an acceleration sensor to collect the user's motion data. The acceleration sensor is installed on the user or related equipment, and the user's motion data is obtained by measuring acceleration; the collected motion data is preprocessed, including noise removal, filtering, downsampling and other operations to improve data quality and accuracy.
[0183] Step S33: Time-synchronizing the standardized physiological parameter data with the motion data to obtain multi-modal time series data;
[0184] The embodiment of the present invention time-aligns the standardized physiological parameter data and motion data to ensure that they correspond at the same time point; performs time synchronization operation to align the physiological parameter data and motion data according to the same time step to obtain multimodal time series data, in which each time point contains relevant information of the physiological parameters and motion data.
[0185] Step S34: using a Kalman filter algorithm to eliminate motion interference in the multimodal time series data, thereby obtaining preliminary purified physiological data;
[0186] The embodiment of the present invention uses a Kalman filter algorithm to process multimodal time series data. The Kalman filter algorithm is a recursive filtering algorithm that can be used to estimate the system state and eliminate noise. In the Kalman filter, motion data is regarded as noise, and motion interference in the multimodal time series data is eliminated by establishing a dynamic model and a measurement model to obtain preliminary purified physiological data.
[0187] Step S35: Using independent component analysis technology to separate physiological signals and residual noise from the initially purified physiological data, thereby obtaining purified physiological data
[0188] The circuit of the present invention uses independent component analysis (ICA) technology to process the initially purified physiological data. ICA is a commonly used signal processing method that can separate mixed signals into independent components. In independent component analysis, the initially purified physiological data is regarded as a mixed signal, and the physiological signal and residual noise are separated by estimating the mixing matrix and the separation matrix to obtain purified physiological data.
[0189] Step S36: Send the purified physiological data to the cloud for in-depth analysis and long-term trend tracking, thereby obtaining in-depth analysis data.
[0190] The embodiment of the present invention sends the purified physiological data to a cloud computing platform to ensure the secure transmission and storage of the data, performs in-depth analysis in the cloud, and uses various algorithms and models to further process and analyze the purified physiological data, such as machine learning, deep learning, statistical analysis, etc.; performs long-term trend tracking, performs time series analysis on the purified physiological data, and explores patterns, trends, and correlations in the data to obtain deeper insights and predictions.
[0191] The present invention can unify the physiological parameter data of different acquisition time points to the same time scale and perform data standardization through time alignment and normalization processing; standardized physiological parameter data helps to eliminate time differences and dimensional differences, provide a consistent data benchmark, and facilitate subsequent data analysis and comparison. By collecting acceleration sensor data, the user's motion data can be obtained, and the preprocessing operation can include filtering, noise removal, etc. to improve the quality and accuracy of the motion data; the motion data provides information about the user's activity level, posture changes, etc., and provides a basis for subsequent multimodal data fusion. By synchronizing the standardized physiological parameter data and motion data, the time axes of the two data sources can be aligned to ensure that they correspond at the same time point; the multimodal time series data obtained in this way combines physiological parameters and motion information, provides a more comprehensive data perspective, and helps to reveal the correlation and influence between physiology and motion. The Kalman filter algorithm can eliminate the interference of motion on physiological parameters by modeling and estimating the motion data in the multimodal time series data; by removing motion noise, preliminary purified physiological data is obtained, and the accuracy and reliability of physiological data are improved. Independent component analysis technology can decompose the initially purified physiological data into independent physiological signals and residual noise components. By separating physiological signals and noise, the characteristics and change trends of physiological signals can be further extracted to obtain cleaner and more interpretable purified physiological data. By sending the purified physiological data to the cloud for in-depth analysis, cloud computing power and algorithm models can be used to perform more complex and comprehensive analysis of the data, including pattern recognition, anomaly detection, and health status assessment of physiological data; at the same time, long-term trend tracking can help monitor changes and trends in an individual's physiological state, providing more in-depth health management and early warning functions.
[0192] Preferably, step S36 includes the following steps:
[0193] Step S361: Send the purified physiological data to the cloud, and perform identity authentication and data integrity check on the cloud, so as to obtain physiological characteristic cloud data;
[0194] The embodiment of the present invention transmits the purified physiological data to the cloud computing platform via the network to ensure the secure transmission and storage of the data; performs identity authentication in the cloud to verify the identity of the user to ensure the privacy and security of the data; and performs a data integrity check to ensure that the received data is complete and correct.
[0195] Step S362: decompressing the physiological characteristic cloud data and merging it with the pre-acquired user history data in time series to form long-term physiological characteristic data;
[0196] The embodiment of the present invention decompresses the received physiological characteristic cloud data, restores it to the original format of data, and pre-acquires the user's historical data, including past physiological characteristic data; merges the decompressed cloud data with the pre-acquired user historical data in a time series to ensure that the data is organized in chronological order to form long-term physiological characteristic data.
[0197] Step S363: using an autoregressive integrated moving average model algorithm to analyze the long-term trend and periodic pattern of the physiological indicators of the long-term physiological characteristic data, thereby obtaining trend analysis result data;
[0198] The embodiment of the present invention uses an autoregressive integrated moving average model algorithm to process long-term physiological characteristic data, and the model can analyze the long-term trend and periodic pattern of time series data. In the model, autoregressive modeling, integration and moving average operations are performed on the long-term physiological characteristic data to obtain the long-term trend analysis result data of the physiological index.
[0199] Step S364: using a gradient boosting tree to perform feature importance analysis on the long-term physiological feature data, and identifying the key indicators that have the greatest impact on the health status, thereby obtaining feature importance ranking data;
[0200] The embodiment of the present invention uses a gradient boosting tree algorithm to process long-term physiological feature data. The gradient boosting tree is an integrated learning algorithm that can be used for feature importance analysis and predictive modeling. In the algorithm, a gradient boosting tree model is constructed using long-term physiological feature data. By analyzing the importance scores of each feature in the model, the key indicators that have the greatest impact on health status are identified. Feature importance ranking data is obtained, and the indicators are ranked according to the feature importance scores to guide subsequent in-depth analysis and decision-making.
[0201] Step S365: Merge the trend analysis result data and the feature importance ranking data into the depth analysis data.
[0202] The embodiment of the present invention combines the trend analysis result data and the feature importance ranking data to form the deep analysis data. The merging method can be selected according to specific needs, for example, the trend analysis result data and the feature importance ranking data are spliced as different feature columns, or they are fused together in some way to form a comprehensive deep analysis data set.
[0203] The present invention sends the purified physiological data to the cloud for identity authentication and data integrity check to ensure the security and reliability of the data. The identity authentication in the cloud can verify the identity of the user and ensure that only authorized users can access sensitive physiological data. The data integrity check can identify and exclude errors or damage that may be introduced during data transmission or storage to ensure the integrity of cloud data. The cloud data is decompressed and decoded to restore the compressed data and restore its original format, and then the decompressed physiological characteristic data is merged with the pre-acquired user historical data to form longer-term physiological characteristic data; this can increase the time span of the data, provide more comprehensive historical information, and help analyze the long-term trend of health status. The autoregressive integrated moving average model (ARIMA) algorithm is a commonly used time series analysis method that can be used to explore the long-term trend and periodic pattern of physiological indicators. By applying the ARIMA model, the long-term change trend, seasonal changes and other periodic patterns of physiological indicators can be identified, thereby obtaining the result data of trend analysis, which helps to understand the changes of the user's physiological state over time and discover potential periodic patterns or laws. Gradient Boosting Trees is a machine learning algorithm that can be used to analyze feature importance. By applying the gradient boosting tree algorithm, long-term physiological feature data can be ranked by feature importance, and which physiological indicators have the greatest impact on health status can be identified, which helps to determine key indicators and provides ranking and importance information of physiological features that have the greatest impact on health status. The result data of trend analysis and feature importance ranking data are merged to form deep analysis data, which integrates the long-term trends, periodic patterns of physiological indicators, and importance ranking information of key indicators; such comprehensive data can provide a more in-depth and comprehensive data analysis basis for applications such as health management, disease prediction, and individual monitoring, helping users and medical professionals better understand physiological data, assess health status, and make corresponding decisions or interventions.
[0204] Preferably, step S4 comprises the following steps:
[0205] Step S41: performing correlation analysis on the deep analysis data and the pre-acquired historical health data, thereby constructing a health status model of the user;
[0206] The embodiment of the present invention associates the deep analysis data with the pre-acquired historical health data, and can use machine learning algorithms, statistical methods or other related analysis techniques to integrate and model the data. A user's health status model is constructed, which combines the deep analysis data with the historical health data to reflect the user's current health status.
[0207] Step S42: using the knowledge graph-based reasoning engine to infer the user's health status on the health status model to generate personalized health report data;
[0208] The reasoning engine based on the knowledge graph in the embodiment of the present invention uses the user's health status model as input, and infers the user's health status according to predefined knowledge and rules; the reasoning engine can use the medical knowledge, disease association and other information in the knowledge graph, combined with the user's health status model for reasoning and analysis; generate personalized health report data, which includes information such as evaluation, suggestions and predictions of the user's health status.
[0209] Step S43: The personalized health report data is transmitted back to the headset for decoding, and the personalized health report data is converted into text using natural language processing technology to obtain user health text data;
[0210] The embodiment of the present invention transmits the personalized health report data back to the headset to ensure that the user can receive the report data, decode it in the headset, and convert the report data from binary form into readable text. Use natural language processing technology to convert the personalized health report data into text; this may involve NLP technologies such as text cleaning, syntactic analysis, and semantic understanding to extract key information from the report data.
[0211] Step S44: Convert the user's health text data into voice prompts through a speech synthesis module, thereby obtaining health prompt data.
[0212] The embodiment of the present invention uses a speech synthesis module to convert user health text data into voice prompts so as to convey health information to the user through headphones; the speech synthesis module can use text-to-speech (TTS) technology to convert text into natural speech with high readability and naturalness. According to user needs and the content of the health report, corresponding voice prompts are generated to provide customized health advice, warnings or other related information.
[0213] The present invention can establish a user's health status model by associating the deep analysis data with the pre-acquired historical health data. This model can comprehensively evaluate the user's health status based on multiple data sources and multiple factors; by analyzing and mining the association between the deep analysis data and the historical health data, the user's health status can be more accurately understood, including chronic disease risks, changes in living habits and physiological indicators. The reasoning engine based on the knowledge graph can use the data and knowledge in the health status model to perform reasoning and inference, thereby generating personalized health report data. The reasoning engine can analyze the user's health data based on the user's health status model and related knowledge, and provide targeted suggestions, predictions and explanations. Such personalized health report data can help users better understand their health status, understand possible risks and suggestions, and promote active health management and behavior changes. The personalized health report data is transmitted back to the headset and decoded, and then the data is converted into user health text data using natural language processing technology. Such text data can provide a user-readable health report so that the user can intuitively understand and obtain health information. By conveying the personalized health report to the user in text form, a more direct, easy-to-understand and reference health guidance can be provided to support the user's health management and decision-making. Through the speech synthesis module, the user's health text data is converted into voice prompts to provide users with personalized health prompt data. Such health prompt data can be directly conveyed to users by voice, allowing users to obtain health information without having to look at the screen. Voice prompts can provide health advice, warnings or reminders in verbal form, allowing users to more easily receive and apply health information and promote positive health behaviors and decisions.
[0214] Therefore, the embodiments should be regarded as illustrative and non-restrictive from all points, and the scope of the present invention is limited by the appended claims rather than the above description, and it is therefore intended that all changes falling within the meaning and range of equivalent elements of the application documents are included in the present invention.
[0215] The above description is only a specific embodiment of the present invention, so that those skilled in the art can understand or implement the present invention. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but should conform to the widest scope consistent with the principles and novel features invented herein.
Claims
1. An ear health monitoring method based on earphones, It is characterized in that The following steps are involved: Step S1: collecting raw photoelectric signal data through a photoplethysmography sensor integrated in the headset, wherein the photoplethysmography sensor includes a plurality of micro LEDs with different wavelengths and a high-sensitivity micro photodiode; Perform signal enhancement on the original photoelectric signal data to obtain high-quality photoelectric signal data; Step S2: Performing physiological characteristic analysis on the high-quality photoelectric signal data to obtain physiological characteristic data including heart rate, heart rate variability, blood oxygen saturation, body temperature and blood pressure estimation; The physiological characteristic data is transmitted to the signal processing module through the flexible circuit board, and preliminary filtering and feature extraction are performed to obtain the physiological parameter data; Step S3: performing multimodal fusion analysis based on the physiological parameter data, and eliminating motion interference based on the pre-acquired motion data to obtain purified physiological data; Send purified physiological data to the cloud for in-depth analysis and long-term trend tracking to obtain in-depth analysis data; Step S4: Evaluate health status based on the in-depth analysis data and generate personalized health report data; The personalized health report data is transmitted back to the headset, and the user's current health status is displayed through voice to obtain health reminder data.
2. The earphone-based ear health monitoring method according to claim 1, It is characterized in that Step S1 includes the following steps: Step S11: Initializing the photoplethysmography sensor in the headset to obtain sensor status data, wherein the initialization includes activating micro LEDs of different wavelengths and a high-sensitivity photodiode; Step S12: performing self-calibration on the initialized photoplethysmography sensor according to the sensor state data, thereby obtaining calibration parameter data; Step S13: using the micro LED to emit light of a preset wavelength, and receiving reflected light through the micro photodiode, thereby obtaining original photoelectric signal data; Step S14: amplifying and filtering the original photoelectric signal data according to the calibration parameter data, so as to obtain preliminary photoelectric signal data; Step S15: performing wavelength-based separation on the preliminary photoelectric signal data to obtain multi-wavelength photoelectric signal data; Step S16: Calculating the signal-to-noise ratio and signal quality index of the multi-wavelength photoelectric signal data to obtain signal quality evaluation data; Step S17: Perform quality judgment on the signal quality assessment data according to a preset quality threshold. If the quality is lower than the preset threshold, re-collection is triggered to obtain high-quality photoelectric signal data.
3. The ear health monitoring method based on earphone according to claim 2, It is characterized in that Step S15 includes the following steps: Step S151: performing multi-scale decomposition on the preliminary photoelectric signal data using discrete wavelet transform, thereby obtaining signal component data of different frequency bands; Step S152: performing independent high-frequency noise removal processing on the signal component of each wavelength according to the signal component data by using a soft threshold method, thereby obtaining preliminarily denoised multi-wavelength signal data; Step S153: using independent component analysis technology to separate the signal components corresponding to different physiological processes on the multi-wavelength signal data, thereby obtaining physiological signal component data; Step S154: Perform short-time Fourier transform on the physiological signal component data to obtain multi-wavelength photoelectric signal data.
4. The earphone-based ear health monitoring method according to claim 3, It is characterized in that Step S2 includes the following steps: Step S21: segmenting the high-quality photoelectric signal data into time windows, and dynamically adjusting the window size using an adaptive window algorithm, thereby obtaining photoelectric signal segmented data; Step S22: performing physiological characteristic analysis on the high-quality photoelectric signal segmentation data to obtain physiological characteristic data including heart rate, heart rate variability, blood oxygen saturation, body temperature and blood pressure estimation; Step S23: performing abnormality detection and preliminary classification on the physiological characteristic data by machine learning technology, thereby obtaining abnormal labeling data; Step S24: performing lossless compression encoding on the physiological characteristic data to obtain physiological characteristic compressed data; Step S25: Packing the physiological characteristic compressed data and the abnormal mark data, and adding a timestamp and a user ID, thereby obtaining a physiological characteristic data packet; Step S26: The physiological characteristic data packet is transmitted to the signal processing module through the flexible printed circuit board, and preliminary filtering and feature extraction are performed to obtain physiological parameter data.
5. The earphone-based ear health monitoring method according to claim 4, It is characterized in that Step S22 includes the following steps: Step S221: performing peak detection and heart rate calculation on the photoelectric signal segmented data, and removing outliers based on median filtering, so as to obtain preliminary heart rate data; Step S222: Calculating the continuous heartbeat intervals according to the preliminary heart rate data, and performing time domain and frequency domain analysis to obtain heart rate variability characteristic data; Step S223: performing ratio calculation based on different wavelengths according to the heart rate variability characteristic data, thereby obtaining blood oxygen saturation data; Step S224: analyzing the infrared light signal intensity variation of the photoelectric signal segmentation data, and estimating the body temperature according to a pre-established temperature-signal intensity calibration curve, thereby obtaining body temperature estimation data; Step S225: performing pulse wave propagation time analysis on the photoelectric signal segment data to obtain blood pressure estimation data; Step S226: combining the preliminary heart rate data, heart rate variability characteristic data, blood oxygen saturation data, body temperature estimation data and blood pressure estimation data into physiological characteristic data.
6. The earphone-based ear health monitoring method according to claim 5, It is characterized in that Step S226 includes the following steps: Step S2261: Time-aligning the preliminary heart rate data, the heart rate variability characteristic data, the blood oxygen saturation data, the body temperature estimation data, and the blood pressure estimation data, thereby obtaining time-aligned time data; Step S2262: normalizing the time-aligned physiological data to obtain normalized physiological characteristic data; Step S2263: performing credibility scoring on the normalized physiological characteristic data and marking low-credibility data, thereby obtaining physiological data quality score data; Step S2264: Filter the normalized physiological characteristic data according to the physiological data quality score data to obtain the physiological characteristic data.
7. The earphone-based ear health monitoring method according to claim 6, It is characterized in that Step S26 includes the following steps: Step S261: transmitting the physiological characteristic data to the signal processing module through the flexible circuit board, and performing integrity check and decoding, so as to restore the physiological characteristic compressed data and abnormal mark data; Step S262: performing Gaussian noise filtering on the physiological characteristic compressed data through a particle filter, thereby obtaining preliminary filtered physiological data; Step S263: performing bandpass filtering on the preliminary filtered physiological data to retain signal components within a preset frequency range, thereby obtaining filtered physiological data; Step S264: repairing the abnormal value of the filtered physiological data according to the abnormal marking data, thereby obtaining physiological abnormality repair data; Step S265: extracting features of the physiological abnormality repair data to obtain preliminary physiological parameter data, wherein the features include average heart rate, maximum heart rate, minimum heart rate, heart rate variability index, average blood oxygen saturation and body temperature change trend; Step S265: abnormally marking the preliminary physiological parameter data according to the abnormal marking data, identifying and marking abnormal events, thereby obtaining marked physiological parameter data.
8. The earphone-based ear health monitoring method according to claim 7, It is characterized in that Step S3 includes the following steps: Step S31: performing time alignment and normalization processing on the physiological parameter data to obtain standardized physiological parameter data; Step S32: using an acceleration sensor to collect user motion data, and pre-processing to obtain motion data; Step S33: Time-synchronizing the standardized physiological parameter data with the motion data to obtain multi-modal time series data; Step S34: using a Kalman filter algorithm to eliminate motion interference in the multimodal time series data, thereby obtaining preliminary purified physiological data; Step S35: Using independent component analysis technology to separate physiological signals and residual noise from the initially purified physiological data, thereby obtaining purified physiological data Step S36: Send the purified physiological data to the cloud for in-depth analysis and long-term trend tracking, thereby obtaining in-depth analysis data.
9. The earphone-based ear health monitoring method according to claim 8, It is characterized in that Step S36 includes the following steps: Step S361: Send the purified physiological data to the cloud, and perform identity authentication and data integrity check on the cloud, so as to obtain physiological characteristic cloud data; Step S362: decompressing the physiological characteristic cloud data and merging it with the pre-acquired user history data in time series to form long-term physiological characteristic data; Step S363: using an autoregressive integrated moving average model algorithm to analyze the long-term trend and periodic pattern of the physiological indicators of the long-term physiological characteristic data, thereby obtaining trend analysis result data; Step S364: using a gradient boosting tree to perform feature importance analysis on the long-term physiological feature data, and identifying the key indicators that have the greatest impact on the health status, thereby obtaining feature importance ranking data; Step S365: Merge the trend analysis result data and the feature importance ranking data into the depth analysis data.
10. The earphone-based ear health monitoring method according to claim 9, It is characterized in that Step S4 includes the following steps: Step S41: performing correlation analysis on the deep analysis data and the pre-acquired historical health data, thereby constructing a health status model of the user; Step S42: using a knowledge graph-based reasoning engine to infer the user's health status on the health status model to generate personalized health report data; Step S43: The personalized health report data is transmitted back to the headset for decoding, and the personalized health report data is converted into text using natural language processing technology to obtain user health text data; Step S44: Convert the user's health text data into voice prompts through a speech synthesis module, thereby obtaining health prompt data.
Citation Information
Cited By
Physiological optical signal credibility evaluation method, device, storage medium and system
CN119770004A
Cerebrovascular disease analysis device based on AI
CN120267256A
