Intelligent multifunctional blood pressure and pulse rate monitoring method, system, equipment and medium

Abstract

The invention relates to an intelligent multifunctional blood pressure and pulse rate monitoring method, system and device and a medium, and belongs to the technical field of health data analysis. The monitoring method comprises the steps that user identity recognition data is obtained and analyzed, and a user identifier is generated; loading historical measurement records of a user, acquiring real-time attitude sensor data, and generating a personalized measurement parameter configuration set; selecting a target blood pressure measuring method; controlling a pressurizing device, and synchronously acquiring a cuff original pressure signal and a pulse wave original signal; extracting pressure oscillation characteristics and calculating an original pulse rate value; performing compensation calculation on the pressure oscillation characteristics to obtain calibrated pressure characteristics; calculating a calibrated systolic pressure value and a calibrated diastolic pressure value based on the calibrated pressure characteristic; and inputting the data into a time sequence analysis model, outputting a user health report, encrypting and storing the user health report to a local database, and synchronizing the user health report to a cloud server through a communication module. The accuracy and continuity of blood pressure measurement can be improved.

Description

Technical Field

[0001] This application relates to the field of health data analysis technology, and in particular to an intelligent multifunctional blood pressure and pulse rate monitoring method, system, device and medium. Background Technology

[0003] Currently, most electronic blood pressure monitors remain at the level of "single measurement, isolated output," lacking effective user identification and continuous accumulation of historical data. This results in each measurement operating in a "memoryless" mode, unable to adaptively optimize based on individual physiological characteristics and usage habits. Especially when different users share a single device, data contamination is highly likely, making it difficult to trace measurement results back to a specific individual, severely impacting the accuracy and reliability of long-term health tracking. Furthermore, the inability to access past measurement records for parameter presets or trend analysis necessitates repeating the same standardized process for each measurement, ignoring significant heterogeneity among individuals in areas such as vascular elasticity, differences in blood pressure between arms, and postural responses, leading to decreased measurement accuracy and unavoidable result bias. Summary of the Invention

[0004] To improve the accuracy and continuity of blood pressure measurement, this application provides an intelligent multifunctional blood pressure pulse rate monitoring method, system, device, and medium.

[0005] Firstly, this application provides an intelligent multifunctional blood pressure and pulse rate monitoring method, which adopts the following technical solution: A smart, multifunctional blood pressure and pulse rate monitoring method, the monitoring method comprising: Acquire and parse user identification data to generate user identifiers; Load the user's historical measurement records associated with the user identifier, and simultaneously acquire real-time attitude sensor data; Based on the user's historical measurement records and posture sensor data, a personalized set of measurement parameter configurations is generated, including arm side parameters, body position parameters, and cuff calibration pressure thresholds. Select the target blood pressure measurement method based on the arm side parameters and body position parameters; The pressurization device is controlled based on the cuff calibration pressure threshold, and the original cuff pressure signal and pulse wave signal are collected simultaneously. The original pressure signal of the cuff is filtered to extract pressure oscillation characteristics, and the original pulse wave signal is analyzed in the frequency domain to calculate the original pulse rate value. Call the body position compensation coefficient library corresponding to the body position parameters to perform compensation calculation on the pressure oscillation characteristics and obtain the calibrated pressure characteristics; Call the arm-side difference calibration model corresponding to the arm-side parameters, and calculate the calibrated systolic blood pressure value and calibrated diastolic blood pressure value based on the calibration pressure characteristics; The calibrated systolic blood pressure value, calibrated diastolic blood pressure value, raw pulse rate value, and the user's historical measurement records are input into the time series analysis model to output a user health report; the user health report includes blood pressure level judgment, health trend prediction, and risk assessment markers; The calibrated systolic blood pressure value, calibrated diastolic blood pressure value, raw pulse rate value, and user health report are encrypted and stored in a local database, and synchronized to a cloud server via a communication module.

[0006] By adopting the above technical solution, a complete technological chain has been constructed, from individual identity authentication to precise physiological measurement and intelligent health insights. This achieves personalization, intelligence, and continuity in the measurement process, not only improving the accuracy and repeatability of individual measurements but also endowing the device with early risk warning and health management capabilities through long-term data accumulation and trend modeling. This technical solution embodies the cutting-edge development direction of artificial intelligence-enabled health management and has significant clinical application value and industrialization prospects.

[0007] Secondly, this application provides an intelligent multifunctional blood pressure and pulse rate monitoring system, which adopts the following technical solution: A smart multifunctional blood pressure and pulse rate monitoring system, the monitoring system comprising: The user identification module is used to acquire and parse user identification data to generate user identifiers. The data loading module is used to load the user's historical measurement records associated with the user identifier, and at the same time acquire real-time attitude sensor data; The parameter configuration module is used to generate a personalized set of measurement parameter configurations based on the user's historical measurement records and posture sensor data, including arm side parameters, body position parameters and cuff calibration pressure thresholds. The measurement method selection module selects the target blood pressure measurement method based on the arm side parameters and body position parameters. The signal acquisition module is used to control the pressurization device based on the cuff calibration pressure threshold and to simultaneously acquire the original cuff pressure signal and the original pulse wave signal. The signal processing module is used to filter the original pressure signal of the cuff, extract the pressure oscillation characteristics, and perform frequency domain analysis on the original pulse wave signal to calculate the original pulse rate value. The body position compensation module is used to call the body position compensation coefficient library corresponding to the body position parameters, perform compensation calculations on the pressure oscillation characteristics, and obtain the calibrated pressure characteristics. The arm side calibration module is used to call the arm side difference calibration model corresponding to the arm side parameters, and calculate the calibration systolic blood pressure value and calibration diastolic blood pressure value based on the calibration pressure characteristics. The health analysis module is used to input the calibrated systolic blood pressure value, calibrated diastolic blood pressure value, raw pulse rate value, and the user's historical measurement records into a time-series analysis model, and output a user health report; wherein, the user health report includes blood pressure level judgment, health trend prediction, and risk assessment markers; The data storage module is used to encrypt and store the calibrated systolic blood pressure value, calibrated diastolic blood pressure value, raw pulse rate value and user health report to the local database, and synchronize them to the cloud server through the communication module.

[0008] Thirdly, this application provides a computer device, which adopts the following technical solution: A computer device includes a memory, a processor, and a computer program stored in the memory, the processor executing the computer program to perform the steps of the method as described in the first aspect.

[0009] Fourthly, this application provides a computer-readable storage medium, which adopts the following technical solution: A computer-readable storage medium storing a computer program that can be loaded by a processor and executed as in any of the methods in the first aspect. Attached Figure Description

[0010] Figure 1 This is a schematic diagram of the first process of an intelligent multifunctional blood pressure and pulse rate monitoring method according to one embodiment of this application.

[0011] Figure 2 This is a schematic diagram of the second process of an intelligent multifunctional blood pressure and pulse rate monitoring method according to one embodiment of this application.

[0012] Figure 3 This is a schematic diagram of the third process of an intelligent multifunctional blood pressure and pulse rate monitoring method according to one embodiment of this application.

[0013] Figure 4 This is a schematic diagram of the fourth process of an intelligent multifunctional blood pressure and pulse rate monitoring method according to one embodiment of this application.

[0014] Figure 5 This is a schematic diagram of the fifth process of an intelligent multifunctional blood pressure and pulse rate monitoring method according to one embodiment of this application.

[0015] Figure 6 This is a schematic diagram of the sixth process of an intelligent multifunctional blood pressure and pulse rate monitoring method according to one embodiment of this application.

[0016] Figure 7This is a schematic diagram of the seventh process of an intelligent multifunctional blood pressure and pulse rate monitoring method according to one embodiment of this application. Detailed Implementation

[0017] To make the purpose, technical solution, and advantages of this application clearer, the following description is provided in conjunction with the appendix. Figures 1-7 The present application will be further described in detail below with reference to embodiments. It should be understood that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the scope of the application.

[0018] Currently, traditional blood pressure monitors generally employ fixed measurement strategies and uniform compensation standards, failing to fully consider changes in body posture during measurement and their impact on blood pressure readings. Clinical studies have shown that changes in body position significantly affect venous return and arterial pressure distribution. For example, when changing from a sitting to a lying position, systolic blood pressure can decrease by 5–10 mmHg. However, most existing devices rely solely on users manually selecting the body position mode. If the user does not set it correctly or if a posture shift occurs during measurement, the system cannot actively detect and dynamically correct it, leading to deviations from standard measurement conditions and unavoidable result bias. Furthermore, due to anatomical differences between the left and right upper arms (e.g., the right subclavian artery originates from the brachiocephalic trunk, while the left arm has a longer path), there is often a 5–10 mmHg difference in blood pressure between the two arms. Some patients even experience greater differences due to subclavian artery stenosis. However, traditional devices lack effective calibration mechanisms for this, often defaulting to the right arm as the standard measurement side. If the user is accustomed to using the left arm, this may introduce systematic errors.

[0019] Furthermore, in terms of health assessment, most devices only provide a comparison of a single measurement with a static reference range, lacking trend modeling and risk prediction functions based on long-term data. This fails to provide users with forward-looking health insights or valuable data for population health monitoring for medical institutions. These issues collectively limit the in-depth application of blood pressure monitoring devices in chronic disease management, making it difficult for them to truly play a role in early warning and personalized intervention.

[0020] Therefore, how to build a comprehensive monitoring system that integrates accurate identification, personalized measurement configuration, multidimensional physiological compensation, intelligent health analysis and safe data management has become a key challenge to improve the accuracy, continuity and clinical applicability of blood pressure measurement.

[0021] Based on this, this application discloses an intelligent multifunctional blood pressure and pulse rate monitoring method.

[0022] Reference Figure 1 A smart, multifunctional blood pressure and pulse rate monitoring method, specifically including: Step S101: Obtain user identification data and parse it to generate user identifiers; User identification data includes at least one of the following: ID card image, social security card NFC signal, or wristband QR code; Specifically, this step uses at least one of the following as identification data: an image of the user's ID card, an NFC signal from their social security card, or a QR code on their wristband. This ensures multimodal compatibility and scenario adaptability for user authentication. Whether it's wristband scanning in a hospital clinical environment, social security card sensing in community health services, or uploading a photo of the user's ID card for home self-testing, the system can flexibly adapt to the identity authentication needs of different usage scenarios. This raw identity data is uniformly transformed into structured information through image recognition, radio frequency signal decoding, or QR code parsing. Furthermore, a unique user identifier (UID) is generated using a hash algorithm or encryption encoding method. This UID serves as the core index throughout the entire monitoring process, achieving precise binding and privacy anonymization of user data.

[0023] In addition, the system automatically creates encrypted user profiles, which not only contain the user's basic identity metadata, but also reserve space for storing dynamic information such as historical measurement records, parameter preferences, and health trend models.

[0024] Step S102: Load the user's historical measurement records associated with the user identifier, and simultaneously acquire real-time attitude sensor data; The system loads historical measurement records associated with the user identifier and simultaneously acquires real-time attitude sensor data from the built-in accelerometer and gyroscope. This process represents a crucial leap from "general measurement" to "personalized, precise measurement." The historical measurement records contain multi-dimensional data such as the user's past blood pressure, pulse rate, measurement time, body position, arm circumference information, cuff size selection, and signal quality feedback. This data constitutes a long-term profile of the user's physiological characteristics.

[0025] Step S103: Based on the user's historical measurement records and posture sensor data, generate a personalized set of measurement parameter configurations, including arm side parameters, body position parameters, and cuff calibration pressure thresholds. Specifically, the posture sensor captures the user's current body posture in real time, such as sitting, lying, standing, or tilting angles. The data is expressed in the form of three-dimensional spatial coordinates and angular velocity. After filtering and posture fusion algorithms (such as complementary filtering or Kalman filtering), the user's current actual body position can be accurately determined.

[0026] Subsequently, the system combines historical data with real-time posture information, using a rule engine or lightweight machine learning model (such as a decision tree or regression model) to dynamically generate a personalized set of measurement parameter configurations. Among these, the arm side parameter determines whether the measurement should be taken using the left or right arm. This is based not only on user preferences but also on statistical results of past bilateral blood pressure differences (clinically, it's common for right arm blood pressure to be slightly higher than left arm, especially in cases of vascular abnormalities such as subclavian artery stenosis). The body position parameter clearly indicates the body posture during the current measurement, as changes in body position significantly affect venous return and arterial pressure distribution; for example, diastolic blood pressure is typically 5–10 mmHg lower in the supine position than in the sitting position. The generation of the cuff calibration pressure threshold comprehensively considers the user's historical arm circumference data and the current body position. It matches the appropriate cuff type by searching a preset arm circumference-cuff size mapping table and adjusts the initial inflation pressure according to the body position to avoid discomfort due to excessive pressure or measurement failure due to excessively low pressure.

[0027] In addition, the system also has the ability to dynamically correct body position: if the sensor detects that the deviation between the actual body position and the preset measurement body position (such as standard sitting posture) exceeds the set threshold (such as trunk tilt greater than 15 degrees), the body position parameters will be automatically updated and the measurement process will be reconfigured, thereby ensuring the standardization of measurement conditions and improving data comparability and accuracy.

[0028] Step S104: Select the target blood pressure measurement method based on the arm side parameters and body position parameters; This step introduces a parameter-driven measurement strategy selection mechanism. For example, when the user is in a supine position and at risk of respiratory interference, the system may activate a low-noise pressurization mode and extend the measurement cycle; when the system detects that the user is an elderly hypertensive patient and historical data shows significant pulse wave conduction delay, it may adjust the pressure release rate to improve the accuracy of oscillatory wave capture.

[0029] Step S105: Control the pressurization device based on the cuff calibration pressure threshold, and simultaneously collect the original cuff pressure signal and the original pulse wave signal. After selecting the measurement method, the system uses the cuff calibration pressure threshold as a benchmark to control a micro-pump to apply graded pressure to the cuff. Simultaneously, pressure and pulse wave sensors are activated to acquire two types of raw signals: the raw cuff pressure signal reflects the change in internal cuff pressure over time, including minute pressure fluctuations caused by arterial pulsation (i.e., pressure oscillation waves); while the raw pulse wave signal acquires peripheral arterial pulsation information through photoplethysmography (PPG) or a piezoelectric sensor, used for independent heart rate estimation. The time synchronization of these two signals is crucial, requiring hardware-level triggering or timestamp alignment technology to ensure data frame synchronization, laying the foundation for subsequent joint analysis.

[0030] Step S106: Filter the original pressure signal of the cuff, extract the pressure oscillation characteristics, and perform frequency domain analysis on the original pulse wave signal to calculate the original pulse rate value. For the raw cuff pressure signal, the system first applies digital filtering, typically using a bandpass filter (e.g., 0.5–20 Hz) to remove interference components such as respiratory motion, muscle tremors, and electronic noise, retaining the pressure oscillation component related to the heart cycle. In the filtered signal, the system identifies a series of periodically occurring pressure fluctuation peaks and troughs. These oscillation amplitudes exhibit a typical "bell-shaped curve" where the cuff pressure first increases and then decreases as it decreases. The maximum value corresponds to the mean arterial pressure, and the raw systolic blood pressure (SBP_raw) and raw diastolic blood pressure (DBP_raw) can be preliminarily estimated using specific proportional rules (e.g., the fixed proportional method or the adaptive inflection point method).

[0031] Meanwhile, frequency domain analysis is performed on the raw pulse wave signal. Fast Fourier Transform (FFT) is commonly used to transform it from the time domain to the frequency domain, identifying the dominant frequency component with the most concentrated energy. This frequency represents the current pulse rate value (unit: beats / minute). Since PPG signals are significantly affected by factors such as blood perfusion and motion artifacts, the system typically combines time-domain peak detection with frequency-domain spectral peak tracking for cross-validation to improve the robustness of pulse rate calculation.

[0032] Step S107: Call the body position compensation coefficient library corresponding to the body position parameters to perform compensation calculation on the pressure oscillation characteristics and obtain the calibrated pressure characteristics; Specifically, the system calls upon a pre-built database of postural compensation coefficients. This database, trained on large-scale clinical research data, records the blood pressure changes of the same population in different positions (such as supine, lateral, sitting, and semi-recumbent), forming a set of standardized compensation vectors. For example, the database shows that when changing from a sitting to a supine position, the average systolic blood pressure drops by approximately 6 mmHg. Based on this, the system applies corresponding positive compensation to the currently extracted pressure oscillation characteristics. This compensation is not a simple addition or subtraction, but rather a weighted adjustment based on the user's individual historical response patterns, achieving a dual optimization of "group patterns + individual adaptation."

[0033] Step S108: Call the arm side difference calibration model corresponding to the arm side parameters, and calculate the calibration systolic blood pressure value and calibration diastolic blood pressure value based on the calibration pressure characteristics; The system utilizes an arm-side difference calibration model to further correct for blood pressure deviations caused by different arm measurements. This model, built upon extensive bi-arm synchronous measurement data, reveals the distribution pattern of the blood pressure difference (ΔP_arm) between the left and right arms and its correlation with factors such as age, vascular elasticity, and the presence of subclavian artery stenosis. The model expression can be formalized as a linear combination (e.g., SBP_cal = SBP_raw + α·ΔP_posture + β·ΔP_arm), essentially a composite function embedding a nonlinear mapping relationship, where the coefficients α and β can be dynamically learned and updated based on the user's historical data. Through this model, the system can output calibrated systolic blood pressure (SBP_cal) and calibrated diastolic blood pressure (DBP_cal) values ​​that more closely reflect real physiological conditions, significantly reducing the risk of misjudgment due to improper body position or arm side selection.

[0034] Step S109: Input the calibrated systolic blood pressure value, calibrated diastolic blood pressure value, raw pulse rate value, and user's historical measurement records into the time series analysis model, and output the user health report; wherein, the user health report includes blood pressure level judgment, health trend prediction, and risk assessment markers; Specifically, the calibrated blood pressure value and raw pulse rate value obtained from the current measurement, along with the historical blood pressure and pulse rate sequence for N consecutive days (N≥30) retrieved through the user identifier, are input into a deep time series analysis model, namely a Long Short-Term Memory (LSTM) network. As a special type of recurrent neural network, LSTM is good at capturing long-term dependencies in time series and can effectively identify the trend, periodicity, and sudden changes in blood pressure fluctuations.

[0035] In this embodiment, the model input includes not only the current value but also derived indicators such as historical extreme values, coefficient of variation, and circadian rhythm characteristics within the sliding window, enabling it to model complex physiological dynamics. The trained LSTM model can output two results: first, it predicts the blood pressure trend over the next K days (K=7), determining whether it may rise, fall, or remain stable; second, it generates risk level labels, such as "normal," "mild fluctuation," "persistently high," and "acute elevation warning," and automatically labels potential cardiovascular event risks by setting thresholds in conjunction with clinical guidelines.

[0036] In addition, the system can combine auxiliary indicators such as heart rate variability (HRV) trends and postural tolerance to generate a comprehensive health report that includes blood pressure level assessment, health trend prediction, and risk assessment markers. This report not only presents users with intuitive visual charts and textual suggestions but can also be accessed remotely by doctors, serving as an important basis for chronic disease management.

[0037] Step S110: Encrypt and store the calibrated systolic blood pressure value, calibrated diastolic blood pressure value, raw pulse rate value and user health report to the local database, and synchronize them to the cloud server through the communication module.

[0038] Specifically, all generated critical data is stored in a local database using high-strength encryption with the AES-256 algorithm. The key is generated by the device's unique hardware fingerprint, ensuring that even if the storage medium is illegally accessed, it cannot be decrypted. Simultaneously, data packets are uploaded to the cloud server via an SSL / TLS encrypted channel through communication modules (such as Wi-Fi, Bluetooth, or cellular networks), enabling cross-device data synchronization and long-term archiving. The cloud platform can further integrate multi-source information such as electronic health records (EHRs), medication records, and exercise and sleep data, supporting broader health data analysis and remote intervention.

[0039] The above implementation constructs a complete technology chain from individual identity authentication to precise physiological measurement and intelligent health insights, realizing the personalization, intelligence, and continuity of the measurement process. This not only improves the accuracy and repeatability of individual measurements but also endows the device with early risk warning and health management capabilities through long-term data accumulation and trend modeling. This technical solution embodies the cutting-edge development direction of artificial intelligence-enabled health management and has significant clinical application value and industrialization prospects.

[0040] Reference Figure 2 As one implementation of step S103, the step of generating a personalized measurement parameter configuration set based on the user's historical measurement records and posture sensor data, including arm side parameters, body position parameters, and cuff calibration pressure thresholds, includes: Step S201: Call the historical measurement records associated with the user identifier and extract historical arm side selection data, historical body position selection data, historical cuff pressure threshold and user arm circumference data. The system retrieves historical measurement records uniquely associated with each user identifier (UID) and extracts three key historical data points: historical arm side selection data, historical body position selection data, and historical cuff pressure thresholds. This process embodies the user-centric personalized design philosophy.

[0041] Specifically, historical arm side selection data reflects the user's preference for using their left or right arm in previous measurements. This preference is often closely related to lifestyle habits, limb function (such as the side recovering after surgery), and even potential differences in vascular anatomy. Historical body position selection data records the user's body posture during previous measurements, such as sitting, lying flat, or semi-recumbent. This information not only reveals the user's usage habits but also indirectly reflects their physical tolerance (e.g., heart failure patients find it difficult to maintain a sitting posture for extended periods). Furthermore, the historical cuff pressure threshold serves as a reference value for the initial inflation pressure used in previous successful measurements, incorporating a comprehensive response to the user's arm circumference, vascular elasticity, and the measurement environment.

[0042] Understandably, by analyzing this long-accumulated behavioral and physiological data, the system can establish a preliminary basis for parameter prediction, avoiding the need to configure from scratch for each measurement and significantly improving user experience and measurement efficiency.

[0043] Step S202: Acquire attitude sensor data in real time, including triaxial accelerometer signals and gyroscope angular velocity signals; The system acquires attitude data in real time from the device's built-in sensors, specifically triaxial accelerometer signals and gyroscope angular velocity signals. These two types of sensors constitute the core components of the inertial measurement unit (IMU), which is the key hardware support for realizing human posture recognition.

[0044] Specifically, a triaxial accelerometer is used to detect the linear acceleration of a device in space. Especially under static or quasi-static conditions, its output mainly reflects the projection components of the gravity vector on the three coordinate axes, which can be used to estimate the tilt angle of the device relative to the Earth coordinate system. A gyroscope, on the other hand, measures the rotational angular velocity of a device about three axes and is good at capturing attitude changes during rapid dynamic motion.

[0045] However, single sensors have limitations: accelerometers are susceptible to interference from motion acceleration, leading to distorted tilt angle calculations; gyroscopes, due to integration drift, accumulate errors over prolonged use. Therefore, both must work collaboratively, leveraging sensor fusion algorithms (such as complementary filtering or extended Kalman filtering) to achieve synergistic effects and consistently output high-precision attitude information even in dynamic environments. This real-time sensing capability allows the system to proactively identify the user's current physical state, rather than passively accepting user input, providing an objective basis for subsequent parameter corrections.

[0046] Step S203: Generate recommended body position parameters based on historical body position selection data, and calculate the real-time body position tilt angle based on triaxial accelerometer signals and gyroscope angular velocity signals. The system generates recommended posture parameters based on historical posture selection data. This means that the system predicts the ideal posture for the current measurement based on the user's most frequently used posture in the past, such as "standard sitting posture" or "supine position". However, this recommended value is only an initial assumption and not a final determination.

[0047] Meanwhile, the system uses accelerometer and gyroscope signals to calculate the current real-time body tilt angle. That is, by decomposing the acceleration signal into a gravity component, the pitch and roll angles of the device relative to the horizontal plane can be calculated. These two angles together describe the spatial orientation of the measurement part. To further improve accuracy, the system can also introduce the angular velocity integral results of the gyroscope to compensate and correct the attitude changes during the dynamic process, and suppress the instantaneous noise caused by brief limb tremors or breathing movements.

[0048] Step S204: If the deviation between the real-time body tilt angle and the recommended body position parameter exceeds a preset deviation threshold, then the body position type corresponding to the real-time body tilt angle is used as the final body position parameter; if it does not exceed the preset deviation threshold, then the recommended body position parameter is retained as the final body position parameter. When the deviation between the real-time body tilt angle and its recommended value exceeds a preset threshold (e.g., 15°), it indicates that the user's current actual body position has significantly deviated from the normal measurement posture, which may affect the accuracy of blood pressure readings (e.g., excessive forward tilt of the trunk leads to neck muscle tension, which in turn affects upper limb blood pressure). At this time, the system will automatically update the real-time detected body position type to the final body position parameters to ensure that the subsequent compensation model calls the body position label that truly reflects the current state.

[0049] Furthermore, if the deviation is within an acceptable range, the original recommended body position parameters are maintained, reflecting respect for user habits and the preservation of measurement continuity. This dual mechanism of "historical guidance + real-time verification" utilizes long-term behavioral patterns while also taking into account immediate physiological states, achieving an organic combination of static experience and dynamic perception.

[0050] Step S205: Generate recommended arm-side parameters based on historical arm-side selection data; wherein, if the user has no historical arm-side selection data, the recommended arm-side parameters are set to the default value. Based on historical arm selection data, the system generates recommended arm parameters, that is, to determine whether the left or right arm should be used first for this measurement.

[0051] Clinical studies have shown that blood pressure in the right arm is slightly higher than in the left arm in most people, with an average difference of about 5–10 mmHg. Some individuals may even have significant asymmetry due to conditions such as subclavian artery stenosis. Therefore, consistently using one arm can help improve the consistency of longitudinal data comparisons.

[0052] In this embodiment, if the user has a clear history of arm-side usage, the system will continue that preference; however, if it is a first-time user or historical data is missing, the arm-side parameters will be automatically set to preset default values ​​(such as the right arm) to ensure the startability of the measurement process. This design avoids the problem of process interruption due to lack of historical data, demonstrating the system's fault tolerance and robustness.

[0053] To further enhance operational reliability, the system can also integrate an infrared ranging sensor for non-contact detection of the user's upper limb position distribution, determining whether the actual arm placement matches the recommended arm side. If a mismatch is detected (e.g., the system recommends the right arm but the user places the device on their left arm), an audible and visual alarm is immediately triggered, and the parameter generation process is paused to prevent erroneous measurements, forming a dual safety loop of "software recommendation + hardware verification."

[0054] Step S206: Based on the final body position parameters and user arm circumference data, output the cuff calibration pressure threshold through the pre-configured cuff pressure calculation model. The system uses the determined final body position parameters and user arm circumference data to generate a cuff calibration pressure threshold by calling the cuff pressure calculation model. This step directly affects the safety and effectiveness of the measurement process. Excessive cuff pressure can cause user discomfort and even nerve compression, and may even lead to complete arterial occlusion, preventing the capture of effective pressure oscillations. Conversely, insufficient pressure will not adequately block blood flow, resulting in measurement failure or inaccurate results. Therefore, the initial inflation pressure must be scientifically set.

[0055] In this embodiment, the computational model comprehensively considers two core factors: first, the influence of body position, as the hemodynamic state differs under different body positions. For example, venous return increases and mean arterial pressure rises slightly when lying down, thus requiring an appropriate increase in initial pressure; second, arm circumference, as a larger arm circumference requires a higher external pressure to effectively compress the brachial artery. The model introduces a linear or nonlinear combination of body position coefficients (e.g., 1.0 for sitting, 1.05 for lying down, and 0.98 for standing due to gravity) with arm circumference diameter, and performs weighted calculations using clinically calibrated weighting coefficients to output a personalized calibration pressure threshold. These weighting coefficients have been validated through large-scale clinical trials, ensuring good generalization performance across different populations (e.g., children, adults, obese individuals), making the pressure setting both physiologically sound and individually adaptable.

[0056] Step S207: Integrate the final body position parameters, recommended arm side parameters, and cuff calibration pressure threshold to generate a personalized set of measurement parameter configurations.

[0057] The system integrates the key parameters output from each of the aforementioned stages to form a complete set of personalized measurement parameter configurations. This set is not merely a simple summary of multiple independent parameters, but rather an organic integration of multi-dimensional decision-making results, representing the system's comprehensive understanding of the current measurement situation and its optimal response strategy.

[0058] Understandably, this configuration set is then passed to subsequent blood pressure measurement modules to guide processes such as pressurization control, signal acquisition, and data calibration, ensuring the entire measurement process operates within the correct parameter framework. For example, postural parameters determine which set of postural compensation coefficients to use, arm-side parameters affect the selection of the inter-arm variability correction model, and the cuff pressure threshold directly controls the pump's start-up pressure and pressurization rate.

[0059] In the above embodiments, the traditional blood pressure measurement process, which relies on manual experience, is transformed into an automated, standardized, and personalized digital decision-making process. This significantly reduces data deviations caused by inconsistent measurement conditions and improves the comparability and reliability of measurement results. Simultaneously, the system, through closed-loop feedback design, multi-sensor fusion, and the introduction of clinical calibration models, ensures the scientific validity, safety, and practicality of the technical solution. It is not only suitable for home health management scenarios but can also be widely applied in demanding environments such as community health services, remote medical monitoring, and clinical research, providing solid technical support for achieving accurate blood pressure monitoring.

[0060] Reference Figure 3 As one embodiment of step S104, the step of selecting a target blood pressure measurement method based on arm side parameters and body position parameters includes: Step S301: Load arm side parameters and body position parameters, and simultaneously call the historical measurement method dataset associated with the user identifier; Among them, the arm side parameter clearly indicates whether the measurement will be based on the left or right arm, while the body position parameter describes the user's current body posture, such as sitting, lying or standing. These two parameters directly affect the hemodynamic state and signal acquisition quality.

[0061] Meanwhile, the historical measurement method dataset stores the specific measurement techniques and their results used by the user in multiple past measurements, such as whether oscillometric methods, pulse wave transit time (PTT), vibration-sensitive algorithms, or rapid boost modes were used. It also includes feedback information such as the success rate, signal quality index, and completion time of these methods under corresponding body positions and arm-side combinations. By accessing this long-accumulated personalized data asset, the system can understand which measurement methods perform better under the user's specific physiological conditions, thus providing an empirical basis for subsequent intelligent selection. For example, if an elderly user fails multiple rapid measurements in a supine position due to respiratory interference, the system can accordingly reduce the priority of such methods, demonstrating its deep learning ability to recognize individual differences.

[0062] Step S302: Generate constraints for position-related measurement methods based on the rule base for matching positional parameters with positional characteristics; The postural characteristic rule base is a structured knowledge set that embeds criteria for judging the adaptability of measurement techniques to different postural states. For example, when the postural parameter is "supine," the user's trunk and limbs are relatively still, but the breathing amplitude is large and chest and abdominal movements may be transmitted to the upper limbs, easily introducing low-frequency vibration noise. Therefore, the rule base will enforce constraints to exclude measurement methods sensitive to small vibrations (such as certain fine oscillation analysis algorithms that rely on high-precision pressure sensors) to avoid artifact interference leading to misjudgment. When the postural position is "standing," the user's ability to maintain balance is limited, especially for the elderly or those who are weak, as it is difficult to remain still for a long time. In this case, the rule base requires priority to be given to rapid measurement methods, that is, to adopt a high-speed pressurization and rapid depressurization strategy to complete blood pressure estimation in a short time and reduce data drift caused by postural instability or fatigue.

[0063] It should be noted that these rules are not simply summaries of experience, but rather standardized judgment logic derived from extensive clinical trials and ergonomic studies, ensuring high repeatability and safety. Through this rule-matching mechanism, the system proactively avoids inappropriate measurement methods, constructing the first layer of "hard screening" barrier.

[0064] Step S303: Query the arm side physiological difference database based on arm side parameters to obtain the adaptation weights of arm side related measurement methods; This step focuses on the inherent differences in human anatomy and hemodynamics, reflecting a technological upgrade from the "structural homogeneity assumption" to "individual vascular characteristic modeling." Studies have shown that there are systematic differences in the length of vascular pathways, arterial elasticity, and branching structures between the left and right upper arms. In particular, the right subclavian artery originates directly from the aortic arch, resulting in a shorter blood flow path and faster pressure transmission, while blood flow in the left upper arm must pass through the aortic arch, resulting in a slightly longer pulse wave propagation time (PWV).

[0065] Therefore, when using non-invasive blood pressure estimation methods based on pulse wave propagation time (PTT), the left arm often provides more stable time delay characteristics, making it suitable for high-precision continuous monitoring. Conversely, the traditional pressure oscillation method (oscillometric method) typically yields more reliable systolic and diastolic blood pressure determination points in the right arm due to higher signal strength and a better signal-to-noise ratio. The arm-side physiological difference database is constructed based on such clinical research data, recording the statistical characteristics of the performance of different measurement methods in the left and right arms. When the arm-side parameter is the left upper arm, the system automatically assigns a higher adaptation weight to pulse wave propagation velocity measurement methods; when it is the right upper arm, the weight of the pressure oscillation measurement method is increased.

[0066] Understandably, this differentiated weighting mechanism is not a simple preference setting, but a scientific decision support based on the calibration of hemodynamic models and measured data, which makes the selection of measurement methods more in line with actual physiological conditions.

[0067] In step S304, by combining the historical measurement method dataset, the position-related measurement method constraints, and the arm-side related measurement method adaptation weights, a target blood pressure measurement method instruction is generated through the measurement method selector and output to the blood pressure measurement control module.

[0068] The measurement method selector employs a hierarchical fusion decision architecture, combining the advantages of both rule-driven and data-driven approaches. It first determines if a historical successful record exists that perfectly matches the current body position and arm position. If such a record exists, it directly reuses that historical method, reflecting the robust principle of "prioritizing proven effective methods" and avoiding unnecessary technology switching risks. If no perfect matching record exists, a comprehensive scoring mechanism is activated to quantitatively evaluate all candidate measurement methods.

[0069] Specifically, the scoring model comprehensively considers three dimensions: first, the weight of historical usage frequency, which reflects the past success rate and stability of a method on the user; second, the compliance with body position constraints, which measures whether the method meets the safety and applicability requirements under the current body position (such as whether it is excluded by the rule base); and third, the weight of arm side adaptation, which reflects the theoretical performance advantage of the method on the current measuring arm.

[0070] In this embodiment, these three factors are fused using a weighted summation method. The weight coefficients (α, β, γ) are calibrated using large-scale clinical data to ensure balance and interpretability among different dimensions. For example, historical experience has the highest weight (α=0.5), reflecting individualization priority; postural compliance is secondary (β=0.3), ensuring basic safety; and arm-side compatibility serves as a supplement (γ=0.2), optimizing accuracy potential. Finally, the system selects the measurement method with the highest overall score as the target scheme and generates standardized control instructions.

[0071] Finally, the target blood pressure measurement method command is output to the blood pressure measurement control module, completing the closed-loop transmission from decision-making to execution. This command includes specific measurement mode identifiers, parameter configuration suggestions (such as pressurization rate, sampling frequency, signal filtering strategy, etc.), and is strictly matched with the downstream hardware control system interface to ensure that the command can be accurately parsed and executed. For example, if "rapid oscillometric method + right arm priority" is selected, the control module will start the high-speed air pump, adjust the depressurization gradient, and activate the dedicated right arm signal processing channel, thereby achieving coordinated optimization of the entire process.

[0072] The above implementation achieves intelligent, personalized, and safe selection of blood pressure measurement technology pathways. It transforms the originally static and fixed measurement process into an adaptive system that can adjust in real time according to the user's condition. This not only significantly improves the measurement success rate and data reliability in complex scenarios but also fully integrates clinical medical knowledge, biomechanical characteristics, and machine learning concepts, forming a highly interpretable and engineering-feasible technological closed loop.

[0073] Reference Figure 4 As one implementation of step S106, the steps of filtering the original cuff pressure signal, extracting pressure oscillation features, and performing frequency domain analysis on the original pulse wave signal to calculate the original pulse rate value include: Step S401: Receive the original pressure signal from the cuff and the original pulse wave signal; The original pressure signal of the cuff is acquired by a built-in pressure sensor, reflecting the change curve of the air pressure inside the cuff over time. This includes not only the small pressure oscillations caused by the periodic pulsation of the brachial artery (i.e., pressure oscillation waves), but also low-frequency drift caused by respiratory movements, muscle tremors, cuff loosening, and high-frequency noise from the electronic circuitry itself. The original pulse wave signal is usually acquired by a photoplethysmography (PPG) sensor, which records the fluctuations in light absorption rate of the finger or wrist tissue due to changes in blood volume during each heartbeat cycle. Its morphology is affected by various factors such as blood perfusion, skin contact pressure, and ambient light.

[0074] Step S402: Perform multi-level adaptive filtering on the original pressure signal of the cuff to eliminate motion artifacts and high-frequency environmental noise, and generate a pre-processed pressure signal. In this process, multi-stage filtering is not a simple cascading operation, but rather a targeted suppression that is carried out in a layered and progressive manner according to the type of noise and the characteristics of the signal.

[0075] Specifically, the first stage uses a moving average filter, which is mainly used to eliminate sudden impulse noise (such as spike interference caused by sudden user movement or equipment collision). By averaging the data points within a local time window, it smooths out instantaneous outliers in the signal while preserving the overall trend.

[0076] The second stage introduces wavelet thresholding denoising technology, a nonlinear time-frequency analysis method that effectively separates high-frequency noise from useful oscillation components in a signal. Wavelet transform decomposes the signal into sub-bands of different scales, where the high-frequency sub-bands mainly correspond to random noise. By setting an appropriate threshold to zero their coefficients and then reconstructing the signal, the signal-to-noise ratio can be significantly improved without compromising the detailed features of the pressure oscillation wave.

[0077] The third stage employs dynamic compensation filtering based on accelerometer signals, which is one of the key innovations of this solution: when the user makes slight limb movements during the measurement process, the accelerometer captures triaxial motion information in real time and uses it as a reference input to an adaptive noise cancellation algorithm (such as an LMS or RLS filter) to estimate and subtract motion-induced artifacts. For example, when the arm is slightly raised, it causes a brief increase in the cuff pressure reading. The system can identify this movement trend through the acceleration signal and compensate in reverse in the pressure signal, thereby restoring the true intravascular pressure change. This "sensor collaboration + adaptive modeling" mechanism enables the system to maintain high measurement stability even in non-stationary states.

[0078] Step S403: Extract pressure oscillation features from the preprocessed pressure signal; The extraction of pressure oscillation characteristics is first achieved through Hilbert transform, a mathematical tool that converts real-valued signals into their analytical form, thereby generating an instantaneous amplitude envelope. This envelope clearly depicts the typical "bell-shaped curve" of arterial pulsation amplitude as the cuff pressure gradually decreases. Subsequently, the system identifies the pressure oscillation peak corresponding to each cardiac cycle using a local maximum detection algorithm, forming a complete peak sequence and fitting the envelope trend. The inflection point or maximum slope point of this envelope is often used to estimate systolic and diastolic blood pressure.

[0079] Step S404: The original pulse wave signal is resampled and aligned, and then converted to the frequency domain by fast Fourier transform to calculate the fundamental frequency value corresponding to the main spectrum energy peak. Since the pulse wave sensor and the pressure sensor may use different hardware sampling rates (e.g., 100Hz for PPG and 50Hz for pressure signal), direct comparison or joint analysis will lead to time misalignment, so resampling and alignment processing must be performed.

[0080] Specifically, the system uses the sampling rate of the cuff pressure signal as a benchmark and performs linear interpolation resampling on the pulse wave signal. This involves redistributing the original PPG data points according to timestamps onto a time grid consistent with the pressure signal, ensuring that both have corresponding values ​​at every moment for subsequent synchronous analysis. After alignment, the system applies Fast Fourier Transform (FFT) to convert the time-domain pulse wave signal to the frequency domain, revealing its frequency composition structure. In the spectrogram, the dominant frequency component corresponding to the heartbeat exhibits a clear energy peak, but this peak may drift or split due to respiratory variability, motion artifacts, or other physiological disturbances.

[0081] To address this, the system employs a window-width-controllable integral energy search strategy: within a sliding window with a width of 0.5 Hz, the power spectral density (PSD) integral energy of each frequency band is calculated, and the center frequency with the largest energy integral is identified as the fundamental frequency corresponding to the main spectral energy peak. This method is more stable than simply taking the point with the largest amplitude, effectively suppressing spectral leakage and harmonic interference, and is particularly suitable for elderly individuals with large heart rate fluctuations or for measurement scenarios after exercise.

[0082] Step S405: Calculate the original pulse rate value based on the fundamental frequency value and mark it with a timestamp.

[0083] The fundamental frequency (FQ) is the number of heartbeats per second (Hz), multiplied by 60 to obtain the pulse rate (bpm). This value represents the instantaneous heart rate level at the moment of measurement and is one of the important indicators for assessing cardiovascular status. The timestamp is crucial; it not only records the accurate moment of pulse rate calculation but also establishes a spatiotemporal correlation with simultaneously acquired information such as blood pressure and postural parameters, enabling all physiological data to be stored, reviewed, and analyzed within a unified time coordinate system. For example, when generating a health report, the system can match blood pressure and pulse rate data within the same measurement period using timestamps to determine if there is a stress response pattern of increased heart rate accompanied by elevated blood pressure, or to identify abnormal phenomena such as nighttime bradycardia.

[0084] The above implementation achieves the transformation from raw noise signals to high-reliability physiological parameters. By integrating modern digital signal processing technology with classical blood pressure measurement theory, various data processing algorithms are used to improve data quality. At the same time, mechanisms such as acceleration compensation and resampling alignment are used to cope with complex interference in real-world usage environments, thereby improving the measurement accuracy and stability of the device in uncontrolled scenarios.

[0085] Reference Figure 5 As one implementation of step S108, the step of calling the arm-side difference calibration model corresponding to the arm-side parameters and calculating the calibrated systolic blood pressure value and calibrated diastolic blood pressure value based on the calibration pressure characteristics includes: Step S501: Load arm side parameters and calibrate pressure characteristics; Among them, the arm-side parameters clearly indicate whether the left or right arm was used in this measurement. This information is not only the basis for subsequent model calls, but also the starting point of the entire calibration process. The calibration pressure characteristics are pressure oscillation wave envelope data after body position compensation, which has initially eliminated the influence of hydrostatic pressure differences and changes in blood distribution caused by changes in sitting or lying posture, representing a relatively pure arterial pulsation pressure response under the current measurement conditions.

[0086] Step S502: Call the corresponding arm-side difference calibration model according to the arm-side parameters; wherein, the arm-side difference calibration model contains the mapping relationship between the left arm physiological feature database and the right arm physiological feature database; The design of the arm-side difference calibration model stems from a profound understanding of the differences in the vascular anatomy of the upper limb in clinical medicine: there are systematic differences in the origin, course, branching angle, and hemodynamic characteristics of the arteries in the left and right upper arms. For example, the blood supply to the right upper arm comes directly from the first major branch of the aortic arch—the brachiocephalic trunk, which has a short blood flow path, high velocity, and high inertia, exhibiting a stronger pressure transmission response during cuff compression. In contrast, the left upper arm is supplied by the left subclavian artery, whose initial segment forms a large angle with the aortic arch, and some individuals are at risk of "subclavian steal syndrome," resulting in relatively sluggish blood flow and pressure waveforms that are easily affected by local stenosis or compression effects.

[0087] Therefore, the left arm physiological feature database stores key anatomical parameters such as the subclavian artery angle and the depth distribution parameters of the brachial artery in subcutaneous tissue, used to assess blood flow resistance and signal attenuation. The right arm physiological feature database focuses on parameters reflecting arterial elasticity and blood flow inertia, such as peak blood flow velocity and vessel wall thickness gradient values ​​in the brachiocephalic trunk. Both databases are built based on large-scale clinical ultrasound imaging and simultaneous blood pressure measurement data, possessing high statistical representativeness and physiological relevance. When the system recognizes that the current measurement is for the left arm, it automatically activates the matching logic of the left arm feature database, and vice versa, thereby achieving precise modeling tailored to each arm.

[0088] Step S503: Input the calibration pressure characteristics into the arm side difference calibration model, and calculate the vascular compliance compensation factor through a multilayer perceptron neural network; Among them, the Multilayer Perceptron (MLP), as a feedforward neural network, possesses powerful nonlinear mapping capabilities, enabling it to learn implicit physiological patterns from complex input features. Its input layer receives engineered feature parameters across multiple dimensions, including time-domain statistics of calibrated pressure features, such as the mean reflecting the overall pressure level, variance reflecting the pulsation amplitude fluctuation, and skewness describing waveform symmetry (used to identify abnormal pulse morphology). It also introduces the ratio of pressure change rates during the cuff compression and decompression phases. This indicator indirectly reflects the dynamics of vascular filling and emptying under different external pressure conditions and is an important proxy variable for assessing vascular compliance.

[0089] Next, after sufficient training, the neural network outputs a scalar vascular compliance compensation factor δ, whose value is limited to the range of [0.85, 1.15], representing the elasticity adjustment coefficient relative to the standard vascular state. For example, if a user's left arm is stiffened and compliance is reduced due to arteriosclerosis, δ may approach 0.85, indicating that the original pressure characteristics need to be moderately amplified to restore the true blood pressure; conversely, if the vascular elasticity is good, δ is close to 1.0 or even slightly higher than 1.0. This compensation factor is not a simple amplification factor, but a comprehensive correction parameter that integrates the vascular wall deformation capacity, blood viscosity resistance, and pulse wave propagation characteristics, and has clear physiological significance.

[0090] Step S504: Correct and calibrate the pressure feature peak sequence based on the vascular compliance compensation factor to generate the final pressure feature vector; After obtaining the vascular compliance compensation factor, the system applies it to the peak sequence of calibrated pressure features, performing point-by-point correction to generate the final pressure feature vector. This process does not uniformly scale all data points, but rather adaptively adjusts the pressure peak value for each cardiac cycle based on its corresponding local vascular state.

[0091] For example, if a significantly low compensation factor is found in certain heartbeats during continuous measurements, it may indicate transient vasospasm or sympathetic nerve excitation. The system can apply stronger correction to these abnormal points while preserving the original fluctuation trend. The resulting pressure feature vector not only retains the morphological characteristics of the original pressure oscillations but also incorporates the dynamic compensation results of individual vascular biomechanical characteristics, becoming the optimal input for subsequent blood pressure calculations. This correction mechanism effectively overcomes the systematic errors introduced by using fixed compensation coefficients in traditional methods, and is particularly effective in elderly hypertensive patients, individuals with diabetic microvascular complications, or those recovering from surgery.

[0092] Step S505: Calculate the calibrated systolic pressure value and the calibrated diastolic pressure value based on the final pressure characteristic vector.

[0093] The system calculates the calibrated systolic blood pressure (SBP) and calibrated diastolic blood pressure (DBP) based on the extreme points in the final pressure feature vector, namely the maximum value P_max and the minimum value P_min of the pressure oscillation envelope, combined with a clinically calibrated physical model. This calculation model does not simply take the inflection point of the envelope, but introduces a coupling term between the cuff reference pressure difference ΔP and the compensation factor δ, forming a dynamic correction expression.

[0094] Specifically, α and β are empirical coefficients calibrated through numerous clinical controlled trials, ensuring good generalization performance across different populations and device platforms. By embedding the compensation factor δ into the blood pressure calculation formula, the final result not only reflects the current pressure oscillation amplitude but also the deformation response of the blood vessel wall under external pressure, thus more closely approximating the true intra-arterial pressure level. For example, even if two users have the same pressure oscillation peak, if their vascular compliance differs significantly (δ is 0.9 and 1.1 respectively), the calculated blood pressure value will be adjusted accordingly, avoiding misjudgment of "same waveform, same blood pressure".

[0095] The above implementation achieves deep integration from macroscopic measurement signals to microscopic vascular characteristics, transforming previously overlooked anatomical differences between the left and right arms into quantifiable, modelable, and compensable calibration criteria, significantly improving the accuracy and individual adaptability of blood pressure measurement. Especially in long-term health management scenarios, this method effectively eliminates data jumps caused by changing the measuring arm, enhancing the continuity and reliability of blood pressure trend analysis.

[0096] Reference Figure 6 As one implementation of step S109, the step of inputting the calibrated systolic blood pressure value, calibrated diastolic blood pressure value, raw pulse rate value, and user historical measurement records into the time series analysis model includes: Step S601: Load the calibrated systolic blood pressure value, calibrated diastolic blood pressure value, raw pulse rate value, and user historical measurement records, and merge them to generate a comprehensive time series dataset; The calibrated blood pressure value eliminates systematic errors caused by differences in body position, arm side selection, and vascular compliance, representing the closest estimate to the true arterial pressure under the current condition. The raw pulse rate value reflects the basic state of autonomic nervous system regulation and cardiac function. Historical measurement records typically contain multiple measurements taken by the user over a continuous period of 30 days or even longer, covering blood pressure fluctuation patterns at different time periods (such as morning peak and nighttime) and under different physiological states (such as after exercise and at rest).

[0097] Understandably, by integrating current measurements with these historical data on a timeline, the system constructs a complete data view containing both current and past information. This integrated time series not only includes the numerical values ​​themselves but also carries metadata such as the timestamp of each record, measurement conditions (e.g., body position, arm side), and signal quality score, providing rich contextual support for subsequent advanced analysis.

[0098] Step S602: Perform time dimension alignment processing on the comprehensive time series dataset to generate normalized time series data; This step is a crucial preprocessing step to ensure the effectiveness of subsequent model analysis. Because user measurement behavior is highly irregular (it may involve multiple measurements in a day, or no measurements for several days), the raw data exhibits a sparse and non-uniform distribution over time. Directly inputting this data into the model can lead to disordered time dependencies or biased feature extraction.

[0099] Therefore, the system uses the timestamp of each record as a benchmark to implement standardized reconstruction of the time dimension. Specifically, the entire time series is first divided into a unified time grid (such as three fixed time periods: morning, noon, and evening each day). For missing time points, linear interpolation is used to fill in the gaps. For example, if no measurement is taken on a certain day but data is available on the two days before and after, the system estimates the possible blood pressure level for that day based on the linear trend of adjacent values, avoiding misjudgment of the trend due to data gaps. At the same time, all physiological parameters (systolic blood pressure, diastolic blood pressure, and pulse rate) are uniformly resampled to a preset sampling frequency (such as once a day), forming an evenly spaced time series structure. In addition, this process also implicitly includes the ability to identify outliers. For example, when a measurement value deviates significantly from the interpolation prediction range, a data reliability verification mechanism can be triggered to prevent erroneous data from interfering with the overall trend judgment.

[0100] Step S603: Input the normalized time series data into the pre-configured time series analysis model, and extract dynamic trend features and risk probability vectors through the sliding window mechanism; This stage is the core intelligent engine for generating the entire health report, tasked with uncovering potential patterns and early warning signals from the data.

[0101] In this embodiment, the sliding window mechanism is not a simple slice of fixed size, but rather the window span is adaptively adjusted according to the length of the user's historical data: for new users with only short-term data, a smaller window (e.g., 7 days) is used to quickly respond to changes; while for older users with stable records over several months, a larger window (e.g., 30 days) is used to capture long-term trends and avoid excessive warnings due to short-term fluctuations. Within this window, the system calculates several key dynamic indicators, among which the blood pressure slope change coefficient is used to quantify the strength of the trend of blood pressure rising or falling over time. For example, if the morning systolic blood pressure shows an increasing trend for several consecutive days and the slope exceeds a threshold, it may indicate poor drug control or deteriorating lifestyle; the pulse rate variability reflects the stability of heart rate regulation. Low variability is often associated with sympathetic hyperactivity and decreased cardiovascular regulatory capacity, and is a risk precursor to potential cardiac events.

[0102] At the same time, the model also outputs a risk probability vector, which predicts the probability distribution of hypertensive crisis, arrhythmia, or the need for medical intervention within the next 7 days based on statistical learning algorithms (such as LSTM or survival analysis models). These features together constitute a multidimensional characterization of the user's cardiovascular status, including both observable trends and potential uncertainties, providing a scientific basis for the generation of health reports.

[0103] Step S604: Based on dynamic trend features and risk probability vectors, blood pressure level is determined, health trend is predicted, and risk assessment is marked, and a user health report is synthesized.

[0104] The system synthesizes the final health report based on the extracted dynamic trend features and risk probability vectors. This synthesis process is not a simple accumulation of information, but a structured and semantic content generation logic.

[0105] First, blood pressure levels are mapped to clear discrete classification labels, such as "normal", "high normal", "grade 1 hypertension", and "grade 2 hypertension", which are automatically determined based on preset hypertension prevention and control standards, making it easy for users to quickly understand their current status.

[0106] Secondly, the health trend prediction uses an exponential smoothing model to generate a visualized trend curve. This model assigns higher weight to recent data, enabling it to sensitively reflect the latest trends while smoothing out random fluctuations, and outputting a prediction of blood pressure trends for the next week. For example, the system may suggest, "Your systolic blood pressure has been slowly rising over the past two weeks. We recommend you increase monitoring and consult your doctor."

[0107] Finally, the risk assessment label automatically generates a warning label based on a preset probability threshold: when the probability of a certain type of event (such as nocturnal hypertension) in the risk probability vector exceeds the critical value (such as 30%), the system will mark it as "moderate risk" and provide a brief explanation and suggested measures.

[0108] The entire user health report is organized into a coherent text using natural language generation technology, balancing professionalism and readability. It can be used as a reference for users' self-health management and as an important basis for remote medical consultations.

[0109] In the above implementation, static measurement is upgraded to dynamic insight. The system improves the foresight and personalization of health assessment through advanced technologies such as adaptive sliding window, trend slope analysis and risk probability modeling. It is especially suitable for scenarios such as long-term management of chronic diseases, health monitoring of the elderly and home telemedicine.

[0110] Reference Figure 7 As a further implementation of the intelligent multifunctional blood pressure and pulse rate monitoring method, the monitoring method also includes: Step S701: Periodically scan the user health report update records in the local database. When the number of update reports within a preset period exceeds a preset threshold, trigger a group health analysis command. The local database uses an incremental update mechanism to store health reports uploaded by terminal devices. Each report includes a precise timestamp, user identifier, and a summary of measurement parameters. The system executes scans periodically via a scheduled task (such as a Cron-based background job), typically hourly or daily, to maintain a stable monitoring rhythm. The scanning process employs a sliding time window algorithm to count the number of new reports within a specific area or institution within a specified period (e.g., the past 24 hours).

[0111] Specifically, the threshold for determining whether to trigger analysis is not a fixed constant, but rather dynamically calculated based on the amount of historical data from the same period using exponential smoothing. This method can effectively filter out seasonal fluctuations and daily baseline changes, identifying truly significant data anomalies. For example, in a community nursing home scenario, if historical data shows an average of 15 new reports per day, but suddenly increases to 50 reports on a given day, exceeding three times the standard deviation, the system will determine it as an abnormal event, which is highly likely to reflect a sudden health crisis (such as a flu-induced blood pressure fluctuation).

[0112] At this point, the system automatically generates a group health analysis command, publishes it as an internal message queue event, carrying metadata such as timestamp, region code, and data mutation coefficient, and initiates subsequent high-level analysis processes. This triggering mechanism based on data density mutation avoids the waste of computing resources caused by indiscriminately processing all data, and realizes on-demand resource access and response priority ranking.

[0113] Step S702: Retrieve the historical health dataset of the target user group from the cloud server; wherein the target user group is determined based on at least one dimension, including age stratification, disease tags, or geographic location. Specifically, the target group is not defined by random sampling, but rather by precise screening based on at least one dimension, such as age stratification, disease labels, or geographic location, to ensure that the subjects analyzed have clinical relevance and public health significance. The cloud server adopts a high-performance distributed columnar storage architecture, supporting efficient reading and joint querying of massive amounts of structured and semi-structured health data.

[0114] In terms of age stratification, the system follows the World Health Organization standards, dividing the population into three major categories: adolescents, adults, and the elderly. Furthermore, it applies the K-means clustering algorithm to subdivide subgroups based on indicators such as blood pressure level and heart rate variability, and identifies high-risk subgroups (such as the elderly with prehypertension).

[0115] Regarding disease labels, the system is based on the International Classification of Diseases, 11th Revision (ICD-11) coding system. The system can automatically identify and collect users with chronic diseases such as hypertension, diabetes, and coronary heart disease, and associate their complication risks through knowledge graph technology (such as the increased probability of stroke in hypertensive patients).

[0116] In terms of geographic location, a GIS (Geographic Information System) is used to divide the data into grid units of 500 meters. The analysis range is dynamically adjusted using population heat maps to achieve fine-grained control of spatial granularity. During data retrieval, the system executes multi-dimensional joint filtering statements through a distributed query engine to quickly extract complete historical data sets that meet the criteria. For example: "Select all measurement data from the past 6 months for all users aged 65 and above, diagnosed with primary hypertension, and residing in grid G3305 within a certain urban area."

[0117] Step S703: Perform multimodal feature fusion processing on the historical health dataset to generate a population feature library; wherein, the population feature library includes a blood pressure fluctuation pattern matrix, a pulse rate abnormality correlation map, and a medication feedback vector. Specifically, the blood pressure fluctuation pattern matrix separates high-frequency (short-term fluctuations) and low-frequency (long-term trends) components by performing wavelet transform on the 24-hour time series of systolic and diastolic blood pressure for each user. This results in the construction of an m×n feature matrix, where m represents the number of samples and n covers multiple dimensions such as the diurnal rhythm coefficient (reflecting the proportion of blood pressure drop at night, used to identify "non-dipping" hypertension), blood pressure variability (standard deviation, measuring stability), and the probability of morning surge (a sharp rise in blood pressure in the early morning, a high-risk factor for cardiovascular and cerebrovascular events). These features together characterize the dynamic profile of the group's blood pressure.

[0118] The pulse rate abnormality correlation map requires the introduction of graph neural network concepts. First, the key indicator of individual pulse rate variability (HRV), RMSSD (reflecting parasympathetic activity), is calculated. Then, a graph structure is constructed with users as nodes and spatiotemporal correlations as edge weights. For example, if user A experiences a significant heart rate abnormality and user B experiences a similar event within 48 hours, and the two are geographically close or have close social relationships, the connection weight between them is strengthened. By learning the topological characteristics of this map through a graph convolutional network (GCN), the system can discover potential propagation paths of group autonomic nervous system disorders.

[0119] Meanwhile, the medication feedback vector quantifies the actual effect of drug intervention, including medication adherence score (the ratio of actual medication frequency to prescription frequency) and changes in blood pressure before and after medication (ΔSBP_med, ΔDBP_med), forming a three-dimensional vector to assess the combined effect of drug efficacy and behavioral adherence.

[0120] Ultimately, these three heterogeneous features are integrated into a unified population feature library through feature cascading, forming a multidimensional representation system that includes both physiological dynamics and behavioral intervention information, providing rich input for epidemiological modeling.

[0121] Step S704: Input the population feature database into the pre-trained epidemiological prediction model and output a regional disease risk warning report; The epidemiological prediction model employs an advanced spatiotemporal graphical neural network (ST-GNN) architecture, capable of simultaneously capturing spatial propagation patterns and temporal evolution trends. Its spatial dependency modeling module utilizes Chebyshev multinomial approximation technology to calculate disease transmission weights between adjacent geographic grids, simulating potential spread paths of chronic diseases such as hypertension in the population. The temporal dynamics capture module uses gated recurrent units (GRUs) to perform long-term dependency learning on temporal features such as blood pressure and pulse rate, identifying patterns of upward trends or periodic deterioration.

[0122] It should be noted that the model introduces medication feedback vectors as external covariates through an attention mechanism, assigning different weights to the differences in intervention responses among different users, thereby more realistically reflecting the impact of policy interventions on population health. The final output warning report contains multiple levels of information: first, a risk heatmap, visually marking high-risk areas in GeoJSON format (such as grid cells where the mean systolic blood pressure consistently exceeds 140 mmHg), allowing decision-makers to intuitively grasp the distribution trend; second, a warning index Rt (effective reproduction number), defined as the ratio of the number of newly added high-risk cases in the current week to the average of the previous four weeks. A red warning is triggered when Rt > 1.2, indicating that the health risk is spreading rapidly; third, a disease association tree, showing the probabilistic path of hypertension towards complications such as stroke and heart failure, supporting forward-looking health management planning. The entire prediction process not only relies on data-driven approaches but also integrates epidemiological transmission theory, significantly improving the scientific rigor and foresight of the warning.

[0123] Step S705: Generate a set of tiered intervention strategies based on the regional disease risk warning report, and push strategy execution instructions to the terminal of the designated medical institution through the communication module.

[0124] In this embodiment, the tiered intervention strategy engine employs a hybrid reasoning model combining rule-based reasoning (RBR) and case-based reasoning (CBR) to ensure that the strategy conforms to medical standards while also being context-adaptable. For example, when Rt > 1.5 and the elderly population accounts for more than 40%, the system automatically triggers a Level 1 response strategy: initiating a community doctor home screening program and deploying mobile health monitoring vehicles; when Rt is between 1.2 and 1.5, a Level 2 response is initiated, such as opening a green channel for cardiovascular specialists at tertiary hospitals for residents in the area; and for areas with a specific complication risk probability exceeding 30%, a Level 3 health education strategy is pushed, including distributing personalized educational materials or smart electronic pillboxes to improve medication adherence.

[0125] Specifically, these strategies are pushed to healthcare institution terminal devices (such as healthcare PDAs and on-call workstations) via a lightweight MQTT protocol in the form of structured JSON instruction packets. The messages include fields such as action type, target area, priority, and validity period. Upon receiving the instruction, the terminal verifies its source legitimacy through digital signature to prevent malicious tampering. After confirmation, medical staff execute the corresponding task. This automated instruction push mechanism significantly reduces the time delay from early warning to intervention, achieving standardized, precise, and real-time public health response.

[0126] The above implementation achieves intelligent transformation from individual health data to regional public health decision-making, breaking down information barriers between family health terminals and the primary healthcare system. Utilizing artificial intelligence and big data technologies, it transforms massive amounts of scattered physiological data into actionable public health guidelines, significantly improving the timeliness and accuracy of chronic disease prevention and control, and optimizing the efficiency of medical resource allocation. Furthermore, the medication feedback and blood pressure fluctuation data accumulated by the system can, in turn, assist health management departments in optimizing drug procurement strategies, promoting the transformation of the healthcare system from a "treatment-centered" to a "health-centered" approach, and providing core technological support for building a smart and resilient modern public health governance system.

[0127] This application also discloses an intelligent multifunctional blood pressure and pulse rate monitoring system.

[0128] A smart multifunctional blood pressure and pulse rate monitoring system, the monitoring system includes: The user identification module is used to acquire and parse user identification data to generate user identifiers. The data loading module is used to load the user's historical measurement records associated with the user identifier, and at the same time acquire real-time attitude sensor data. The parameter configuration module is used to generate a personalized set of measurement parameter configurations based on the user's historical measurement records and posture sensor data, including arm side parameters, body position parameters and cuff calibration pressure thresholds. The measurement method selection module allows you to choose the target blood pressure measurement method based on arm side parameters and body position parameters. The signal acquisition module is used to control the pressurization device based on the cuff calibration pressure threshold, and to simultaneously acquire the original cuff pressure signal and the original pulse wave signal. The signal processing module is used to filter the raw pressure signal from the cuff, extract pressure oscillation characteristics, and perform frequency domain analysis on the raw pulse wave signal to calculate the raw pulse rate value. The body position compensation module is used to call the body position compensation coefficient library corresponding to the body position parameters, perform compensation calculations on the pressure oscillation characteristics, and obtain the calibrated pressure characteristics. The arm side calibration module is used to call the arm side difference calibration model corresponding to the arm side parameters, and calculate the calibration systolic blood pressure value and calibration diastolic blood pressure value based on the calibration pressure characteristics. The health analysis module is used to input calibrated systolic blood pressure, calibrated diastolic blood pressure, raw pulse rate, and the user's historical measurement records into the time series analysis model, and output a user health report. The user health report includes blood pressure level judgment, health trend prediction, and risk assessment markers. The data storage module is used to encrypt and store the calibrated systolic blood pressure value, calibrated diastolic blood pressure value, raw pulse rate value and user health report to the local database, and synchronize them to the cloud server through the communication module.

[0129] The intelligent multifunctional blood pressure and pulse rate monitoring system of this application embodiment can implement any of the above methods, and the specific working process of each module in the system can refer to the corresponding process in the above method embodiment.

[0130] In the several embodiments provided in this application, it should be understood that the provided methods and systems can be implemented in other ways. For example, the system embodiments described above are merely illustrative; for example, the division of a certain module is merely a logical functional division, and in actual implementation there may be other division methods, such as multiple modules can be combined or integrated into another system, or some features can be ignored or not executed.

[0131] This application also discloses a computer device.

[0132] A computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement an intelligent multifunctional blood pressure and pulse rate monitoring method as described above.

[0133] This application also discloses a computer-readable storage medium.

[0134] A computer-readable storage medium storing a computer program that can be loaded by a processor and executed as described above in any of the intelligent multifunctional blood pressure and pulse rate monitoring methods.

[0135] The computer-readable storage medium can be any tangible medium that contains or stores a program that can be used by or in connection with an instruction execution system, apparatus, or device; the program code contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to wireless, wire, optical fiber, RF, etc., or any suitable combination thereof.

[0136] Although this application has been described herein in conjunction with various embodiments, those skilled in the art, by reviewing the accompanying drawings, disclosure, and appended claims, will understand and implement other variations of the disclosed embodiments in carrying out the claimed application. In the claims, the word "comprising" does not exclude other components or steps, and "a" or "an" does not exclude a plurality. A single processor or other unit can implement several functions listed in the claims. While different dependent claims may recite certain measures, this does not mean that these measures cannot be combined to produce a good effect.

[0137] The above are all preferred embodiments of this application and are not intended to limit the scope of protection of this application. Any feature disclosed in this specification (including the abstract and drawings) may be replaced by other equivalent or similar features unless specifically stated otherwise. That is, unless specifically stated otherwise, each feature is only one example of a series of equivalent or similar features.

Claims

1. A smart, multifunctional method for monitoring blood pressure and pulse rate, characterized in that, The monitoring method includes: Acquire and parse user identification data to generate user identifiers; Load the user's historical measurement records associated with the user identifier, and simultaneously acquire real-time attitude sensor data; Based on the user's historical measurement records and posture sensor data, a personalized set of measurement parameter configurations is generated, including arm side parameters, body position parameters, and cuff calibration pressure thresholds. Select the target blood pressure measurement method based on the arm side parameters and body position parameters; The pressurization device is controlled based on the cuff calibration pressure threshold, and the original cuff pressure signal and pulse wave signal are collected simultaneously. The original pressure signal of the cuff is filtered to extract pressure oscillation characteristics, and the original pulse wave signal is analyzed in the frequency domain to calculate the original pulse rate value. Call the body position compensation coefficient library corresponding to the body position parameters to perform compensation calculation on the pressure oscillation characteristics and obtain the calibrated pressure characteristics; Call the arm-side difference calibration model corresponding to the arm-side parameters, and calculate the calibrated systolic blood pressure value and calibrated diastolic blood pressure value based on the calibration pressure characteristics; The calibrated systolic blood pressure value, calibrated diastolic blood pressure value, raw pulse rate value, and the user's historical measurement records are input into the time series analysis model to output a user health report; the user health report includes blood pressure level judgment, health trend prediction, and risk assessment markers; The calibrated systolic blood pressure value, calibrated diastolic blood pressure value, raw pulse rate value, and user health report are encrypted and stored in a local database, and synchronized to a cloud server via a communication module.

2. The intelligent multifunctional blood pressure and pulse rate monitoring method according to claim 1, characterized in that, Based on the user's historical measurement records and posture sensor data, the steps to generate a personalized set of measurement parameter configurations, including arm side parameters, body position parameters, and cuff calibration pressure thresholds, include: Retrieve historical measurement records associated with the user identifier, and extract historical arm side selection data, historical body position selection data, historical cuff pressure thresholds, and user arm circumference data; Real-time acquisition of attitude sensor data, including triaxial accelerometer signals and gyroscope angular velocity signals; Recommended body position parameters are generated based on the historical body position selection data, and the real-time body position tilt angle is calculated based on the triaxial accelerometer signal and the gyroscope angular velocity signal. If the deviation between the real-time body tilt angle and the recommended body position parameter exceeds a preset deviation threshold, the body position type corresponding to the real-time body tilt angle will be used as the final body position parameter; if it does not exceed the preset deviation threshold, the recommended body position parameter will be retained as the final body position parameter. Recommended arm-side parameters are generated based on the historical arm-side selection data; wherein, if the user has no historical arm-side selection data, the recommended arm-side parameters are set to the default value. Based on the final body position parameters and user arm circumference data, the cuff calibration pressure threshold is output through a pre-configured cuff pressure calculation model. By integrating the final body position parameters, recommended arm side parameters, and cuff calibration pressure threshold, a personalized set of measurement parameter configurations is generated.

3. The intelligent multifunctional blood pressure and pulse rate monitoring method according to claim 2, characterized in that, The steps for selecting a target blood pressure measurement method based on the arm side parameters and body position parameters include: Load the arm side parameters and body position parameters, and simultaneously call the historical measurement method dataset associated with the user identifier; Based on the body position parameter matching body position characteristic rule library, generate body position related measurement method constraint conditions; Based on the arm-side parameters, query the arm-side physiological difference database to obtain the adaptation weights of relevant arm-side measurement methods; Combining the historical measurement method dataset, the position-related measurement method constraints, and the arm-side related measurement method adaptation weights, the target blood pressure measurement method instruction is generated through the measurement method selector and output to the blood pressure measurement control module.

4. The intelligent multifunctional blood pressure and pulse rate monitoring method according to claim 1, characterized in that, The steps of filtering the original pressure signal from the cuff, extracting pressure oscillation features, and performing frequency domain analysis on the original pulse wave signal to calculate the original pulse rate value include: Receive raw pressure signal from the cuff and raw pulse wave signal; The original pressure signal of the cuff is subjected to multi-level adaptive filtering to eliminate motion artifacts and high-frequency environmental noise, and a pre-processed pressure signal is generated. Extract pressure oscillation features from the preprocessed pressure signal; The original pulse wave signal is resampled and aligned, then converted to the frequency domain using a fast Fourier transform, and the fundamental frequency value corresponding to the main spectrum energy peak is calculated. The original pulse rate value is calculated based on the fundamental frequency value and a timestamp is marked.

5. The intelligent multifunctional blood pressure and pulse rate monitoring method according to claim 1, characterized in that, The steps of calling the arm-side difference calibration model corresponding to the arm-side parameters and calculating the calibrated systolic and diastolic blood pressure values ​​based on the calibration pressure characteristics include: Load the arm-side parameters and the calibration pressure characteristics; The corresponding arm-side difference calibration model is invoked based on the arm-side parameters; wherein, the arm-side difference calibration model contains the mapping relationship between the left arm physiological feature database and the right arm physiological feature database; The calibration pressure characteristics are input into the arm-side differential calibration model, and the vascular compliance compensation factor is calculated through a multilayer perceptron neural network. Based on the vascular compliance compensation factor, the pressure feature peak sequence is corrected and calibrated to generate the final pressure feature vector; The calibrated systolic blood pressure and calibrated diastolic blood pressure values ​​are calculated based on the final pressure feature vector.

6. The intelligent multifunctional blood pressure and pulse rate monitoring method according to claim 5, characterized in that, The steps of inputting the calibrated systolic blood pressure value, calibrated diastolic blood pressure value, raw pulse rate value, and the user's historical measurement records into the time series analysis model include: Load the calibrated systolic blood pressure value, calibrated diastolic blood pressure value, raw pulse rate value, and the user's historical measurement records, and merge them to generate a comprehensive time series dataset; Perform time dimension alignment processing on the comprehensive time series dataset to generate normalized time series data; The normalized time series data is input into a pre-configured time series analysis model, and dynamic trend features and risk probability vectors are extracted through a sliding window mechanism. Based on the dynamic trend features and risk probability vector, blood pressure level is determined, health trend is predicted, and risk assessment is marked, and the user's health report is synthesized.

7. A smart multifunctional blood pressure and pulse rate monitoring method according to any one of claims 1 to 6, characterized in that, The monitoring method also includes: Periodically scan the user health report update records in the local database. When the number of update reports within a preset period exceeds a preset threshold, trigger a group health analysis command. Retrieve historical health datasets of the target user group from a cloud server; wherein the target user group is determined based on at least one dimension, including age stratification, disease tags, or geographic location; The historical health dataset is subjected to multimodal feature fusion processing to generate a population feature library; wherein, the population feature library includes a blood pressure fluctuation pattern matrix, a pulse rate abnormality correlation map, and a medication feedback vector; The population feature database is input into a pre-trained epidemiological prediction model, which outputs a regional disease risk warning report. Based on the regional disease risk warning report, a set of tiered intervention strategies is generated, and strategy execution instructions are pushed to the terminal of the designated medical institution through the communication module.

8. An intelligent multifunctional blood pressure and pulse rate monitoring system, characterized in that, The monitoring system includes: The user identification module is used to acquire and parse user identification data to generate user identifiers. The data loading module is used to load the user's historical measurement records associated with the user identifier, and at the same time acquire real-time attitude sensor data; The parameter configuration module is used to generate a personalized set of measurement parameter configurations based on the user's historical measurement records and posture sensor data, including arm side parameters, body position parameters and cuff calibration pressure thresholds. The measurement method selection module selects the target blood pressure measurement method based on the arm side parameters and body position parameters. The signal acquisition module is used to control the pressurization device based on the cuff calibration pressure threshold and to simultaneously acquire the original cuff pressure signal and the original pulse wave signal. The signal processing module is used to filter the original pressure signal of the cuff, extract the pressure oscillation characteristics, and perform frequency domain analysis on the original pulse wave signal to calculate the original pulse rate value. The body position compensation module is used to call the body position compensation coefficient library corresponding to the body position parameters, perform compensation calculations on the pressure oscillation characteristics, and obtain the calibrated pressure characteristics. The arm side calibration module is used to call the arm side difference calibration model corresponding to the arm side parameters, and calculate the calibration systolic blood pressure value and calibration diastolic blood pressure value based on the calibration pressure characteristics. The health analysis module is used to input the calibrated systolic blood pressure value, calibrated diastolic blood pressure value, raw pulse rate value, and the user's historical measurement records into a time-series analysis model, and output a user health report; wherein, the user health report includes blood pressure level judgment, health trend prediction, and risk assessment markers; The data storage module is used to encrypt and store the calibrated systolic blood pressure value, calibrated diastolic blood pressure value, raw pulse rate value and user health report to the local database, and synchronize them to the cloud server through the communication module.

9. A computer device, characterized in that: The method includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the program, implements the method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The computer program is stored that can be loaded by a processor and executed as described in any one of claims 1 to 7.

Images