A wearable blood pressure monitoring device steady state baseline self-calibration method

Abstract

The application discloses a kind of wearable blood pressure monitoring equipment steady-state baseline self-calibration method, to 200Hz acquisition pulse pressure signal, first by temperature subsection compensation noise reduction;Extract five steady-state parameters such as pulse period fluctuation coefficient, waveform similarity, all meet the standard for 5s continuously, then determine steady-state window;Eliminate abnormal points and solve baseline by mean, avoid numerical jump by linear weighted smoothing update baseline;According to the historical baseline drift rate, the calibration sensitivity is self-adaptively adjusted;When baseline offset exceeds the threshold, trigger abnormality and open artificial calibration channel.The application only relies on pressure waveform to identify steady state, without additional sensing hardware, synchronously compensates creep and circuit zero drift, significantly reduces the probability of false calibration, extends the artificial calibration period from 4h to more than 8h, and the effective rate of 24h monitoring data is more than 95%, suitable for long-term static non-invasive blood pressure monitoring during sleep, postoperative bed rest, etc.

Description

Technical Field

[0001] This invention belongs to the field of wearable medical device signal calibration and non-invasive continuous blood pressure monitoring technology. Specifically, it relates to a steady-state baseline self-calibration method for wearable blood pressure monitoring devices. It is applicable to lightweight upper arm cuffless blood pressure monitoring devices without acceleration, photoelectric, or ECG assistance. It is particularly suitable for quasi-static 24-hour long-term continuous monitoring scenarios such as postoperative bed rest, nighttime sleep, and resting in the elderly with chronic diseases. It is mainly used to eliminate the cumulative baseline error caused by mechanical creep of flexible pressure sensors and zero-point drift of circuits. Background Technology

[0002] Currently, the baseline calibration technology for wearable cuffless blood pressure monitoring devices is relatively fixed. The mainstream calibration methods in the industry are mainly divided into three categories: fixed-cycle forced calibration, multi-sensor motion-triggered calibration, and manual calibration. All of these have significant technical defects and cannot meet the long-term static monitoring needs of lightweight pure pressure array devices.

[0003] The existing publicly available patents and literature have the following technical defects: Fixed-period calibration patents, represented by CN109965861, use fixed time intervals to force the baseline to zero, without considering the actual physiological state of the human body to determine the timing of calibration. This makes it easy to trigger miscalibration in non-steady-state scenarios such as limb activity and slight body movements, introducing secondary measurement errors and resulting in poor long-term monitoring accuracy and stability. Mainstream motion-triggered calibration schemes rely on auxiliary sensors such as accelerometers, gyroscopes, and photoelectric sensors to identify motion states, resulting in high hardware costs and high device power consumption. They are also not compatible with the minimalist hardware architecture of the first-generation pure pressure array, leading to low mass production compatibility. Traditional manual calibration requires users to frequently use medical cuff devices for manual recalibration, which is cumbersome, requires high human intervention, and cannot achieve 24-hour unattended continuous monitoring.

[0004] Meanwhile, existing calibration technologies generally suffer from technical shortcomings: most only make simple corrections for instantaneous baseline shifts, failing to address the dual cumulative errors caused by the mechanical creep and sinking of the polymer film and the slow drift of the circuit zero point resulting from long-term wear of flexible piezoresistive sensors; furthermore, there are currently no existing patents or published literature, and the only publicly available solutions rely solely on the temporal characteristics of the pressure pulse wave itself without any auxiliary sensing support for physiological steady-state adaptive calibration. This leads to severe error accumulation in lightweight cuffless devices during long-term monitoring, high frequency of manual calibration, and low data effectiveness, greatly limiting the application of these devices in static medical monitoring scenarios. Based on the aforementioned technological gaps, this invention proposes a novel technical solution of pure pressure waveform steady-state recognition and dynamic on-demand self-calibration. Summary of the Invention

[0005] This invention aims to overcome the shortcomings of existing fixed-cycle calibration, such as easy error correction, high cost of multi-sensor calibration, high frequency of manual calibration, and inability to eliminate long-term cumulative drift errors. It provides a steady-state baseline self-calibration method for wearable blood pressure monitoring devices. Based on the multi-dimensional temporal characteristics of pure pressure pulse waves, it intelligently identifies the physiological steady-state windows of the human body at rest and during sleep. Without any auxiliary sensors or manual intervention, it achieves baseline adaptive dynamic reset, simultaneously offsetting the cumulative errors of sensor mechanical creep and circuit zero-point drift, significantly reducing the frequency of manual calibration and ensuring 24-hour long-term static high-precision stable monitoring.

[0006] A method for steady-state baseline self-calibration of a wearable blood pressure monitoring device includes the following steps: S1. Signal Acquisition and Preprocessing: Raw pulse pressure signals are continuously acquired at a frequency of 200Hz through a multi-channel pure pressure array of the upper arm brachial artery. The temperature segmentation compensation algorithm is used to complete the pre-temperature drift correction, filter out power frequency noise and high frequency spikes, and output stable and continuous time-series pulse waveform data. S2. Multi-dimensional steady-state feature parameter extraction: Real-time frame-by-frame extraction of five core steady-state evaluation parameters of the pulse wave, namely pulse cycle fluctuation coefficient, waveform peak variation coefficient, baseline amplitude fluctuation range, waveform integrity, and waveform similarity between adjacent cycles; S3. Quantitative determination of physiological homeostasis window: A preset homeostasis threshold standard is used. When the pulse cycle fluctuation coefficient is <3%, the peak variation coefficient is <2%, the baseline fluctuation amplitude is <1.5 mmHg within 5 consecutive seconds, and the waveform integrity and similarity to the waveform of adjacent cycles meet the standard, the current state is determined to be an effective homeostasis calibration window for human rest / sleep. The optimal values ​​for each quantitative threshold in this invention have been obtained through multiple batches of human static clinical trials. S4. Steady-state baseline truth value solution: Lock the effective steady-state calibration window, remove a small number of discrete abnormal sampling points within the window, and calculate the true zero-point baseline of the current sensor through the mean fitting algorithm, which serves as the baseline calibration benchmark. S5. Progressive Baseline Smooth Update: The old and new baselines are adaptively replaced by a 1-second linear weighted transition method. The weight of the old baseline is gradually reduced while the weight of the new baseline is synchronously increased to achieve a smooth baseline transition and prevent sudden changes in blood pressure values ​​and waveform distortion. S6. Adaptive adjustment of calibration sensitivity: Real-time monitoring of the recent baseline drift rate of the equipment. The faster the drift rate, the higher the steady-state recognition trigger sensitivity and the more frequent the calibration. When the drift is stable, the calibration frequency is reduced, achieving low power consumption, high precision and on-demand calibration. S7. Over-limit Abnormal Calibration: Real-time monitoring of single baseline offset. When the baseline offset exceeds the preset limit threshold, an abnormality prompt is triggered, and the manual precision calibration entry for the medical cuff is retained, forming a dual guarantee mechanism with fully automatic calibration as the main method and manual calibration as a backup.

[0007] The key technologies of this invention include: 1. Steady-state identification and baseline calibration are completed solely based on the timing characteristics of pressure pulse waves, without relying on any auxiliary sensors such as acceleration, ECG, or photoelectric sensors; 2. Quantitative standard for steady-state calibration window: continuously meeting the threshold values ​​of multi-dimensional steady-state parameters for 5 seconds, optimizing the optimal values ​​for clinical trials; 3. Baseline update adopts a 1-second linear weighted smooth transition mechanism to ensure the continuity of monitoring data; 4. Synchronous compensation for the dual cumulative errors of sensor mechanical creep baseline sinking and circuit zero-point drift; 5. Adaptable to the first-generation pure pressure array device, extending the manual calibration cycle from 4 hours to more than 8 hours.

[0008] Unlike all trigger calibration schemes that rely on multimodal sensing such as acceleration, photoelectric, and ECG, this method determines physiological steady state solely based on the time-domain characteristics of the pressure pulse wave itself. It is fully compatible with the minimalist hardware architecture of the first-generation pure pressure array, requiring no additional hardware cost or power consumption. It abandons the rigid, industry-standard fixed-interval forced calibration mode, using the actual sleep and resting steady-state periods as the sole calibration trigger time, eliminating secondary errors introduced by miscalibration in non-steady-state scenarios from the source. It achieves simultaneous correction of two types of long-term accumulated errors: baseline sinking due to polymer mechanical creep in flexible sensors and slow drift of circuit zero points, overcoming the limitations of single error correction in existing technologies and completely solving the problem of long-term monitoring accuracy attenuation. It abandons the traditional step-type forced baseline replacement mode, adopting a linear weighted smooth transition mechanism to update the baseline, ensuring a continuous blood pressure monitoring trend curve without abrupt changes or data distortion, improving the integrity and usability of long-term monitoring data. It can dynamically adjust the calibration trigger sensitivity according to the device's real-time drift rate, achieving intelligent on-demand calibration; simultaneously, it sets baseline over-limit prompts and a manual calibration entry point, balancing intelligent monitoring with clinical safety.

[0009] Compared with existing technologies, the beneficial effects of the present invention are as follows.

[0010] 1. Completely avoids the defects of traditional calibration technology: Abandoning the fixed-cycle forced calibration mode, calibration is triggered only during the human body's true rest and sleep steady state, fundamentally solving the industry problem of secondary errors introduced by miscalibration in non-steady-state scenarios, and greatly improving calibration accuracy.

[0011] 2. Perfectly adapted to the hardware architecture of the first-generation product: It adopts a pure algorithm lightweight optimization solution, which does not require the addition of any sensors or hardware structures, does not increase the mass production BOM cost, and does not increase the device's operating power consumption. It is fully adapted to the low-cost, lightweight, and low-power mass production positioning of the first-generation device.

[0012] 3. Significantly improved long-term monitoring stability: Simultaneously eliminates two core cumulative errors: baseline sinking due to mechanical creep of flexible sensors under long-term pressure and slow drift of circuit zero position, effectively suppressing the attenuation of long-term monitoring accuracy and achieving 24-hour continuous high-precision static monitoring.

[0013] 4. Excellent monitoring data integrity: It adopts a 1-second linear weighted progressive baseline update mechanism to avoid the problems of sudden changes in blood pressure curve and data distortion caused by direct replacement of traditional baselines, and ensures the continuous integrity of long-term monitoring trend data.

[0014] 5. Significantly reduce manual operation and maintenance costs: Enables routine fully automated unattended calibration, significantly reducing the frequency of manual equipment calibration and operation and maintenance costs, and is suitable for long-term unattended scenarios such as postoperative monitoring, home-based elderly care, and sleep monitoring. Attached Figure Description

[0015] Figure 1 This is a flowchart illustrating the overall process of the steady-state baseline self-calibration method for the wearable blood pressure monitoring device of the present invention. Detailed Implementation

[0016] This embodiment is adapted to the first-generation 10-channel equidistant distributed flexible pressure array cuffless blood pressure monitoring device and is applied to nighttime sleep and postoperative bed rest quasi-static 24-hour continuous monitoring scenarios.

[0017] Step 1, Device Wearing and Signal Acquisition: The device wearing pre-tightening force is controlled at 15~20kPa. After the initial baseline calibration is completed, the upper arm brachial artery pulse pressure signal is continuously acquired at a frequency of 200Hz. Temperature drift interference is eliminated by the pre-temperature segmented compensation algorithm, and a stable pre-processed waveform is output.

[0018] Step 2: Real-time monitoring of steady-state parameters: The device calculates multi-dimensional steady-state parameters such as pulse cycle fluctuation coefficient, peak variation coefficient, and baseline fluctuation amplitude in real time to continuously monitor waveform stability.

[0019] Step 3, Steady-state window determination: When the device detects that the preset steady-state thresholds are met for 5 consecutive seconds (period fluctuation coefficient <3%, peak variation coefficient <2%, baseline fluctuation <1.5 mmHg), it determines that it has entered the effective physiological steady-state calibration window.

[0020] Step 4: Baseline True Value Calculation: Remove random outlier sampling points within the window and use mean fitting to solve for the current accurate baseline null value.

[0021] Step 5, Smooth Baseline Update: The old and new baselines are smoothly replaced by a linear weighted transition within 1 second, without any baseline step jumps or sudden changes in blood pressure data.

[0022] Step 6, Adaptive Sensitivity Adjustment: The device fine-tunes the subsequent steady-state recognition trigger sensitivity based on the baseline drift amplitude over the past 3 hours, and automatically increases the calibration frequency when the drift rate is too high.

[0023] Step 7, Effect Verification: Actual testing showed that the equipment requires no manual intervention in daily use, the manual calibration cycle has been stably extended from 4 hours to more than 8 hours, and the effectiveness of 24-hour static monitoring data is ≥95%; it only triggers a slight abnormality prompt when the baseline deviation exceeds the limit threshold, ensuring monitoring accuracy and safety of use.

Claims

1. A method for steady-state baseline self-calibration of a wearable blood pressure monitoring device, characterized in that, A cuffless upper arm blood pressure monitoring device using a pure pressure array without additional auxiliary sensors includes the following steps: S1. Signal preprocessing: Collect raw signals of brachial artery pressure and pulse from multiple channels, and after temperature drift pre-correction and noise reduction filtering, output continuous and stable time-series pulse waveform data. S2. Steady-state feature extraction: Real-time extraction of five steady-state evaluation parameters: pulse cycle fluctuation coefficient, waveform peak variation coefficient, baseline amplitude fluctuation range, waveform integrity, and similarity between adjacent cycles. S3. Steady-state window determination: When all five steady-state characteristic parameters meet the preset threshold conditions within 5 consecutive seconds, it is determined that the body is entering the effective physiological steady-state calibration window for rest or sleep. S4. Baseline True Value Calculation: Within the effective steady-state calibration window, discrete outlier sampling points are eliminated, and the current true zero-point baseline value of the sensor is calculated using a mean fitting algorithm. S5. Smooth Baseline Update: A linear weighted transition method is used to complete the adaptive replacement of the old and new baselines. The weight of the old baseline gradually decreases and the weight of the new baseline gradually increases, so as to achieve a baseline update without jumps. S6. Sensitivity Adaptive Adjustment: Based on the historical baseline drift rate of the device, the steady-state recognition trigger sensitivity and calibration frequency are dynamically adjusted to achieve accurate calibration on demand; S7. Abnormal backup calibration: When the baseline offset exceeds the preset limit threshold, an abnormal prompt is triggered and the manual calibration channel is retained, achieving dual protection of fully automatic calibration and manual backup.

2. The steady-state baseline self-calibration method for wearable blood pressure monitoring devices according to claim 1, characterized in that, In step S3, the preset thresholds for steady-state determination are: pulse cycle fluctuation coefficient < 3%, peak variation coefficient < 2%, and baseline fluctuation amplitude < 1.5 mmHg.

3. The steady-state baseline self-calibration method for wearable blood pressure monitoring devices according to claim 1, characterized in that, In step S5, the duration of the baseline linear weighted transition update is 1 second.

4. The steady-state baseline self-calibration method for wearable blood pressure monitoring devices according to claim 1, characterized in that, In step S1, the sampling frequency of the pressure pulse signal is 200Hz.

5. The steady-state baseline self-calibration method for wearable blood pressure monitoring devices according to claim 1, characterized in that, The entire process does not rely on auxiliary data from acceleration, electrocardiogram, or photoelectric sensors; steady-state identification and baseline calibration are completed solely through the temporal characteristics of the pressure waveform itself.

6. The steady-state baseline self-calibration method for wearable blood pressure monitoring devices according to claim 1, characterized in that, After calibration using this method, the manual recalibration cycle of the equipment is extended from 4 hours to more than 8 hours.

7. The steady-state baseline self-calibration method for wearable blood pressure monitoring devices according to claim 1, characterized in that, The method is applicable to quasi-static 24-hour continuous blood pressure monitoring scenarios such as postoperative bed rest, nighttime sleep, and resting in the elderly with chronic diseases.

Images