Blood pressure continuous measurement abnormity early warning method and system based on AI data analysis

Abstract

The invention relates to the field of computer medical engineering, and discloses a blood pressure continuous measurement abnormity early warning method and system based on AI data analysis, and the method comprises the steps: collecting a pulse wave signal, triaxial accelerometer data, an ambient light intensity signal and a skin conductance signal through a sensor on a preset wearable device; generating multi-scale physiological features; systolic pressure and diastolic pressure are calculated, and blood vessel stiffness is calculated; calculating the main arteriosclerosis risk level; when the main arteriosclerosis risk level is higher than a preset risk level, a user is prompted through the wearable device. According to the invention, the pulse wave, the triaxial accelerometer, the ambient light intensity and the skin conductance signal are synchronously acquired and analyzed to obtain the systolic pressure, the diastolic pressure dynamic trend and the blood vessel stiffness, and the main arteriosclerosis risk level can be accurately quantified. In addition, a'physiological parameter-angiosclerosis-risk decision 'closed-loop management mechanism is established, and automatic and active health management of the user is realized.

Description

Technical Field

[0001] The present invention relates to the field of computer medical engineering, and more specifically, to a method and system for warning of abnormalities in continuous blood pressure measurement based on AI data analysis. Background Art

[0002] With the increasing incidence of cardiovascular diseases, continuous blood pressure monitoring and early warning of arteriosclerosis have become research hotspots for wearable medical devices. In existing technologies, blood pressure measurement mainly relies on intermittent cuff measurement techniques such as the Korotkoff sound method or the oscillometric method, which have defects such as limited measurement frequency and inability to reflect dynamic blood pressure changes. In recent years, wearable devices based on photoplethysmography (PPG) have attempted to estimate continuous blood pressure through pulse wave transit time, but their accuracy is easily affected by motion artifacts, individual differences and environmental factors, and lack in-depth analysis of vascular elasticity parameters.

[0003] Therefore, a new technical solution is needed that can accurately measure blood pressure and perform intelligent analysis for users in complex environments. Summary of the Invention

[0004] In order to solve the above technical problems, this application is proposed to provide a method and system for abnormal warning of continuous blood pressure measurement based on AI data analysis, which can accurately measure blood pressure for users in complex environments and perform intelligent analysis.

[0005] In a first aspect, the present invention provides a method for warning of abnormalities in continuous blood pressure measurement based on AI data analysis, comprising: collecting a user's pulse wave signal, three-axis accelerometer data, ambient light intensity signal, and skin conductance signal through sensors on a preset wearable device; generating multi-scale physiological characteristics of the user based on the user's pulse wave signal, three-axis accelerometer data, ambient light intensity signal, and skin conductance signal; calculating the user's systolic and diastolic blood pressures, and calculating the user's vascular stiffness based on the user's multi-scale physiological characteristics; calculating the user's aortic sclerosis risk level based on the user's systolic and diastolic blood pressures and vascular stiffness; and when the user's aortic sclerosis risk level is higher than a preset risk level, prompting the user through the wearable device.

[0006] Optionally, in the aforementioned blood pressure continuous measurement abnormality warning method based on AI data analysis, the user's multi-scale physiological characteristics are generated based on the user's pulse wave signal, three-axis accelerometer data, ambient light intensity signal and skin conductance signal, including: according to the preset first frequency interval, second frequency interval and third frequency interval, the user's pulse wave signal is decomposed into a first inherent modal function, a second inherent modal function and a third inherent modal function, wherein the frequency of any point in the first frequency interval is lower than the frequency of any point in the second frequency interval, and the frequency of any point in the second frequency interval is lower than the frequency of any point in the third frequency interval; based on the first inherent modal function, the user's electrocardiogram baseline characteristics are calculated; based on the second inherent modal function, the user's electromyography interference characteristics are calculated; based on the third inherent modal function, the user's microvascular reflection characteristics are calculated; based on the user's electrocardiogram baseline characteristics, myoelectric interference characteristics and microvascular reflection characteristics, the user's multi-scale physiological characteristics are generated.

[0007] Optionally, in the aforementioned blood pressure continuous measurement abnormality warning method based on AI data analysis, before decomposing the user's pulse wave signal into the first intrinsic mode function, the second intrinsic mode function, and the third intrinsic mode function according to the preset first frequency interval, the second frequency interval, and the third frequency interval, the method further includes: obtaining the user's pulse wave signal; setting the target function according to the user's pulse wave signal; Where x(t) represents the user's pulse wave signal, t represents time, ω k ,σ k is the center frequency and width of the kth frequency interval, is the first-order inverse of time t, δ(t) is the Dirac function, j is a preset imaginary number, * represents a convolution operation, F(x(t)) is the Fourier transform of the user's pulse wave signal x(t), and e is a preset natural constant; according to the objective function Determine the center frequency and width of the kth frequency interval when k is 1, 2, or 3; and determine the first frequency interval, the second frequency interval, and the third frequency interval based on the center frequency and width of the kth frequency interval.

[0008] Optionally, in the aforementioned blood pressure continuous measurement abnormality warning method based on AI data analysis, before decomposing the user's pulse wave signal into the first intrinsic modal function, the second intrinsic modal function, and the third intrinsic modal function according to the preset first frequency interval, the second frequency interval, and the third frequency interval, it also includes: calculating the user's movement amplitude according to the user's three-axis accelerometer data; judging whether the user's movement amplitude exceeds the preset amplitude; when the user's movement amplitude does not exceed the preset amplitude, entering the process of decomposing the user's pulse wave signal into the first intrinsic modal function, the second intrinsic modal function, and the third intrinsic modal function; when the user's movement amplitude exceeds the preset amplitude, filtering the user's pulse wave signal, and the processed user's pulse wave signal Wherein, · represents multiplication, x(t) represents the user's pulse wave signal, γ is a preset filter coefficient, Δt is the time interval for the sensor to collect the user's pulse wave signal, and a(t) is the user's three-axis accelerometer data. The processed user's pulse wave signal is then decomposed into the first intrinsic mode function, the second intrinsic mode function, and the third intrinsic mode function.

[0009] Optionally, in the aforementioned blood pressure continuous measurement abnormality warning method based on AI data analysis, the multi-scale physiological characteristics of the user are generated based on the user's pulse wave signal, three-axis accelerometer data, ambient light intensity signal and skin conductance signal, and also include: extracting the power spectral density peak in the z-axis direction from the user's three-axis accelerometer data as the user's motion noise characteristic; calculating the acceleration power spectrum entropy based on the user's three-axis accelerometer data as the user's dynamic and static distinguishing feature; when the user's motion amplitude exceeds the preset amplitude, extracting the waveform envelope entropy of the user's pulse wave signal as the user's arteriosclerosis indication feature; generating the user's multi-scale physiological characteristics based on the user's ECG baseline characteristics, myoelectric interference characteristics, and microvascular reflection characteristics, including: generating the user's multi-scale physiological characteristics based on the user's ECG baseline characteristics, myoelectric interference characteristics, microvascular reflection characteristics, motion noise characteristics, and arteriosclerosis indication characteristics.

[0010] Optionally, in the aforementioned blood pressure continuous measurement abnormality warning method based on AI data analysis, the multi-scale physiological characteristics of the user are generated based on the user's pulse wave signal, three-axis accelerometer data, ambient light intensity signal and skin conductance signal, and also include: calculating the flatness of the light intensity spectrum of the user's environment according to the user's ambient light intensity signal as the user's ambient light interference characteristic; calculating the user's skin conductance response peak and amplitude according to the user's skin conductance signal as the user's skin conductance characteristic; calculating the approximate entropy of the user's skin conductance signal as the user's pressure characteristic; generating the user's multi-scale physiological characteristics based on the user's ECG baseline characteristics, myoelectric interference characteristics, and microvascular reflection characteristics, including: generating the user's multi-scale physiological characteristics based on the user's ECG baseline characteristics, myoelectric interference characteristics, microvascular reflection characteristics, motion noise characteristics, arteriosclerosis indication characteristics, ambient light interference characteristics, skin conductance characteristics, and pressure characteristics.

[0011] Optionally, in the aforementioned blood pressure continuous measurement abnormality warning method based on AI data analysis, the systolic and diastolic blood pressures of the user are calculated based on the multi-scale physiological characteristics of the user, including: inputting the multi-scale physiological characteristics of the user into a trained neural network to obtain the systolic and diastolic blood pressures of the user, wherein the loss function of the neural network is the weighted sum of the Huber loss function and the DTW loss function.

[0012] Optionally, in the aforementioned method for abnormal early warning of continuous blood pressure measurement based on AI data analysis, the step of calculating the user's vascular stiffness includes: extracting an energy operator from the user's microvascular reflex characteristics. Wherein, s(t) represents the microvascular reflection characteristic of the user; the pulse wave propagation time PTT of the user is extracted from the three-axis accelerometer data of the user, and the blood vessel radius R(t) = α1PTT + α2 of the user is calculated based on the pulse wave propagation time PTT of the user, where α1 and α2 are clinical experience values; the vascular elastic modulus of the user is calculated based on the arteriosclerosis indicator characteristic ENT of the user Among them, E normal The elastic modulus of the conventional blood vessels is preset, ENT avg is the preset mean value of the arteriosclerosis indicator feature of the population; based on the motion noise feature PSD of the user, the user's vascular wall thickness h(t) = h0 + βPSD is calculated, where h0 is the preset reference thickness and β is the preset weight coefficient; the user's vascular stiffness is calculated Wherein, P represents the blood vessel wall pressure of the user, Equivalent to the energy operator The derivative of , · represents multiplication.

[0013] In a second aspect, the present invention provides a blood pressure continuous measurement abnormality warning system based on AI data analysis, comprising: an acquisition module, which acquires a user's pulse wave signal, three-axis accelerometer data, ambient light intensity signal, and skin conductance signal through sensors on a preset wearable device; a feature module, which generates multi-scale physiological characteristics of the user based on the user's pulse wave signal, three-axis accelerometer data, ambient light intensity signal, and skin conductance signal; a calculation module, which calculates the user's systolic and diastolic blood pressures, and the user's vascular stiffness based on the user's multi-scale physiological characteristics; an analysis module, which calculates the user's aortic sclerosis risk level based on the user's systolic and diastolic blood pressures and vascular stiffness; and a prompt module, which prompts the user through the wearable device when the user's aortic sclerosis risk level is higher than a preset risk level.

[0014] The above one or more technical solutions of the present invention have at least one or more of the following beneficial effects:

[0015] According to the technical solution of the present invention, by synchronously collecting and analyzing pulse waves, three-axis accelerometers, ambient light intensity and skin conductance signals, the dynamic trends of systolic and diastolic blood pressure and vascular stiffness are obtained, which can accurately quantify the risk level of aortic sclerosis. A closed-loop management mechanism of "physiological parameters-vascular sclerosis-risk decision-making" is established to achieve automated and proactive health management of users. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] The above and other purposes, features, and advantages of the present application will become more apparent through a more detailed description of the embodiments of the present application in conjunction with the accompanying drawings. The accompanying drawings are intended to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the drawings, the same reference numerals generally represent the same components or steps.

[0017] Figure 1 Flowchart of a method for warning abnormalities in continuous blood pressure measurement based on AI data analysis according to an embodiment of the present application;

[0018] Figure 2 This is a partial flow chart of a method for warning of abnormal continuous blood pressure measurement based on AI data analysis according to an embodiment of the present application;

[0019] Figure 3 This is another layout flow chart of the abnormal warning method for continuous blood pressure measurement based on AI data analysis according to an embodiment of the present application;

[0020] Figure 4This is another partial flowchart of the method for abnormal early warning of continuous blood pressure measurement based on AI data analysis according to an embodiment of the present application;

[0021] Figure 5 This is a block diagram of a blood pressure continuous measurement abnormality warning system based on AI data analysis according to an embodiment of the present application. DETAILED DESCRIPTION

[0022] Some embodiments of the present invention are described below with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are only used to explain the technical principles of the present invention and are not intended to limit the scope of protection of the present invention.

[0023] like Figure 1 As shown, one embodiment of the present invention provides a method for warning of abnormal continuous blood pressure measurement based on AI data analysis, comprising:

[0024] In step S110, the user's pulse wave signal, three-axis accelerometer data, ambient light intensity signal, and skin conductance signal are collected through sensors on the preset wearable device.

[0025] This embodiment overcomes the limitations of a single sensor by simultaneously collecting pulse wave, triaxial accelerometer, ambient light intensity, and skin conductance signals. The triaxial accelerometer can be used to dynamically compensate for motion artifacts, eliminating pulse wave signal distortion caused by user activity. Ambient light intensity data can be used to calibrate the photosensor baseline drift in real time, suppressing ambient light interference. Skin conductance signals reflect sympathetic nerve activity, helping to distinguish whether blood pressure fluctuations are due to physiological stress responses. Experiments have shown that compared with traditional single-modality PPG solutions, this embodiment reduces blood pressure calculation errors in dynamic scenarios such as walking and stair climbing. In this application, the data type can be image data, text data, or any other processable data type.

[0026] Step S120 , generating a multi-scale physiological feature of the user based on the user's pulse wave signal, triaxial accelerometer data, ambient light intensity signal, and skin conductance signal.

[0027] Step S130 , calculating the user's systolic and diastolic blood pressures, and the user's vascular stiffness based on the user's multi-scale physiological characteristics.

[0028] Step S140 , calculating the user's aortic sclerosis risk level based on the user's systolic blood pressure, diastolic blood pressure, and vascular stiffness.

[0029] In this embodiment, systolic blood pressure, diastolic blood pressure and vascular stiffness are key parameters for determining the risk level of aortic sclerosis. Based on the above parameters, the user's risk level of aortic sclerosis can be accurately determined.

[0030] Step S150: When the user's aortic sclerosis risk level is higher than a preset risk level, the user is prompted via the wearable device.

[0031] In this embodiment, when the risk value exceeds a preset threshold, the wearable device triggers a multi-level alert through vibration, LED flashing, and app push notifications, guiding the user to seek medical attention or make lifestyle adjustments promptly. Clinical testing has shown that the technical solution of this embodiment significantly improves the detection rate of aortic sclerosis compared to traditional arm blood pressure monitors, with a significantly reduced false alarm rate. If the aortic sclerosis risk level exceeds the preset risk level for multiple consecutive monitoring sessions, the wearable device provides the user with medication information and calculates the time and dosage for medication based on the specific values ​​of systolic and diastolic blood pressure and vascular stiffness. The wearable device also activates heart rate monitoring and issues a warning message if the heart rate exceeds a preset upper limit, preventing patients with aortic sclerosis from engaging in strenuous exercise that could harm their health.

[0032] According to the technical solution of this embodiment, by integrating the dynamic trends of systolic and diastolic blood pressure with vascular stiffness, the risk level of aortic sclerosis can be accurately quantified, and a closed-loop management mechanism of "physiological parameters-vascular sclerosis-risk decision-making" is established to achieve automated and proactive health management of users.

[0033] like Figure 2 As shown, one embodiment of the present invention provides a method for warning abnormalities in continuous blood pressure measurement based on AI data analysis. Compared with the previous embodiment, the method for warning abnormalities in continuous blood pressure measurement based on AI data analysis in this embodiment includes step S120:

[0034] Step S210: Acquire the user's pulse wave signal.

[0035] Step S220: Setting the target function based on the user's pulse wave signal Where x(t) represents the user's pulse wave signal, t represents time, and ω k ,σ k is the center frequency and width of the kth frequency interval, is the first-order inverse of time t, δ(t) is the Dirac function, j is a preset imaginary number, * represents a convolution operation, F(x(t)) is the Fourier transform of the user's pulse wave signal x(t), and e is a preset natural constant.

[0036] In this embodiment, The Hilbert transform kernel is used to construct the analytical signal. The objective function projects the signal into the frequency domain through Fourier transform and uses the energy concentration characteristics of the Hilbert spectrum and the Dirac function to minimize energy leakage in each frequency interval and avoid cross-interference (modal aliasing) between the characteristics of different frequency intervals.

[0037] Step S230, according to the objective function

[0038] Determine the center frequency and width of the kth frequency interval when k is 1, 2, or 3.

[0039] Step S240: Determine a first frequency interval, a second frequency interval, and a third frequency interval according to the center frequency and width of the kth frequency interval.

[0040] Step S250: Decompose the user's pulse wave signal into a first intrinsic mode function, a second intrinsic mode function, and a third intrinsic mode function according to the preset first frequency interval, second frequency interval, and third frequency interval, wherein the frequency of any point in the first frequency interval is lower than the frequency of any point in the second frequency interval, and the frequency of any point in the second frequency interval is lower than the frequency of any point in the third frequency interval.

[0041] In this embodiment, variational mode decomposition (VMD) is used to decompose the pulse wave signal.

[0042] Step S260: Calculate the user's ECG baseline characteristics based on the first intrinsic mode function.

[0043] Step S270: Calculate the user's electromyographic interference characteristics based on the second intrinsic mode function.

[0044] Step S280: Calculate the user's microvascular reflection characteristics based on the third intrinsic mode function.

[0045] In this embodiment, the pulse wave is decomposed into three characteristics: low frequency (ECG baseline), medium frequency (myographic interference), and high frequency (microvascular reflex). The low frequency band (typically 0.01-0.5Hz) reflects the long-term regulation of blood pressure by the autonomic nervous system and is nonlinearly negatively correlated with arterial stiffness. The medium frequency band (typically 0.5-5Hz) dynamically calculates systolic blood pressure, which helps reduce calculation errors. The high frequency band (typically 5-20Hz) reflects capillary resistance and helps indicate capillary microcirculatory disorders.

[0046] Step S290 , generating a multi-scale physiological feature of the user based on the user's ECG baseline feature, myoelectric interference feature, and microvascular reflex feature.

[0047] According to the technical solution of this embodiment, the pulse wave signal is decomposed into the first intrinsic mode function, the second intrinsic mode function, and the third intrinsic mode function, and three characteristics of low frequency (ECG baseline), medium frequency (myoelectric interference), and high frequency (microvascular reflex) are calculated and spliced ​​together to form the user's multi-scale physiological characteristics, which is conducive to accurately identifying the user's blood pressure condition.

[0048] like Figure 3As shown, one embodiment of the present invention provides a method for warning of abnormal continuous blood pressure measurement based on AI data analysis. Compared with the previous embodiment, the method for warning of abnormal continuous blood pressure measurement based on AI data analysis in this embodiment further includes, before step S250:

[0049] Step S310 : Calculating the user's movement amplitude based on the user's three-axis accelerometer data.

[0050] Step S320: Determine whether the user's movement range exceeds a preset range.

[0051] Step S330: When the user's movement range does not exceed the preset range, proceed to step S250.

[0052] In this embodiment, when the movement amplitude does not exceed the standard, a decomposition mechanism driven by an objective function is adopted to retain the multi-scale physiological characteristics of the pulse wave (ECG baseline / myoelectric interference / microvascular reflex).

[0053] Step S340: When the user's movement amplitude exceeds a preset amplitude, the user's pulse wave signal is filtered. The processed user's pulse wave signal Wherein, x(t) represents the user's pulse wave signal, γ is a preset filter coefficient, Δt is the time interval for the sensor to collect the user's pulse wave signal, a(t) is the user's three-axis accelerometer data, and · represents multiplication, and then proceeds to step S250.

[0054] In this embodiment, the formula It corresponds to the instantaneous rate of change of the acceleration signal and achieves a higher signal-to-noise ratio than traditional low-pass filters by minimizing the contamination of the pulse wave by motion noise.

[0055] The technical solution of this embodiment directly inverts the time-frequency characteristics of motion interference using accelerometer data, enabling precise separation of myoelectric interference from microvascular reflexes in motion scenarios, preserving a pure ECG baseline. Clinical testing has shown that when myoelectric interference intensity exceeds 30μV, the technical solution of this embodiment can maintain a systolic blood pressure calculation error of less than ±5mmHg, significantly outperforming solutions that rely solely on hardware filtering.

[0056] One embodiment of the present invention provides a method for warning of abnormal continuous blood pressure measurement based on AI data analysis. Compared to the previous embodiment, the method for warning of abnormal continuous blood pressure measurement based on AI data analysis in this embodiment further includes, in step S120:

[0057] (1) The power spectrum density peak in the z-axis direction is extracted from the user's three-axis accelerometer data as the user's motion noise feature.

[0058] In this embodiment, by analyzing the power spectrum density peak in the z direction (vertical direction) of the three-axis accelerometer, it is helpful to accurately locate the frequency characteristics of periodic movements such as walking / running. Compared with traditional full-axis power spectrum analysis, the false alarm rate of motion noise identification is significantly reduced.

[0059] (2) Calculate the acceleration power spectrum entropy based on the user's three-axis accelerometer data as the user's dynamic and static distinguishing feature.

[0060] In this embodiment, the acceleration power spectrum entropy can be used as a criterion for distinguishing whether the user is dynamic or static. For example, if the acceleration power spectrum entropy exceeds 1.2, it can be determined that the user is in a static scene, and if it is lower than 1.2, it can be determined that the user is walking or running.

[0061] (3) When the user's movement amplitude exceeds a preset amplitude, the waveform envelope entropy of the user's pulse wave signal is extracted as an arteriosclerosis indication feature of the user.

[0062] In this embodiment, the characteristics of arteriosclerosis are more obvious when the user is in motion. The envelope of arteriosclerosis patients becomes smoother due to decreased vascular elasticity. Therefore, the waveform envelope entropy of the pulse wave signal is extracted as the user's arteriosclerosis indication feature.

[0063] Step S130 includes generating multi-scale physiological features of the user based on the user's electrocardiogram baseline features, myoelectric interference features, microvascular reflection features, motion noise features, and arteriosclerosis indication features.

[0064] According to the technical solution of this embodiment, the multi-scale physiological features generated based on ECG baseline features, myoelectric interference features, microvascular reflection features, motion noise features, and arteriosclerosis indication features are conducive to accurately identifying the user's blood pressure condition.

[0065] One embodiment of the present invention provides a method for warning of abnormal continuous blood pressure measurement based on AI data analysis. Compared to the previous embodiment, the method for warning of abnormal continuous blood pressure measurement based on AI data analysis in this embodiment further includes, in step S120:

[0066] (1) Based on the user's ambient light intensity signal, the flatness of the light intensity spectrum of the user's environment is calculated as the user's ambient light interference feature.

[0067] In this embodiment, the flatness of the light intensity spectrum is used as a feature, and the effect of light intensity fluctuation on the baseline drift of the signal can be eliminated during the analysis process by quantifying the ambient light frequency domain fluctuation.

[0068] (2) Based on the user's skin conductance signal, the user's skin conductance response peak value and amplitude are calculated as the user's skin conductance characteristics.

[0069] In this embodiment, the peak value and amplitude of the skin conductance response reflect the intensity of sympathetic nerve activation and are positively correlated with arterial stiffness.

[0070] (3) Calculate the approximate entropy of the user's skin conductance signal as the user's stress feature.

[0071] In this embodiment, the approximate entropy of the skin conductance signal reflects the degree of primary nerve disorder of the user, and the degree of primary nerve disorder of patients with arteriosclerosis is often higher.

[0072] Step S130 includes generating multi-scale physiological characteristics of the user based on the user's electrocardiogram baseline characteristics, myoelectric interference characteristics, microvascular reflection characteristics, motion noise characteristics, arteriosclerosis indication characteristics, ambient light interference characteristics, skin conductance characteristics, and pressure characteristics.

[0073] According to the technical solution of this embodiment, the multi-scale physiological characteristics generated based on the user's electrocardiogram baseline characteristics, myoelectric interference characteristics, microvascular reflection characteristics, motion noise characteristics, arteriosclerosis indication characteristics, ambient light interference characteristics, skin conductance characteristics, and pressure characteristics are conducive to accurately identifying the user's blood pressure condition.

[0074] One embodiment of the present invention provides a method for warning of abnormal continuous blood pressure measurement based on AI data analysis. Compared to the previous embodiment, the method for warning of abnormal continuous blood pressure measurement based on AI data analysis in this embodiment includes step S140:

[0075] The user's multi-scale physiological features are input into a trained neural network to obtain the user's systolic and diastolic blood pressures. The loss function of the neural network is the weighted sum of the Huber loss function and the DTW loss function.

[0076] In this embodiment, the Huber loss function uses piecewise square / linear penalty for outliers (such as sudden changes in pulse wave spikes caused by motion artifacts), which is beneficial to reducing the systolic blood pressure prediction error. The DTW loss function is beneficial to aligning the pulse wave timing features of different users and solving the feature misalignment problem caused by inconsistent time steps in traditional neural networks.

[0077] like Figure 4 As shown, one embodiment of the present invention provides a method for warning of abnormal continuous blood pressure measurement based on AI data analysis. Compared with the previous embodiment, the method for warning of abnormal continuous blood pressure measurement based on AI data analysis in this embodiment includes step S140:

[0078] Step S410: extracting energy operators from the user's microvascular reflection features Where s(t) represents the user's microvascular reflex feature, and · represents multiplication.

[0079] In this embodiment, an energy operator is calculated through nonlinear differential operation to capture the transient energy changes of microvascular reflection waves, thereby improving the sensitivity to local fluctuations in arterial stiffness compared to traditional PWV schemes.

[0080] Step S420 , extracting the user's pulse wave propagation time PTT from the user's triaxial accelerometer data, and calculating the user's blood vessel radius R(t)=α1PTT+α2 according to the user's pulse wave propagation time PTT, where α1 and α2 are clinical experience values.

[0081] In this embodiment, the triaxial accelerometer data is converted into dynamic changes in blood vessel radius, which is helpful in eliminating manual measurement errors of traditional palpation methods.

[0082] Step S430: Calculate the user's vascular elastic modulus based on the user's arteriosclerosis indicator feature ENT Among them, E normal The elastic modulus of the conventional blood vessels is preset, ENT avg It is the mean value of the preset population arteriosclerosis indicator characteristics.

[0083] In this embodiment, dynamic adjustment of the elastic modulus is beneficial for sensitively detecting dynamic changes in arterial stiffness.

[0084] Step S440 , calculating the user's blood vessel wall thickness h(t)=h0+βPSD based on the user's motion noise feature PSD, where h0 is a preset reference thickness and β is a preset weight coefficient.

[0085] In this embodiment, when the motion noise is high, the β coefficient is dynamically adjusted through Kalman filtering, which can effectively suppress the contamination of the h(t) calculation caused by the high-frequency vibration of the accelerometer.

[0086] Step S450: Calculate the user's vascular stiffness Where P represents the user's blood vessel wall pressure, Equivalent to the energy operator The derivative of , · represents multiplication.

[0087] According to the technical solution of this embodiment, the vascular stiffness formula implicitly incorporates the nonlinear coupling relationship between arterial stiffness, microvascular reflected energy, and vascular radius. Compared with traditional models, it is more consistent with hemodynamic laws and can more accurately calculate the user's vascular stiffness.

[0088] like Figure 5 As shown, one embodiment of the present invention provides a blood pressure continuous measurement abnormality warning system based on AI data analysis, including:

[0089] The acquisition module 510 collects the user's pulse wave signal, three-axis accelerometer data, ambient light intensity signal and skin conductance signal through the sensors on the preset wearable device.

[0090] This embodiment overcomes the limitations of single sensors by simultaneously collecting pulse wave, triaxial accelerometer, ambient light intensity, and skin conductance signals. The triaxial accelerometer dynamically compensates for motion artifacts, eliminating pulse wave signal distortion caused by user activity. Ambient light intensity data is used to calibrate the photoelectric sensor baseline drift in real time, suppressing ambient light interference. Skin conductance signals reflect sympathetic nerve activity, helping to distinguish whether blood pressure fluctuations are due to physiological stress responses. Experiments have shown that compared with traditional single-modality PPG solutions, this embodiment significantly reduces blood pressure calculation errors in dynamic scenarios such as walking and stair climbing.

[0091] The feature module 520 generates a multi-scale physiological feature of the user based on the user's pulse wave signal, triaxial accelerometer data, ambient light intensity signal, and skin conductance signal.

[0092] The calculation module 530 calculates the user's systolic and diastolic blood pressures, and the user's vascular stiffness based on the user's multi-scale physiological characteristics.

[0093] The analysis module 540 calculates the user's aortic sclerosis risk level based on the user's systolic blood pressure, diastolic blood pressure, and vascular stiffness.

[0094] In this embodiment, systolic blood pressure, diastolic blood pressure and vascular stiffness are key parameters for determining the risk level of aortic sclerosis. Based on the above parameters, the user's risk level of aortic sclerosis can be accurately determined.

[0095] The prompt module 550 prompts the user through the wearable device when the user's aortic sclerosis risk level is higher than a preset risk level.

[0096] In this embodiment, when the risk value exceeds a preset threshold, the wearable device generates a multi-level alert through vibration, LED flashing, and app notifications, guiding the user to seek medical attention or make lifestyle adjustments. Clinical testing has shown that this embodiment's technical solution significantly improves the detection rate of aortic sclerosis compared to traditional arm blood pressure monitors, while significantly reducing the false alarm rate.

[0097] According to the technical solution of this embodiment, by integrating the dynamic trends of systolic and diastolic blood pressure with vascular stiffness, the risk level of aortic sclerosis can be accurately quantified, and a closed-loop management mechanism of "physiological parameters-vascular sclerosis-risk decision-making" is established to achieve automated and proactive health management of users.

[0098] The basic principles of the present application have been described above in conjunction with specific embodiments. However, it should be noted that the advantages, strengths, and effects mentioned in this application are merely illustrative and not restrictive, and it should not be assumed that these advantages, strengths, and effects are required of each embodiment of this application. In addition, the specific details disclosed above are merely illustrative and facilitating understanding, and are not restrictive. The above details do not limit this application to necessarily being implemented using the above specific details.

[0099] The block diagrams of the devices, devices, equipment, and systems involved in this application are merely illustrative examples and are not intended to require or imply that they must be connected, arranged, or configured in the manner shown in the block diagrams. As will be appreciated by those skilled in the art, these devices, devices, equipment, and systems can be connected, arranged, or configured in any manner. Words such as "include," "comprise," "have," and the like are open-ended words, meaning "including but not limited to," and can be used interchangeably therewith. The words "or" and "and" used herein refer to the words "and / or" and can be used interchangeably therewith, unless the context clearly indicates otherwise. The word "such as" used herein refers to the phrase "such as but not limited to," and can be used interchangeably therewith.

[0100] It should also be noted that in the apparatus, device, and method of the present application, each component or each step can be decomposed and / or recombined, and such decomposition and / or recombination should be regarded as equivalent solutions of the present application.

[0101] The above description of the disclosed aspects is provided to enable any person skilled in the art to make or use the present application. Various modifications to these aspects will be readily apparent to those skilled in the art, and the general principles defined herein may be applied to other aspects without departing from the scope of the present application. Therefore, the present application is not intended to be limited to the aspects shown herein, but rather to be accorded the widest scope consistent with the principles and novel features disclosed herein.

[0102] The above description has been provided for the purpose of illustration and description. Furthermore, this description is not intended to limit the embodiments of the present application to the forms disclosed herein. Although a number of example aspects and embodiments have been discussed above, those skilled in the art will recognize certain variations, modifications, alterations, additions, and sub-combinations thereof.

Claims

1. A method for warning of abnormalities in continuous blood pressure measurement based on AI data analysis, including: The user's pulse wave signal, three-axis accelerometer data, ambient light intensity signal, and skin conductance signal are collected through the sensors on the preset wearable device; generating a multi-scale physiological feature of the user based on the user's pulse wave signal, triaxial accelerometer data, ambient light intensity signal, and skin conductance signal; calculating the user's systolic and diastolic blood pressures, and calculating the user's vascular stiffness based on the user's multi-scale physiological characteristics; calculating the user's aortic sclerosis risk level based on the user's systolic blood pressure, diastolic blood pressure, and vascular stiffness; When the user's aortic sclerosis risk level is higher than a preset risk level, the user is prompted via the wearable device.

2. The method for abnormal early warning of continuous blood pressure measurement based on AI data analysis according to claim 1, wherein: Generate a multi-scale physiological feature of the user based on the user's pulse wave signal, triaxial accelerometer data, ambient light intensity signal, and skin conductance signal, including: Decomposing the user's pulse wave signal into a first intrinsic mode function, a second intrinsic mode function, and a third intrinsic mode function according to a preset first frequency interval, a second frequency interval, and a third frequency interval, wherein the frequency of any point in the first frequency interval is lower than the frequency of any point in the second frequency interval, and the frequency of any point in the second frequency interval is lower than the frequency of any point in the third frequency interval; Calculating an electrocardiogram baseline characteristic of the user based on the first intrinsic mode function; Calculating a myoelectric interference feature of the user based on the second intrinsic mode function; calculating a microvascular reflex characteristic of the user based on the third intrinsic mode function; Based on the user's electrocardiogram baseline characteristics, myoelectric interference characteristics, and microvascular reflection characteristics, a multi-scale physiological characteristic of the user is generated.

3. The method for abnormal early warning of continuous blood pressure measurement based on AI data analysis according to claim 2, wherein: Before decomposing the user's pulse wave signal into the first intrinsic mode function, the second intrinsic mode function, and the third intrinsic mode function according to the preset first frequency interval, the second frequency interval, and the third frequency interval, the method further includes: obtaining a pulse wave signal of the user; Setting an objective function according to the user's pulse wave signal Where x(t) represents the user's pulse wave signal, t represents time, ω k ,σ k is the center frequency and width of the kth frequency interval, is the first-order inverse of time t, δ(t) is the Dirac function, j is a preset imaginary number, * represents a convolution operation, F(x(t)) is the Fourier transform of the user's pulse wave signal x(t), and e is a preset natural constant; According to the objective function Determine the center frequency and width of the kth frequency interval when k is 1, 2, or 3; The first frequency interval, the second frequency interval, and the third frequency interval are determined according to the center frequency and width of the kth frequency interval.

4. The method for abnormal early warning of continuous blood pressure measurement based on AI data analysis according to claim 2, wherein: Before decomposing the user's pulse wave signal into the first intrinsic mode function, the second intrinsic mode function, and the third intrinsic mode function according to the preset first frequency interval, the second frequency interval, and the third frequency interval, the method further includes: Calculating the user's movement amplitude based on the user's three-axis accelerometer data; Determining whether the user's movement amplitude exceeds a preset amplitude; When the user's movement amplitude does not exceed the preset amplitude, decomposing the user's pulse wave signal into a first intrinsic mode function, a second intrinsic mode function, and a third intrinsic mode function; When the user's movement amplitude exceeds the preset amplitude, the user's pulse wave signal is filtered and the processed user's pulse wave signal is filtered. Wherein, · represents multiplication, x(t) represents the user's pulse wave signal, γ is a preset filter coefficient, Δt is the time interval for the sensor to collect the user's pulse wave signal, and a(t) is the user's three-axis accelerometer data. The processed user's pulse wave signal is then decomposed into a first intrinsic mode function, a second intrinsic mode function, and a third intrinsic mode function.

5. The method for abnormal early warning of continuous blood pressure measurement based on AI data analysis according to claim 4, wherein: Generating a multi-scale physiological feature of the user based on the user's pulse wave signal, triaxial accelerometer data, ambient light intensity signal, and skin conductance signal, further comprising: Extracting a power spectral density peak in the z-axis direction from the three-axis accelerometer data of the user as a motion noise feature of the user; Calculating acceleration power spectrum entropy based on the three-axis accelerometer data of the user as a dynamic and static distinguishing feature of the user; When the user's movement amplitude exceeds the preset amplitude, extracting the waveform envelope entropy of the user's pulse wave signal as an arteriosclerosis indication feature of the user; Generate multi-scale physiological features of the user based on the user's electrocardiogram baseline features, myoelectric interference features, and microvascular reflex features, including: Multi-scale physiological characteristics of the user are generated based on the user's electrocardiogram baseline characteristics, myoelectric interference characteristics, microvascular reflection characteristics, motion noise characteristics, and arteriosclerosis indication characteristics.

6. The method for abnormal early warning of continuous blood pressure measurement based on AI data analysis according to claim 5, wherein: Generating a multi-scale physiological feature of the user based on the user's pulse wave signal, triaxial accelerometer data, ambient light intensity signal, and skin conductance signal, further comprising: Calculating, based on the ambient light intensity signal of the user, the flatness of the light intensity spectrum of the user's environment as the ambient light interference feature of the user; Calculating, according to the user's skin conductance signal, a skin conductance response peak value and an amplitude of the user as a skin conductance feature of the user; calculating an approximate entropy of a skin conductance signal of the user as a stress feature of the user; Generate multi-scale physiological features of the user based on the user's electrocardiogram baseline features, myoelectric interference features, and microvascular reflex features, including: Multi-scale physiological characteristics of the user are generated based on the user's electrocardiogram baseline characteristics, myoelectric interference characteristics, microvascular reflection characteristics, motion noise characteristics, arteriosclerosis indication characteristics, ambient light interference characteristics, skin conductance characteristics, and pressure characteristics.

7. The method for early warning of abnormal continuous blood pressure measurement based on AI data analysis according to claim 6, wherein: Calculating the systolic and diastolic blood pressures of the user according to the multi-scale physiological characteristics of the user, including: The multi-scale physiological features of the user are input into a trained neural network to obtain the systolic and diastolic blood pressures of the user, wherein the loss function of the neural network is a weighted sum of a Huber loss function and a DTW loss function.

8. The method for early warning of abnormal continuous blood pressure measurement based on AI data analysis according to claim 6, wherein: Calculating the user's vascular stiffness, including: Extracting an energy operator from the user's microvascular reflection characteristics Wherein, s(t) represents the microvascular reflex characteristics of the user; Extracting the user's pulse wave propagation time PTT from the user's triaxial accelerometer data, and calculating the user's blood vessel radius R(t)=α1PTT+α2 based on the user's pulse wave propagation time PTT, where α1 and α2 are clinical experience values; Calculate the user's vascular elastic modulus based on the user's arteriosclerosis indicator feature ENT Among them, E normal The elastic modulus of the conventional blood vessels is preset, ENT avg It is the mean value of the pre-set population arteriosclerosis indicator characteristics; Calculating the user's blood vessel wall thickness h(t)=h0+βPSD based on the user's motion noise feature PSD, where h0 is a preset reference thickness and β is a preset weight coefficient; Calculating the user's vascular stiffness Wherein, P represents the blood vessel wall pressure of the user, Equivalent to the energy operator The derivative of , · represents multiplication.

9. A continuous blood pressure measurement abnormality warning system based on AI data analysis, including: The acquisition module collects the user's pulse wave signal, three-axis accelerometer data, ambient light intensity signal and skin conductance signal through the sensors on the preset wearable device; a feature module, generating a multi-scale physiological feature of the user based on the user's pulse wave signal, triaxial accelerometer data, ambient light intensity signal, and skin conductance signal; a calculation module, configured to calculate the user's systolic and diastolic blood pressures, and the user's vascular stiffness, based on the user's multi-scale physiological characteristics; an analysis module, calculating the user's aortic sclerosis risk level based on the user's systolic blood pressure, diastolic blood pressure, and vascular stiffness; The prompt module prompts the user through the wearable device when the user's aortic sclerosis risk level is higher than a preset risk level.

Images