Continuous pulse signal processing and cardiac function evaluation system

Abstract

The invention provides a continuous pulse signal processing and cardiac function evaluation system, which comprises a data acquisition module, a preprocessing module, a signal evaluation module, a feature extraction module and a cardiac function evaluation module, and is characterized in that a sample data set is established by acquiring a sample of a continuous pulse signal and evaluating the cardiac function of the sample; a noise threshold is set for continuous pulse signals in a sample data set for filtering and noise reduction processing, normalization processing is carried out, a preprocessed data set is formed, the pulse quality score of each continuous pulse signal is calculated according to the amplitude and the power spectral density of the continuous pulse signals, the continuous pulse signals are screened according to the pulse quality scores, and the continuous pulse signals are obtained. Time domain feature extraction and frequency domain feature extraction are conducted on the screened continuous pulse signals, a cardiac function evaluation model is established according to a neural network and trained, the continuous pulse signals collected by the collection module, the age and the gender serve as input of the evaluation model, and cardiac function evaluation is completed.

Description

Technical Field

[0001] The invention relates to the technical field of cardiac function assessment, and in particular to a continuous pulse signal processing and cardiac function assessment system. Background Art

[0002] Cardiac diseases have become the leading cause of death worldwide, posing a threat to human health worldwide and a serious public health issue. With the development of social economy, people's lifestyles have undergone tremendous changes. With the aging of the population and the acceleration of urbanization, the problem of unhealthy lifestyles of residents has become increasingly prominent, and the prevalence and incidence of heart disease have continued to increase. It is necessary and urgent to strengthen the monitoring and evaluation of the health status of heart function.

[0003] Continuous pulse signal monitoring is a key technology in the monitoring and evaluation of cardiac function. Traditional cardiac function evaluation methods rely on intermittent electrocardiograms and echocardiograms. Although these methods can provide important cardiac health information, they cannot achieve long-term and continuous cardiac monitoring. How to analyze the pathological information in the continuous pulse signal and provide more accurate and real-time functional health status evaluation has become an urgent problem to be solved.

[0004] In the prior art, publication number CN106539562A discloses a method for evaluating cardiac function by pulse wave, comprising the following steps: (1) obtaining a fingertip photoelectric volumetric pulse wave signal of a human body using a photoelectric volumetric pulse wave sensor; (2) amplifying and filtering the fingertip photoelectric volumetric pulse wave signal; (3) separating an alternating current; (4) performing analog-to-digital conversion on the alternating current, and resampling the converted digital signal to obtain a pulse wave waveform; (5) determining a pulse rate PR using characteristic points on the pulse wave waveform; (6) segmenting the signal using the ascending branch starting point of the pulse wave waveform, and finally obtaining waveform characteristic parameters reflecting the cardiovascular resistance and microvascular microcirculation state of the human body, and then obtaining parameters reflecting cardiac function: cardiac output, cardiac output index, stroke volume per stroke, and heart rate index. However, although this method amplifies and filters the pulse signal, the operation is too simple to effectively eliminate the signal quality problems caused by the environment and equipment during signal acquisition. At the same time, the analysis of cardiac function based on the characteristics of the pulse wave waveform cannot capture the maintenance details in the pulse signal in detail, which may cause deviations in the judgment of cardiac function, and does not have the ability to automatically learn and adapt to the data characteristics of different patients.

[0005] The above information disclosed in this Background section is only for enhancement of understanding of the background of the present disclosure and therefore it may contain information that does not constitute the prior art that is already known to one of ordinary skill in the art. Summary of the invention

[0006] The purpose of the present invention is to provide a continuous pulse signal processing and cardiac function assessment system to solve the problems raised in the above background technology.

[0007] To achieve the above object, the present invention provides the following technical solutions:

[0008] A continuous pulse signal processing and cardiac function assessment system, comprising:

[0009] A data acquisition module, wherein the data acquisition module obtains continuous pulse signal samples of samplers of different ages and genders, evaluates the cardiac function of each person, establishes a sample data set, and maps the continuous pulse signal, age, gender and timestamp in the sample data set one by one, and simultaneously obtains the continuous pulse signal and age and gender information of the person to be evaluated;

[0010] A preprocessing module, wherein the preprocessing module analyzes the noise characteristics of the continuous pulse signal in the sample data set, sets a noise threshold to perform filtering and noise reduction processing, filters out the noise caused by equipment collection, and then normalizes the continuous pulse signal after the noise reduction processing to form a preprocessing data set;

[0011] A signal evaluation module, the signal evaluation module is used to obtain the amplitude and power spectrum density of the continuous pulse signal of the preprocessing data set, calculate the average value and variance of the amplitude of the continuous pulse signal, and the median value of the low-high frequency ratio, calculate the pulse quality score of each continuous pulse signal, screen the continuous pulse signal according to the pulse quality score, and use the screened continuous pulse signal as the analysis data set;

[0012] A feature extraction module, which is used to perform time domain feature analysis on the continuous pulse signal in the analysis data set, and then extract the frequency domain features of the pulse signal by fast Fourier transform, and form a feature data set with the time domain features and the frequency domain features;

[0013] A cardiac function assessment module is used to establish a cardiac function assessment model through a neural network, and the model is trained by taking the characteristic data set of continuous pulse signals and age and gender as the input of the model and taking the cardiac function assessment result as the output of the model. The pulse signal to be assessed and the age and gender collected by the acquisition module are input into the trained assessment model to complete the assessment of cardiac function.

[0014] Furthermore, the specific steps of performing filtering and noise reduction processing by setting a noise threshold are:

[0015] Firstly, the continuous pulse signal is transformed by wavelet transform and decomposed into a series of approximate coefficients and detail coefficients with different frequencies;

[0016]

[0017] Where X(t) is the continuous pulse signal in the sample data set, c j,k is the approximate coefficient, d j,k is the detail coefficient, is the scale function, is the wavelet function, t is the acquisition time point of the continuous pulse signal, j and k are the scale parameter and displacement parameter respectively;

[0018] By setting the noise threshold to process the wavelet coefficients, the detail coefficients affected by noise can be removed;

[0019]

[0020] Among them, d′ j,k is the detail coefficient after denoising, τ is the set noise threshold;

[0021] Use the processed wavelet coefficients to perform inverse wavelet transform and reconstruct the denoised signal;

[0022]

[0023] Among them, X ′ (t) is the continuous pulse signal of the sample after denoising.

[0024] Furthermore, the calculation formula for normalizing the continuous pulse signal after noise reduction is:

[0025]

[0026] Among them, X is the normalized pulse signal, X(t) is the denoised pulse signal, and x min is the minimum value in the original pulse signal data, X max is the maximum value in the original pulse signal data.

[0027] Furthermore, the calculation formula for the mean value and variance of the amplitude of the continuous pulse signal is:

[0028]

[0029] Among them, X fic is the amplitude of the icth main peak wave of the continuous pulse signal, nc is the number of main peak waves of the continuous pulse signal, μ c is the average amplitude of the main peak wave of the continuous pulse signal, σ c is the variance of the amplitude of the main peak wave of the continuous pulse signal;

[0030] The low-high frequency ratio median calculation formula is:

[0031]

[0032] Where q is the median of the low-high frequency ratio, P(g) is the power spectrum density at frequency g, α1 and α2 are the frequency ranges of the continuous pulse signal, and g1 and β2 are the low-frequency ranges of the continuous pulse signal;

[0033] The calculation formula of the continuous pulse quality score is:

[0034] PF=γ1*μ c +γ2*σ c +γ1*q

[0035] Among them, PF is the continuous pulse quality score, γ1, γ2, and γ3 are the average value, variance, and proportion of the median low-high frequency ratio of the amplitude of the continuous pulse signal to the quality score.

[0036] Furthermore, the process of the time domain feature analysis is as follows: by segmenting the continuous pulse signal into periods, obtaining the time domain features of the pulse signal of each period, setting a window of constant width according to the length and characteristics of the pulse signal, placing the new sampling point at the highest position of the window, calculating the average value and standard deviation of the continuous pulse signal period, and obtaining the segmented period of the continuous pulse signal through the average value and standard deviation of the period. The specific calculation formula is as follows;

[0037]

[0038] Where A(i) is the cycle mean of the continuous pulse signal, B(i) is the cycle standard deviation of the continuous pulse signal, w is the width of the window set to be constant, Xq(tk) is the continuous pulse signal in the analysis sample, i is the end point position of the window, tk is the signal of the tkth point in the window, and CV is the segmentation period of the continuous pulse signal;

[0039] The specific formula for extracting the frequency domain characteristics of the pulse signal by fast Fourier transform is:

[0040]

[0041] Among them, VS(m) is the mth component in the frequency domain signal of the periodic pulse signal, Xq(in) is the inth sample in the continuous pulse signal, N is the total number of samples, m represents the mth frequency component in the frequency domain, 0≤m≤N-1, and j is an imaginary unit.

[0042] Furthermore, the specific method of establishing the cardiac function assessment model through the neural network is:

[0043] The cardiac function assessment model established by the neural network is based on a long short-term memory neural network, and the neural network model includes an input gate, a forget gate and an output gate;

[0044] The equation of the forget gate is:

[0045] f t =δ(W f ·[h t―1 ,x t ]+b f )

[0046] Among them, f t is the current switch state of the forget gate, δ represents the sigmoid function, W f is the forget gate weight, b f is the forget gate bias parameter, h t―1 is the hidden state at the previous moment, x t Input switch for current data;

[0047] Determine how much of the unit state at the previous moment needs to be retained to the current moment;

[0048] The equation for the input gate is:

[0049] i t =δ(W i ·[h t―1 ,x t ]+b i )

[0050] Among them, i t is the current switch state of the input gate, δ represents the sigmoid function, W i is the input gate weight, b i is the input gate bias parameter, h t―1 is the output hidden state of the previous moment, x t The pulse features, age, and gender information in the feature data set;

[0051]

[0052] in, Candidate cell state, tanh represents the hyperbolic tangent function, W c is the candidate cell weight, b c is the candidate cell state bias parameter, h t―1 is the hidden state at the previous moment, x t Input switch for current data;

[0053] Determine how much of the network's input data needs to be saved to the unit state at the current moment;

[0054] The update equation is;

[0055]

[0056] Among them, C t Current cell state, Ct―1 is the cell state at the previous moment;

[0057] The equation for the output gate is:

[0058] o t =(W o ·[h t―1 ,x t ]+b o )

[0059] h t =o t *tanh(C t )

[0060] Among them, t is the current switch state of the output gate, δ represents the sigmoid function, W o is the output gate weight, b o is the output gate bias parameter, h t―1 is the hidden state at the previous moment, x t is the current data input switch, h t The results of cardiac function assessment;

[0061] Controls how much of the current unit state needs to be output to the current output value;

[0062] Among them, the mathematical expressions of the sigmoid function and the hyperbolic tangent function are;

[0063]

[0064] Among them, δ(x γ ) is the sigmoid function, x γ is the input of sigmoid function, tanh(C t ) is the hyperbolic tangent function, C t Input to the function.

[0065] Compared with the prior art, the present invention has the following beneficial effects:

[0066] The present invention acquires samples of continuous pulse signals, evaluates the cardiac function of the samples, establishes a sample data set, sets a noise threshold for the continuous pulse signals in the sample data set to perform filtering and noise reduction processing, and performs normalization processing to form a preprocessing data set, calculates the pulse quality score of each continuous pulse signal according to the amplitude and power spectrum density of the continuous pulse signal, screens the continuous pulse signals according to the pulse quality score, extracts time domain features and frequency domain features from the screened continuous pulse signals, establishes a cardiac function evaluation model according to a neural network and performs training, uses an acquisition module to collect continuous pulse signals and age and gender as inputs of the evaluation model, and completes the evaluation of cardiac function.

[0067] The present invention sets a noise threshold for the continuous pulse signal in the sample data set to perform filtering and noise reduction processing, and performs normalization processing. Through wavelet transform, the signal can be observed at different frequency levels, and the useful parts of the noise and the signal can be effectively separated, while retaining the important features of the signal. The pulse quality score of each continuous pulse signal is calculated according to the amplitude and power spectral density of the continuous pulse signal. The continuous pulse signal can be screened according to the pulse quality score, and the pulse signal with good quality can be selected for analysis, which greatly improves the accuracy of cardiac function evaluation. The pulse signal is extracted in time domain and frequency domain as the basis for cardiac function evaluation, which can enhance the representation of data in many aspects, improve the accuracy and adaptability of cardiac function evaluation, and establish a cardiac function evaluation model based on a neural network, effectively capture the complex dynamics of the healthy state of cardiac function changing over time, and have the ability to automatically learn and adapt to the data characteristics of different patients. BRIEF DESCRIPTION OF THE DRAWINGS

[0068] Figure 1 It is a schematic diagram of the overall system structure of the present invention. DETAILED DESCRIPTION

[0069] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with specific embodiments.

[0070] It should be noted that, unless otherwise defined, the technical terms or scientific terms used in the present invention should be understood by people with ordinary skills in the field to which the present invention belongs. The words "first", "second" and similar words used in the present invention do not indicate any order, quantity or importance, but are only used to distinguish different components. "Include" or "comprise" and similar words mean that the elements or objects appearing before the word include the elements or objects listed after the word and their equivalents, without excluding other elements or objects. "Connect" or "connected" and similar words are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to indicate relative positional relationships. When the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0071] Example:

[0072] See also Figure 1 , the present invention provides a technical solution:

[0073] A continuous pulse signal processing and cardiac function evaluation system comprises a data acquisition module, a preprocessing module, a signal evaluation module, a feature extraction module and a cardiac function evaluation module, wherein;

[0074] The data acquisition module obtains continuous pulse signal samples of samplers of different ages and genders, evaluates the cardiac function of each person, establishes a sample data set, and maps the continuous pulse signal, age, gender and timestamp in the sample data set one by one, and simultaneously obtains the continuous pulse signal and age and gender information of the person to be evaluated.

[0075] The pulse signal is a vibration signal caused by the heart pumping blood and the blood impacting the blood vessels. It captures the amount of blood pumped by each heartbeat by sensing the expansion of the blood vessel wall. It is a linear, non-stationary weak physiological signal. There is a close relationship between the pulse signal and heart function. Therefore, the pulse signal is an important physiological parameter reflecting the heart function and the state of the circulatory system.

[0076] Pulse wave signals can usually be collected using photoplethysmography, which is a non-invasive pulse wave detection technology that uses photoelectric elements to detect changes in blood volume caused by heartbeats on the surface of human skin, further reflecting the pulse wave morphology to obtain a pulse signal.

[0077] As we age, blood vessel walls may lose elasticity, which makes the pulse wave propagate faster, and the pulse waveform may show a higher pulse pressure. In addition, women usually have smaller hearts than men and may have a slightly higher resting heart rate. This means that when evaluating cardiac function, gender differences may lead to different normal ranges of heart rate and cardiac output. Therefore, when using pulse signals to evaluate cardiac function, it is usually necessary to consider the individual's age and gender, which helps to interpret the changes in the pulse signal more accurately, and it is also necessary to adjust the interpretation of the pulse waveform analysis according to these factors that may affect the pulse characteristics due to physiological differences.

[0078] The preprocessing module analyzes the noise characteristics of the continuous pulse signal in the sample data set, sets the noise threshold to perform filtering and noise reduction processing, filters the noise caused by equipment collection, and then normalizes the continuous pulse signal after noise reduction processing to form a preprocessing data set.

[0079] Continuous pulse signals are easily disturbed and generate noise during the acquisition process, including baseline drift caused by skin temperature changes, environmental noise, measurement movements, and circuit noise. These noises make it difficult to use pulse signals directly for further diagnosis and monitoring. The processing of noisy pulse signals is very important and has practical value. Wavelet transform can analyze signals at different scales, making it particularly suitable for processing biological signals with instantaneous characteristics, such as pulse signals. Through wavelet transform, signals can be observed at different frequency levels, effectively separating noise and useful parts of the signal. By setting an appropriate threshold, wavelet coefficients below the threshold can be eliminated, which are usually considered noise. This method is particularly suitable for removing random noise while retaining important features of the signal, such as the basic rhythm and instantaneous beating characteristics of the heart.

[0080] In this embodiment, the specific steps of performing filtering and noise reduction processing by setting the noise threshold are:

[0081] Firstly, the continuous pulse signal is transformed by wavelet transform and decomposed into a series of approximate coefficients and detail coefficients with different frequencies;

[0082]

[0083] Where X(t) is the continuous pulse signal in the sample data set, c j,k is the approximate coefficient, d j,k is the detail coefficient, is the scale function, is the wavelet function, t is the acquisition time point of the continuous pulse signal, j and k are the scale parameter and displacement parameter respectively.

[0084] By setting the noise threshold to process the wavelet coefficients, the detail coefficients affected by noise can be removed;

[0085]

[0086] Among them, d′ j,k is the detail coefficient after denoising, and τ is the set noise threshold.

[0087] Use the processed wavelet coefficients to perform inverse wavelet transform and reconstruct the denoised signal;

[0088]

[0089] Among them, X ′ (t) is the continuous pulse signal of the sample after denoising.

[0090] In this embodiment, the calculation formula for normalizing the continuous pulse signal after noise reduction is:

[0091]

[0092] Among them, X is the normalized pulse signal, X(t) is the denoised pulse signal, and X min is the minimum value in the original pulse signal data, X max is the maximum value in the original pulse signal data.

[0093] Normalization is a data mapping method that converts data into the interval of (-1, 1). Due to the different evaluation indicators, their dimensions are also different. Therefore, data normalization is used to make them dimensionless and conduct a comprehensive analysis. Normalization can also solve numerical problems caused by too large values.

[0094] The signal evaluation module is used to obtain the amplitude and power spectral density of the continuous pulse signal of the preprocessing data set, calculate the pulse quality score of each continuous pulse signal by calculating the average value and variance of the amplitude and the median value of the low-high frequency ratio of the continuous pulse signal, filter the continuous pulse signal according to the pulse quality score, and use the filtered continuous pulse signal as the analysis data set.

[0095] In this embodiment, the calculation formula for the average value and variance of the amplitude of the continuous pulse signal is:

[0096]

[0097] Among them, Xf ic is the amplitude of the icth main peak wave of the continuous pulse signal, nc is the number of main peak waves of the continuous pulse signal, μ c is the average amplitude of the main peak wave of the continuous pulse signal, σ c is the variance of the amplitude of the main peak wave of the continuous pulse signal.

[0098] The waveform of a clean pulse signal is smooth and regular, while the amplitude of a pulse signal interfered by a lot of noise fluctuates greatly. The mean and variance of the signal amplitude are significantly different between clean pulse signals and pulse signals interfered by noise. Therefore, the mean and variance of the signal amplitude can be used as features to evaluate the quality of the pulse signal.

[0099] The low-high frequency ratio median calculation formula is:

[0100]

[0101] Among them, q is the median of the low-high frequency ratio, P(g) is the power spectral density at frequency g, α1 and α2 are the frequency ranges of continuous pulse signals, and β1 and β2 are the low-frequency ranges of continuous pulse signals.

[0102] Since the frequency range of the pulse signal is 0-20Hz, most of the energy is concentrated in the frequency range of 1-10Hz. Combined with the interference type of the pulse signal, the low-high frequency ratio is defined as the ratio of the energy of the signal in the frequency range of 1-2.25Hz to the energy in the frequency range of 0-8Hz. The present invention calculates the low-high frequency ratio of each pulse in the pulse signal segment, and uses the median of the low-high frequency ratio as a feature for evaluating signal quality.

[0103] The calculation formula of the continuous pulse quality score is:

[0104] PF=γ1*μ c +γ2*σ c +γ1*q

[0105] Among them, PF is the continuous pulse quality score, γ1, γ2, and γ3 are the weights of the mean value, variance, and median low-high frequency ratio of the amplitude of the continuous pulse signal to the quality score.

[0106] The continuous pulse quality score reflects the overall signal quality of the continuous pulse signal. Although noise reduction processing can improve the clarity of the signal, there may still be quality differences between different signals. These differences may come from different conditions when the signal is captured, individual differences, equipment performance and other factors. Selecting high-quality signals can reduce these influences and thus reduce the errors that may be introduced during the analysis process.

[0107] The feature extraction module is used to perform time domain feature analysis on the continuous pulse signal in the analysis data set, and then extract the frequency domain features of the pulse signal through fast Fourier transform, and form a feature data set with the time domain features and the frequency domain features.

[0108] Extracting time and frequency domain features of pulse signals as the basis for cardiac function assessment can enhance the representation of data in many aspects and improve the accuracy and adaptability of cardiac function assessment. Time domain features reveal the behavior and characteristics of signals in the time dimension, such as heart rate, waveform shape, etc., which are important indicators for directly evaluating cardiac function. Frequency domain features provide information on signal frequency distribution, such as heart rate variability, spectral energy distribution, etc. These features can reflect the autonomic nervous system regulation of the heart under physiological and pathological conditions. The information provided by time domain and frequency domain analysis is complementary to some extent, and their combined use can provide a more comprehensive understanding of the health of cardiac function.

[0109] In this embodiment, the process of time domain feature analysis is as follows: by segmenting the continuous pulse signal into periods, the time domain features of the pulse signal of each period are obtained, a window of constant width is set according to the length and characteristics of the pulse signal, a new sampling point is placed at the highest position of the window, the average value and standard deviation of the continuous pulse signal period are calculated, and the segmentation period of the continuous pulse signal is obtained by the average value and standard deviation of the period. The specific calculation formula is as follows:

[0110]

[0111] Among them, A(i) is the period mean of the continuous pulse signal, B(i) is the period standard deviation of the continuous pulse signal, w is the width of the window set to be constant, Xq(tk) is the continuous pulse signal in the analysis sample, i is the end point position of the window, tk is the signal of the tkth point in the window, and CV is the segmentation period of the continuous pulse signal.

[0112] The specific formula for extracting the frequency domain characteristics of the pulse signal by fast Fourier transform is:

[0113]

[0114] Among them, VS(m) is the mth component in the frequency domain signal of the periodic pulse signal, Xq(in) is the inth sample in the continuous pulse signal, N is the total number of samples, m represents the mth frequency component in the frequency domain, 0≤m≤N-1, and j is an imaginary unit.

[0115] The fast Fourier transform algorithm can convert a signal from the time domain to the frequency domain, revealing the amplitude and phase characteristics of each frequency component in the signal. In cardiac function assessment, fast Fourier transform of the pulse signal and analysis of its frequency domain characteristics are very beneficial because it provides a non-invasive, intuitive and efficient way to understand the health status of cardiac function. The pulse signal is converted from the time domain to the frequency domain, which allows us to see the different frequency components in the signal. Each beat of the heart produces a specific frequency pattern in the pulse signal. For example, the acceleration and deceleration of the heart rate will affect the amplitude and distribution of these frequencies. Analyzing these frequency components can effectively identify heart disease symptoms such as arrhythmia and abnormal heart rate.

[0116] The cardiac function assessment module is used to establish a cardiac function assessment model through a neural network, and the model is trained by using a feature data set of a continuous pulse signal and age and gender as the input of the model, and the cardiac function assessment result as the output of the model. The pulse signal to be assessed, age and gender collected by the acquisition module are input into the trained assessment model to complete the assessment of cardiac function.

[0117] Long short-term memory network is a special type of recurrent neural network, which is very suitable for processing and predicting long-term dependency problems in time series data. In applications using pulse signals to evaluate cardiac function, long short-term memory networks can effectively capture the complex dynamics of cardiac health status over time. Pulse signals are a typical time series data that continuously records the heartbeat and blood flow changes over time. Heart diseases usually form time-dependent patterns in pulse signals. Long short-term memory networks can effectively learn the long-term dependencies in these time series due to their internal gating mechanisms such as forget gates, input gates, and output gates. Through their multi-layer structure and gating mechanism, they can simultaneously process and integrate these complex signal features to provide more accurate cardiac function assessment.

[0118] In long-term health monitoring, especially for patients with chronic heart disease, LSTM networks can process data over a long period of time and learn the development of changes in cardiac function. This ability makes LSTM networks particularly suitable for predicting health trends and possible acute events in cardiac function.

[0119] By combining long short-term memory networks and pulse signals, a powerful cardiac function assessment model can be constructed, which can learn and predict the health of the heart from continuous physiological data and provide accurate and personalized cardiac function assessment results.

[0120] In this embodiment, the specific method of establishing the cardiac function assessment model through the neural network is:

[0121] The cardiac function assessment model established by the neural network is based on a long short-term memory neural network, and the neural network model includes an input gate, a forget gate and an output gate;

[0122] The equation of the forget gate is:

[0123] f t =δ(W f ·[h t―1 ,x t ]+b f )

[0124] Among them, f t is the current switch state of the forget gate, δ represents the sigmoid function, W f is the forget gate weight, b f is the forget gate bias parameter, h t―1 is the hidden state at the previous moment, x t Input switch for current data;

[0125] Determine how much of the unit state at the previous moment needs to be retained to the current moment;

[0126] The equation for the input gate is:

[0127] i t =δ(W i ·[h t―1 ,x t ]+b i )

[0128] Among them, i t is the current switch state of the input gate, δ represents the sigmoid function, W i is the input gate weight, b i is the input gate bias parameter, h t―1 is the output hidden state of the previous moment, x t The pulse features, age, and gender information in the feature data set;

[0129]

[0130] in, Candidate cell state, tanh represents the hyperbolic tangent function, W c is the candidate cell weight, b c is the candidate cell state bias parameter, h t―1 is the hidden state at the previous moment, x t Input switch for current data;

[0131] Determine how much of the network's input data needs to be saved to the unit state at the current moment;

[0132] The update equation is;

[0133]

[0134] Among them, C t Current cell state, C t―1 is the cell state at the previous moment;

[0135] The equation for the output gate is:

[0136] o t =(W o ·[h t―1 ,x t ]+b o )

[0137] h t =o t *tanh(C t )

[0138] Among them, t is the current switch state of the output gate, δ represents the sigmoid function, W o is the output gate weight, b ois the output gate bias parameter, h t―1 is the hidden state at the previous moment, x t is the current data input switch, h t The results of cardiac function assessment;

[0139] Controls how much of the current unit state needs to be output to the current output value;

[0140] Among them, the mathematical expressions of the sigmoid function and the hyperbolic tangent function are;

[0141]

[0142] Among them, δ(x γ ) is the sigmoid function, x γ is the input of sigmoid function, tanh(C t ) is the hyperbolic tangent function, C t Input to the function.

[0143] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters in the formula are set by technicians in this field according to actual conditions.

[0144] The above embodiments may be implemented in whole or in part by software, hardware, firmware or any other combination thereof. When implemented using software, the above embodiments may be implemented in whole or in part in the form of a computer program product. Those skilled in the art may appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein may be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed by hardware or software methods depends on the specific application and design constraints of the technical solution.

[0145] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, and may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0146] The above description is only a specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application.

Claims

1. A continuous pulse signal processing and cardiac function assessment system, characterized in that: include: A data acquisition module, which is used to obtain continuous pulse signal samples of samplers of different ages and genders, evaluate the cardiac function of each person, establish a sample data set, and map the continuous pulse signal, age, gender and timestamp in the sample data set one by one, and simultaneously obtain the continuous pulse signal and age and gender information of the person to be evaluated; A preprocessing module, which is used to analyze the noise characteristics of the continuous pulse signal in the sample data set, set the noise threshold to perform filtering and noise reduction processing, filter the noise caused by equipment collection, and then normalize the continuous pulse signal after the noise reduction processing to form a preprocessing data set; A signal evaluation module, the signal evaluation module is used to obtain the amplitude and power spectrum density of the continuous pulse signal of the preprocessing data set, calculate the average value and variance of the amplitude of the continuous pulse signal, and the median value of the low-high frequency ratio, calculate the pulse quality score of each continuous pulse signal, screen the continuous pulse signal according to the pulse quality score, and use the screened continuous pulse signal as the analysis data set; A feature extraction module, which is used to perform time domain feature analysis on the continuous pulse signal in the analysis data set, and then extract the frequency domain features of the pulse signal by fast Fourier transform, and form a feature data set with the time domain features and the frequency domain features; A cardiac function assessment module is used to establish a cardiac function assessment model through a neural network, and the model is trained by taking the characteristic data set of continuous pulse signals and age and gender as the input of the model and taking the cardiac function assessment result as the output of the model. The pulse signal to be assessed and the age and gender collected by the acquisition module are input into the trained assessment model to complete the assessment of cardiac function.

2. A continuous pulse signal processing and cardiac function assessment system according to claim 1, characterized in that: The specific steps of performing filtering and noise reduction processing by setting the noise threshold are: Firstly, the continuous pulse signal is transformed by wavelet transform and decomposed into a series of approximate coefficients and detail coefficients with different frequencies; Where X(t) is the continuous pulse signal in the sample data set, c j,k is the approximate coefficient, d j,k is the detail coefficient, is the scale function, is the wavelet function, t is the acquisition time point of the continuous pulse signal, j and k are the scale parameter and displacement parameter respectively; By setting the noise threshold to process the wavelet coefficients, the detail coefficients affected by noise can be removed; Among them, d′ j,k is the detail coefficient after denoising, τ is the set noise threshold; Use the processed wavelet coefficients to perform inverse wavelet transform and reconstruct the denoised signal; Among them, X ′ (t) is the continuous pulse signal of the sample after denoising.

3. A continuous pulse signal processing and cardiac function assessment system according to claim 1, characterized in that: The calculation formula for normalizing the continuous pulse signal after noise reduction is: Among them, X is the normalized pulse signal, X(t) is the denoised pulse signal, and X min is the minimum value in the original pulse signal data, X max is the maximum value in the original pulse signal data.

4. A continuous pulse signal processing and cardiac function assessment system according to claim 1, characterized in that: The calculation formula of the mean value and variance of the amplitude of the continuous pulse signal is: Among them, Xf ic is the amplitude of the icth main peak wave of the continuous pulse signal, nc is the number of main peak waves of the continuous pulse signal, μ c is the average amplitude of the main peak wave of the continuous pulse signal, σ c is the variance of the amplitude of the main peak wave of the continuous pulse signal; The low-high frequency ratio median calculation formula is: Where q is the median of the low-high frequency ratio, P(g) is the power spectrum density at frequency g, α1 and α2 are the frequency ranges of the continuous pulse signal, and β1 and β2 are the low-frequency ranges of the continuous pulse signal; The calculation formula of the continuous pulse quality score is: PF=γ1*μ c +γ2*σ c +γ1*q Among them, PF is the continuous pulse quality score, γ1, γ2, and γ3 are the average value, variance, and proportion of the median low-high frequency ratio of the amplitude of the continuous pulse signal to the quality score.

5. A continuous pulse signal processing and cardiac function assessment system according to claim 1, characterized in that: The process of time domain feature analysis is as follows: by segmenting the continuous pulse signal into periods, obtaining the time domain features of the pulse signal of each period, setting a window of constant width according to the length and characteristics of the pulse signal, placing the new sampling point at the highest position of the window, calculating the average value and standard deviation of the continuous pulse signal period, and obtaining the segmentation period of the continuous pulse signal through the average value and standard deviation of the period. The specific calculation formula is as follows; Where A(i) is the cycle mean of the continuous pulse signal, B(i) is the cycle standard deviation of the continuous pulse signal, w is the width of the window set to be constant, Xq(tk) is the continuous pulse signal in the analysis sample, i is the end point position of the window, tk is the signal of the tkth point in the window, and CV is the segmentation period of the continuous pulse signal; The specific formula for extracting the frequency domain characteristics of the pulse signal by fast Fourier transform is: Among them, VS(m) is the mth component in the frequency domain signal of the periodic pulse signal, Xq(in) is the inth sample in the continuous pulse signal, N is the total number of samples, m represents the mth frequency component in the frequency domain, 0≤m≤N-1, and j is an imaginary unit.

6. A continuous pulse signal processing and cardiac function assessment system according to claim 1, characterized in that: The specific method of establishing the cardiac function evaluation model through the neural network is: The cardiac function assessment model established by the neural network is based on a long short-term memory neural network, and the neural network model includes an input gate, a forget gate and an output gate; The equation of the forget gate is: f t =δ(W f ·[h t―1 ,x t ]+b f ) Among them, f t is the current switch state of the forget gate, δ represents the sigmoid function, W f is the forget gate weight, b f is the forget gate bias parameter, h t―1 is the hidden state at the previous moment, x t Input switch for current data; Determine how much of the unit state at the previous moment needs to be retained to the current moment; The equation for the input gate is: i t =δ(W i ·[h t―1 ,x t ]+b i ) Among them, i t is the current switch state of the input gate, δ represents the sigmoid function, W i is the input gate weight, b i is the input gate bias parameter, h t―1 is the output hidden state of the previous moment, x t The pulse features, age, and gender information in the feature data set; in, Candidate cell state, tanh represents the hyperbolic tangent function, W c is the candidate cell weight, b c is the candidate cell state bias parameter, h t―1 is the hidden state at the previous moment, x t Input switch for current data; Determine how much of the network's input data needs to be saved to the unit state at the current moment; The update equation is; Among them, C t Current cell state, C t―1 is the cell state at the previous moment; The equation for the output gate is: o t =(W o ·[h t―1 ,x t ]+b o ) h t =o t *tanh(C t ) Among them, t is the current switch state of the output gate, δ represents the sigmoid function, W o is the output gate weight, b o is the output gate bias parameter, h t―1 is the hidden state at the previous moment, x t is the current data input switch, h t The results of cardiac function assessment; Controls how much of the current unit state needs to be output to the current output value; Among them, the mathematical expressions of the sigmoid function and the hyperbolic tangent function are; Among them, δ(x γ ) is the sigmoid function, x γ is the input of sigmoid function, tanh(C t ) is the hyperbolic tangent function, C t Input to the function.

Images