CN118430732B - Hemodialysis data intelligent processing system based on smart phone application program - Google Patents

Abstract

The present invention relates to the field of data processing technology, and in particular to a hemodialysis data intelligent processing system based on a smart phone application, including: collecting the original signal time series sequence of all parameters in the user's dialysis process; obtaining the fitting function curve of the original signal time series sequence; obtaining an observation error set according to the fitting function curve, obtaining a histogram of the local observation error variance according to the observation error set, and calculating the concentration of each local observation error variance according to the histogram; filtering out non-concentrated items according to the concentration mapping value, and obtaining an increased number of sampling points according to the proportion of non-concentrated items; sampling the original signal time series sequence according to the minimum total number of sampling points and the increased number of sampling points, obtaining a digital signal and displaying it. The present invention saves the sampling rate as much as possible while ensuring the integrity of the effective information, thereby greatly improving the monitoring signal receiving speed of the mobile phone end and reducing the storage pressure.

Description

Intelligent processing system of hemodialysis data based on smartphone application

Technical Field

The present invention relates to the technical field of data processing, and in particular to a hemodialysis data intelligent processing system based on a smart phone application.

Background Art

Smartphone application hemodialysis data intelligent processing system is a technology that uses smartphone App to monitor and manage the hemodialysis process in real time. Through the smartphone connected to the dialysis machine, various parameters in the dialysis process, such as blood pressure, weight, blood flow rate, dialysate flow rate, ultrafiltration volume, conductivity, pH value, etc., are monitored in real time, and the monitored data are analyzed and processed, and corresponding reports and charts are generated for doctors and nurses to evaluate, compare and formulate treatment plans.

The original signal detected by the dialysis machine is an analog signal. Since the analog signal is continuous in time and amplitude, it is easily interfered and distorted, and it is not easy to be processed and stored by digital devices such as computers, so it needs to be converted into a digital signal. The digitization process of analog signals requires three steps, namely sampling, quantization, and encoding, among which sampling is a very important step. Sampling refers to the conversion of continuous analog signals into discrete digital signals. The most widely used method is the uniform sampling method. On this basis, an adaptive sampling method based on signal characteristics is proposed. However, there is a lot of noise in the original analog signal, so the so-called signal characteristics are largely covered, and there are a large number of invalid sampling points in the traditional sampling method. How to accurately select effective points under noise interference to ensure the authenticity of dialysis data and compress the data volume as much as possible is a problem that needs to be solved urgently.

Summary of the invention

The present invention provides a hemodialysis data intelligent processing system based on a smart phone application program to solve the existing problems.

The hemodialysis data intelligent processing system based on a smart phone application of the present invention adopts the following technical solutions:

The present invention provides a hemodialysis data intelligent processing system based on a smart phone application, the system comprising:

The original signal acquisition module collects the original signal time series of all parameters of the user during the dialysis process, including blood pressure, weight, blood flow rate, dialysate flow rate, ultrafiltration volume, conductivity, and pH value;

A fitting function curve acquisition module is used to obtain a fitting function curve of the original signal time series;

A centralization calculation module obtains an observation error set according to a fitting function curve, obtains a histogram of local observation error variances according to the observation error set, and calculates the centralization of each local observation error variance according to the histogram;

Add a sampling point module, obtain the concentration mapping value of each local observation error variance according to the concentration, filter out the non-concentrated items according to the concentration mapping value, and increase the number of sampling points according to the proportion of non-concentrated items;

The digital signal conversion module samples the original signal timing sequence according to the minimum total sampling point number and the increased sampling point number to obtain the digital signal and display it.

Furthermore, the step of obtaining the fitting function curve of the original signal time series includes the following specific steps:

Decompose the original signal time series by EMD decomposition to obtain multiple IMF component signals; sample each IMF component signal at a sampling frequency of P sampling points per second according to a preset sampling frequency P, perform sinusoidal fitting on all sampling points of each IMF component signal using a sine function to obtain a sinusoidal fitting function of each IMF component signal; calculate the mean square error between the sinusoidal fitting function of the IMF component signal obtained by each sampling and the IMF component signal, and through multiple sampling and fitting, use the sinusoidal fitting function corresponding to the minimum mean square error as the fitting function of each IMF component signal;

The fitting functions of all IMF component signals are obtained, and the fitting functions of all IMF component signals are superimposed to obtain the fitting function curve of the original signal time series.

Furthermore, the obtaining of the observation error set includes the following specific steps:

The fitting function curve is used as the state equation of the Kalman filter, and the Kalman filter is used to traverse each original signal in the original signal time series to obtain the predicted amplitude of each original signal. The predicted amplitude of the original signal is obtained by using the state equation of the Kalman filter. The amplitude of the original signal at the next sampling point is predicted through the state of the original signal at each sampling point to obtain the predicted amplitude of the original signal at the next sampling point; the absolute value of the difference between the predicted amplitude of each original signal and the actual signal amplitude is recorded as the prediction error of each original signal to obtain a set of observation errors.

Furthermore, the step of obtaining the histogram of the local observation error variance includes the following specific steps:

The observation errors in the observation error set are organized into an observation error curve in chronological order, and the observation error curve is curve fitted to obtain an error fitting curve, which is composed of a number of error fitting values; a sliding window with a length equal to a preset length Y is slid on the error fitting curve, and the variance of all error fitting values in each sliding window is calculated, which is recorded as the local observation error variance; a histogram composed of the local observation error variances of all sliding windows is obtained, and the horizontal axis of the histogram is the local observation error variance, and the vertical axis is the frequency of each local observation error variance.

Furthermore, the calculation of the centralization of each local observation error variance includes the following specific steps:

According to the left and right statistical values of each local observation error variance, the centralization of each local observation error variance is calculated. The specific calculation formula is:

In the formula, represents the concentration of the local observation error variance c, represents the left statistic of the local observation error variance c, represents the right statistic of the local observation error variance c, represents the frequency of the local observation error variance c, Represents an exponential function with the natural constant e as the base.

Furthermore, the method for obtaining the left statistical value and the right statistical value of each local observation error variance is as follows:

The frequency of the local observation error variance with the largest frequency in the histogram is recorded as the maximum frequency ; For the local observation error variance c, obtain the left statistical value of the local observation error variance c , including: setting a counter, the initial value of the counter is 0, judging the frequency of the first local observation error variance on the left side of the local observation error variance c in the histogram and the maximum frequency Relationship: If the frequency is less than , add 1 to the counter, and continue to determine the frequency of the second local observation error variance on the left side of the local observation error variance c in the histogram until the frequency is greater than or equal to When , stop judging, and record the value of the counter at this time as the left statistical value of the local observation error variance c ;

Similarly, the right statistical value of the local observation error variance c is obtained .

Furthermore, the method of obtaining a centralization mapping value of each local observation error variance according to the centralization, and filtering out non-centralized items according to the centralization mapping value, includes the following specific steps:

The centralization of all local observation error variances is linearly normalized, and the normalized result is recorded as the centralization of each local observation error variance; the centralization mapping value of the local observation error variance is obtained according to the centralization of the local observation error variance, and the specific calculation formula is:

In the formula, represents the centralized mapping value of the local observation error variance c, represents the concentration of the local observation error variance i, Q represents the total number of sliding windows, Indicates rounding to the nearest integer;

If the centralization mapping value of the local observation error variance is equal to the centralization mapping value of the local observation error variance on the left, the sliding window corresponding to the local observation error variance is recorded as a non-centralized item, and all non-centralized items are obtained.

Furthermore, the step of increasing the number of sampling points includes the following specific steps:

The number of increased sampling points S to obtain the non-concentrated items is as follows:

In the formula, represents the number of additional sampling points for non-centralized items, N represents the number of non-centralized items, represents the total number of sliding windows, E represents the minimum total number of sampling points, Indicates rounding to the nearest integer.

Furthermore, the step of obtaining the digital signal includes the following specific steps:

According to the minimum total number of sampling points E, equally spaced sampling is performed on the original signal timing sequence. According to the increased number of sampling points S, equally spaced sampling is performed on the local signal segments corresponding to all non-concentrated items on the original signal timing sequence. The sampling points obtained on the original signal timing sequence are encoded and converted using PCM to obtain a digital signal, which is transmitted to the mobile phone application after conventional preprocessing.

Furthermore, the method for obtaining the minimum total number of sampling points is as follows:

The original signal time series is transformed into the frequency domain using Fourier transform. The signal bandwidth is obtained according to the difference between the maximum and minimum frequencies of the signal in the frequency domain, and then the minimum sampling rate is obtained. The time width in the time domain and the frequency width in the frequency domain are reciprocals of each other. Therefore, the reciprocal of the minimum sampling rate is the minimum number of sampling points A of the single-peak signal in the original signal time series. According to all the extreme points in the original signal, the number of all single-peak signals G is obtained, and then the minimum total number of sampling points of the original signal time series is obtained as E=2A. G.

The beneficial effects of the technical solution of the present invention are as follows: in view of the problem that the amount of data is large and the sampling points are affected by noise, resulting in a large number of invalid samples and serious damage to the original monitoring data in the analog-to-electric conversion of hemodialysis nursing monitoring data, the present invention performs EMD decomposition on the original signal, fits it with the sine function of the minimum mean square error according to the symmetric characteristics of its component signals, and obtains the fitting function curve of the original signal time series by superposition, optimizes the serious distortion problem caused by directly fitting the original signal, performs equal-interval sampling on the original signal time series according to the minimum total number of sampling points, and screens out the variance terms with non-concentrated variance and the sliding windows distributed thereon according to the histogram of the observation error set and the local observation variance, marks the local signal segments where these sliding windows are located in the original signal, and adaptively increases the sampling points for this part of the signal segment. While saving the sampling rate as much as possible, the integrity of the effective information is guaranteed, thereby greatly improving the monitoring signal receiving speed of the mobile phone end and reducing the storage pressure.

BRIEF DESCRIPTION OF THE DRAWINGS

In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

FIG1 is a system block diagram of a hemodialysis data intelligent processing system based on a smart phone application of the present invention;

Figure 2 is a timing sequence of the original blood pressure signal;

FIG3 is a schematic diagram of Gaussian distribution.

DETAILED DESCRIPTION

In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the specific implementation method, structure, features and effects of the hemodialysis data intelligent processing system based on a smartphone application proposed by the present invention are described in detail below in combination with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.

Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

The specific scheme of the hemodialysis data intelligent processing system based on the smart phone application provided by the present invention is described in detail below with reference to the accompanying drawings.

Please refer to FIG1 , which shows a hemodialysis data intelligent processing system based on a smart phone application provided by an embodiment of the present invention. The system includes the following modules:

It should be noted that the minimum sampling rate of PCM sampling coding should be 2B. Although the higher the sampling rate, the more complete the original information is retained, it does not mean that the higher the sampling rate, the better. Too high a sampling rate will lead to increased data redundancy and storage costs, and also increase the complexity of subsequent data processing. Therefore, the present embodiment hopes to retain the integrity of the signal to the greatest extent under the premise of being as close to the minimum sampling rate as possible. To achieve this goal, the effective signal must be identified to save enough sampling rates. Considering that the original signal is divided into low-frequency information and high-frequency information, the low-frequency part is mostly valid information, and the noise is mainly concentrated in the high-frequency part. Due to noise interference, there is a large deviation in directly fitting the low-frequency signal. Therefore, this embodiment first performs EMD decomposition on the original signal time series, and fits the decomposed component signals respectively to obtain a more accurate fitting curve; based on the effective information in the fitting function curve of the EMD component signal is complete enough, the distribution of the observation error is more consistent with the distribution characteristics of Gaussian noise, that is, the more concentrated its local variance is, the higher the obedience of these observation errors to the variance of the Gaussian noise distribution is, and the closer the observation error is to the real noise signal, then when the local variance is not concentrated, there must be some valid information that is not included in the fitting function. This embodiment needs to find this part of the signal and increase the number of sampling points of this part. Therefore, the concentration characteristics of the local observation error variance are obtained by characterizing the left and right statistical values of the sparsity of the adjacent local areas of each local observation error variance, and the frequency characterizing the sparsity of the local observation error variance itself, and then the concentration mapping value of the local observation error variance is obtained; for the signal in the local window where the variance is not concentrated, since the local observation error variance obtained by its sliding window belongs to the variance non-concentrated item, there is a large difference between the observation error of this part of the signal and the actual noise, which also means that the effective information in the fitting function curve of this part is not fully extracted, and a part of it exists in the observation error. Therefore, more sampling points need to be put on this part of the signal to ensure that the number of sampling points is sufficient to retain the integrity of the effective information as much as possible.

The original signal acquisition module S101 is used to acquire the original signal time series of various parameters during the dialysis process.

It should be noted that the smart phone application hemodialysis data intelligent processing system is a technology that uses smart phone apps to monitor and manage the hemodialysis process in real time. Through the smart phone connected to the dialysis machine, various parameters in the dialysis process are monitored in real time, the monitored data are analyzed and processed, and corresponding reports and charts are generated for doctors and nurses to evaluate, compare and formulate treatment plans.

Specifically, when the user undergoes dialysis, the original signal timing sequence corresponding to all parameters is obtained, and the parameters include: blood pressure, weight, blood flow rate, dialysate flow rate, ultrafiltration volume, conductivity, pH value, etc.; this embodiment takes the collected original signal of blood pressure as an example to illustrate, and the original signal timing sequence of the user's blood pressure during the dialysis process is collected by the dialysis machine, and the signal on the original signal timing sequence is continuous; please refer to Figure 2, which shows the original signal timing sequence of blood pressure, wherein the horizontal axis is time and the vertical axis is signal amplitude.

It should be further explained that the original signal timing sequence of each parameter is an analog-to-electric signal, which is an analog signal that changes continuously in time and amplitude. The original signal timing sequence collected by the dialysis machine has a high content of white noise (Gaussian noise). This embodiment will subsequently mainly analyze the impact of white noise on the analog-to-electric conversion process.

The fitting function curve acquisition module S102 is used to acquire the fitting function curve.

It should be noted that PCM sampling coding (pulse code modulation) is a common method for converting analog signals into digital signals. It samples and quantizes analog signals at certain time intervals, and encodes the sampled data into digital signals. It is often used for encoding audio signals. According to the Nyquist sampling theorem, the minimum sampling rate should meet twice the signal bandwidth. In other words, if the signal bandwidth is B, the minimum sampling rate should be 2B. Although the higher the sampling rate, the more complete the original information is retained, it does not mean that the higher the sampling rate, the better. Excessive sampling rate will lead to increased data redundancy and storage costs, and also increase the complexity of subsequent data processing. Therefore, this embodiment hopes to retain the integrity of the signal to the greatest extent possible under the premise of being as close to the minimum sampling rate as possible. To achieve this goal, it is necessary to identify the valid signal in order to save enough sampling rate.

1. Obtain the total number of sampling points of the original signal timing sequence.

Specifically, the original signal time series is transformed into the frequency domain using Fourier transform, and the signal bandwidth is obtained according to the difference between the maximum and minimum frequencies of the signal in the frequency domain, and then the minimum sampling rate is obtained. The time width in the time domain and the frequency width in the frequency domain are reciprocals of each other, so the reciprocal of the minimum sampling rate is the minimum number of sampling points A of the single-peak signal in the original signal time series. According to all the extreme points in the original signal, the number of all single-peak signals G is obtained, and then the minimum total number of sampling points of the original signal time series is obtained as E=2A. G.

2. Obtain the fitting function curve and its total number of sampling points.

It should be noted that the original signal is divided into low-frequency information and high-frequency information. The low-frequency part is mostly valid information, and the noise is mainly concentrated in the high-frequency part. Due to noise interference, there is a large deviation in directly fitting the low-frequency signal. Therefore, this embodiment first performs EMD decomposition on the original signal time series, and fits the decomposed component signals separately to obtain a more accurate fitting curve.

It should be further explained that EMD decomposition is a well-known technology, but the characteristics of the component signal need to be used later, so the general steps are explained: first, the extreme point is obtained, and the upper and lower envelope functions and the local mean function are calculated according to the extreme point. The original signal is repeatedly subtracted from the local mean function to obtain the residual term. Each repetition needs to calculate the first-order difference of the mean of the envelope under the residual term until the first-order difference is 0, that is, the upper and lower envelopes are symmetrical. Then the first component signal is obtained, and the original signal is subtracted from the first component signal to obtain the residual signal, and then the above steps are repeated to decompose the residual signal until all component signals are obtained. The stopping condition is that the last component signal has less than two extreme points, then the decomposition is stopped; the original signal time series is decomposed by EMD decomposition to obtain multiple IMF component signals. Since the IMF component signal must satisfy the first-order difference of 0 during the decomposition process, each component signal is a symmetrical signal. Therefore, this embodiment uses a sine function to fit the component signal.

It should be further explained that the original signal time series is an analog signal, and the signal on the original signal time series is continuous, and theoretically there is no sampling point, so the value can be taken at any time on the time series, that is, there are infinite sampling points on the original signal time series. However, this embodiment is only for calculating the fitting error between the sine fitting function and the component signal, so the sampling frequency P can be directly set, and when the number of sampling points reaches a certain number, the mean square error value will hardly change with the change of the total number of sampling points.

A sampling frequency P is preset, wherein this embodiment is described by taking P=1200 Hz as an example, and this embodiment is not specifically limited, wherein P may be determined according to specific implementation conditions.

Specifically, the original signal time series is decomposed through EMD decomposition to obtain multiple IMF component signals; according to the sampling frequency P, sampling is performed on each IMF component signal at a sampling frequency of 1200 sampling points per second, and all sampling points of each IMF component signal are sinusoidally fitted using a sine function to obtain a sinusoidal fitting function of each IMF component signal; the mean square error between the sinusoidal fitting function of the IMF component signal obtained in each sampling and the IMF component signal is calculated, and through multiple sampling and fitting, when the minimum mean square error is taken, the corresponding sinusoidal fitting function has the highest fitting degree for the IMF component signal, and therefore, the sinusoidal fitting function corresponding to the minimum mean square error is used as the fitting function of each IMF component signal.

Furthermore, the fitting functions of all IMF component signals are obtained, and the fitting functions of all IMF component signals are superimposed to obtain a fitting function curve.

Furthermore, the fitting function curve is a relatively accurate low-frequency baseline function obtained by EMD decomposition. In this embodiment, the sampling rate of low-frequency information is prioritized. Therefore, the total number of sampling points of the fitting function curve is set to E, where E is the minimum total number of sampling points of the original signal time series sequence obtained in step 1.

The centrality calculation module S103 is used to obtain a set of observation errors, obtain a histogram of local observation error variances, and calculate the centrality of each local observation error variance.

1. Obtain the set of observation errors.

It should be noted that this embodiment uses Kalman filtering to perform real-time prediction on the original signal and obtain the fitting error distribution of the original signal to better observe the local fitting effect of the original signal. The Kalman filter is used to traverse the original signal. Note that the Kalman filter used in the present invention is not a denoising process, and the analog signal needs to be converted into a digital signal before denoising can be performed.

Specifically, the fitting function curve is used as the state equation of the Kalman filter, and the Kalman filter is used to traverse each original signal in the original signal time series sequence to obtain the predicted amplitude of each original signal. The predicted amplitude of the original signal is obtained by using the state equation of the Kalman filter. The amplitude of the original signal at the next sampling point is predicted through the state of the original signal at each sampling point to obtain the predicted amplitude of the original signal at the next sampling point; the absolute value of the difference between the predicted amplitude of each original signal and the actual signal amplitude is recorded as the prediction error of each original signal, and the observation error set can be obtained.

2. Obtain a histogram of the local observation error variance.

It should be noted that if the effective information in the fitting function curve based on the EMD component signal is complete enough, the distribution of the observation error is more consistent with the distribution characteristics of Gaussian noise, that is, the more concentrated its local variance is, the higher the degree of compliance of these observation errors with the Gaussian noise distribution variance is. It also means that the closer the observation error is to the real noise signal, then when the local variance is not concentrated, there must be some effective information that is not included in the fitting function. This embodiment needs to find this part of the signal and increase the number of sampling points of this part.

A length Y is preset, wherein this embodiment is described by taking Y=100 as an example, and this embodiment is not specifically limited, wherein Y can be determined according to specific implementation conditions.

Specifically, the observation errors in the observation error set are organized into an observation error curve in chronological order, and the observation error curve is curve fitted to obtain an error fitting curve, where the error fitting curve is composed of a number of error fitting values; a sliding window with a length equal to Y is slid on the error fitting curve, and the variance of all error fitting values in each sliding window is calculated, which is recorded as the local observation error variance; a histogram composed of the local observation error variances of all sliding windows is obtained, where the horizontal axis of the histogram is the local observation error variance, and the vertical axis is the frequency of each local observation error variance.

3. Calculate the centrality of the error variance of each local observation.

It should be noted that the local observation error variance is not concentrated, that is, the number of sliding windows distributed on the variance value is small and sparse. According to the characteristics of Gaussian distribution, that is, on Gaussian distribution, the rough The interval is considered to be a less distributed area. The area is considered to be a concentrated distribution area. Please refer to Figure 3, which shows a schematic diagram of Gaussian distribution. The total distribution quantity in the interval with less distribution accounts for about 1/3 of the central concentrated area.

Specifically, the frequency of the local observation error variance with the largest frequency in the histogram is recorded as the maximum frequency ; For the local observation error variance c, obtain the left statistical value of the local observation error variance c , including: setting a counter, the initial value of the counter is 0, judging the frequency of the first local observation error variance on the left side of the local observation error variance c in the histogram and the maximum frequency Relationship: If the frequency is less than , add 1 to the counter, and continue to determine the frequency of the second local observation error variance on the left side of the local observation error variance c in the histogram until the frequency is greater than or equal to When , stop judging, and record the value of the counter at this time as the left statistical value of the local observation error variance c ; Similarly, obtain the right statistical value of the local observation error variance c ; The larger the value of the counter, the sparser the local observation error variance is in the histogram.

It should be noted that the left statistical value of the local observation error variance c represents the sparsity of the local area to the left of the local observation error variance c in the histogram, the right statistical value of the local observation error variance c represents the sparsity of the local area to the right of the local observation error variance c in the histogram, and the frequency of the local observation error variance c itself represents the sparsity at the local observation error variance c in the histogram. Therefore, the centralization feature representing the degree of concentration at the local observation error variance c is obtained based on the left statistical value, right statistical value and frequency of the local observation error variance c.

Furthermore, the centralization of each local observation error variance is calculated. The specific calculation formula is:

In the formula, represents the concentration of the local observation error variance c, represents the left statistic of the local observation error variance c, represents the right statistic of the local observation error variance c, represents the frequency of the local observation error variance c, Represents an exponential function with the natural constant e as the base.

The larger the value, the sparser the local observation error variance c in the histogram, and the more concentrated the local observation error variance c is. The smaller; Represents the frequency of the local observation error variance c. The smaller the value, the less concentrated the local observation error variance c is in the histogram. The concentration of the local observation error variance c is The smaller the value, the average count value of the local observation error variance c is calculated. The probability of its vertical axis The Euclidean norm of , the smaller the norm is, the sparser the local observation error variance c is, and the worse the centralization of the local observation error variance c is.

The sampling point adding module S104 is used to filter out the non-concentrated items and obtain the number of increased sampling points.

Specifically, the centralization of all local observation error variances is linearly normalized, and the normalized centralization is recorded as the centralization of the local observation error variance. According to the centralization of the local observation error variance, the centralization mapping value of the local observation error variance is obtained. The specific calculation formula is:

In the formula, represents the centralized mapping value of the local observation error variance c, represents the concentration of the local observation error variance i, Q represents the total number of sliding windows, Indicates rounding to the nearest integer.

Furthermore, if the centralization mapping value of the local observation error variance is equal to the centralization mapping value of the local observation error variance on the left, the sliding window corresponding to the local observation error variance is recorded as an unconcentrated item, and all unconcentrated items are obtained, that is, all local windows with unconcentrated variances.

It should be noted that for the signal in the local window with unequal variance, since the local observation error variance obtained by sliding the window belongs to the unequal variance item, there is a large difference between the observation error of this part of the signal and the actual noise. This also means that the effective information in the fitting function curve of this part is not fully extracted, and some of it still exists in the observation error. Therefore, more sampling points need to be placed on this part of the signal.

Specifically, the number of increased sampling points S for obtaining the non-concentrated items is:

In the formula, represents the number of additional sampling points for non-centralized items, N represents the number of non-centralized items, represents the total number of sliding windows, E represents the minimum total number of sampling points, Indicates rounding to the nearest integer.

Characterize the proportion of effective information remaining in the observation error. In order to better retain this part of effective information, when sampling the original analog electrical signal, S sampling points are added to the marked signal segment, and these added sampling points are added to the number of sampling points uniformly distributed in the segment according to the minimum sampling data, and then the sampling is uniformly performed again.

The digital signal conversion module S105 is used to sample the original signal timing sequence to obtain a digital signal and display it.

Specifically, sampling is performed at equal intervals on the original signal timing sequence according to the minimum total number of sampling points E, and sampling is performed at equal intervals on the local signal segments corresponding to all non-concentrated items on the original signal timing sequence according to the increased number of sampling points S; the sampling points obtained on the original signal timing sequence are encoded and converted using PCM to obtain a digital signal, which is transmitted to the mobile phone application after conventional preprocessing.

Through the sampling method of this embodiment, the number of sampling points can be ensured to be sufficient to retain the integrity of effective information. When doctors or patients use mobile phone applications to view hemodialysis data, the receiving speed of the mobile phone can be greatly improved, the storage pressure can be reduced, and the integrity of the monitoring data can be ensured.

The system of the present invention includes an original signal acquisition module, a fitting function curve acquisition module, a centralized calculation module, a sampling point increase module and a digital signal conversion module. In view of the problem that the amount of data is large and the sampling points are affected by noise in the analog-to-electric conversion of hemodialysis nursing monitoring data, there are a large number of invalid samplings, which seriously damage the original monitoring data, the present invention performs EMD decomposition on the original signal, fits it according to the symmetric characteristics of its component signals using the sine function of the minimum mean square error, and obtains the fitting function curve of the original signal time series by superposition, optimizes the serious distortion problem caused by directly fitting the original signal, performs equal-interval sampling on the original signal time series according to the minimum total number of sampling points, and screens out the variance terms with unconcentrated variance and the sliding windows distributed thereon according to the histogram of the observation error set and the local observation variance, marks the local signal segments where these sliding windows are located in the original signal, and adaptively increases the sampling points for this part of the signal segment. While saving the sampling rate as much as possible, the integrity of the effective information is guaranteed, thereby greatly improving the monitoring signal receiving speed of the mobile phone end and reducing the storage pressure.

The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the principles of the present invention should be included in the protection scope of the present invention.

Claims (6)

1. A hemodialysis data intelligent processing system based on a smartphone application, characterized in that the system comprises: The original signal acquisition module collects the original signal time series of all parameters of the user during the dialysis process, including blood pressure, weight, blood flow rate, dialysate flow rate, ultrafiltration volume, conductivity, and pH value; A fitting function curve acquisition module is used to obtain a fitting function curve of the original signal time series; A centralization calculation module obtains an observation error set according to a fitting function curve, obtains a histogram of local observation error variances according to the observation error set, and calculates the centralization of each local observation error variance according to the histogram; Add a sampling point module, obtain the concentration mapping value of each local observation error variance according to the concentration, filter out the non-concentrated items according to the concentration mapping value, and increase the number of sampling points according to the proportion of non-concentrated items; The digital signal conversion module samples the original signal timing sequence according to the minimum total sampling point number and the increased sampling point number, obtains the digital signal and displays it; The calculation of the centralization of each local observation error variance includes the following specific steps: According to the left and right statistical values of each local observation error variance, the centralization of each local observation error variance is calculated. The specific calculation formula is: In the formula, represents the concentration of the local observation error variance c, represents the left statistic of the local observation error variance c, represents the right statistic of the local observation error variance c, represents the frequency of the local observation error variance c, It represents the exponential function with the natural constant e as the base; The method for obtaining the left statistical value and the right statistical value of each local observation error variance is as follows: The frequency of the local observation error variance with the largest frequency in the histogram is recorded as the maximum frequency ; For the local observation error variance c, obtain the left statistical value of the local observation error variance c , including: setting a counter, the initial value of the counter is 0, judging the frequency of the first local observation error variance on the left side of the local observation error variance c in the histogram and the maximum frequency Relationship: If the frequency is less than , add 1 to the counter, and continue to determine the frequency of the second local observation error variance on the left side of the local observation error variance c in the histogram until the frequency is greater than or equal to When , stop judging, and record the value of the counter at this time as the left statistical value of the local observation error variance c ; Similarly, the right statistical value of the local observation error variance c is obtained ; The method of obtaining the centralization mapping value of each local observation error variance according to the centralization and filtering out the non-centralized items according to the centralization mapping value includes the following specific steps: The centralization of all local observation error variances is linearly normalized, and the normalized result is recorded as the centralization of each local observation error variance; the centralization mapping value of the local observation error variance is obtained according to the centralization of the local observation error variance, and the specific calculation formula is: In the formula, represents the centralized mapping value of the local observation error variance c, represents the concentration of the local observation error variance i, Q represents the total number of sliding windows, Indicates rounding to the nearest integer; If the centralization mapping value of the local observation error variance is equal to the centralization mapping value of the local observation error variance on the left, the sliding window corresponding to the local observation error variance is recorded as a non-centralized term, and all non-centralized terms are obtained; The specific steps of increasing the number of sampling points are as follows: The number of increased sampling points S to obtain the non-concentrated items is as follows: In the formula, represents the number of additional sampling points for non-centralized items, N represents the number of non-centralized items, represents the total number of sliding windows, E represents the minimum total number of sampling points, Indicates rounding to the nearest integer. 2. The hemodialysis data intelligent processing system based on a smart phone application according to claim 1, characterized in that the step of obtaining the fitting function curve of the original signal time series comprises the following specific steps: Decompose the original signal time series by EMD decomposition to obtain multiple IMF component signals; sample each IMF component signal at a sampling frequency of P sampling points per second according to a preset sampling frequency P, perform sinusoidal fitting on all sampling points of each IMF component signal using a sine function to obtain a sinusoidal fitting function of each IMF component signal; calculate the mean square error between the sinusoidal fitting function of the IMF component signal obtained by each sampling and the IMF component signal, and through multiple sampling and fitting, use the sinusoidal fitting function corresponding to the minimum mean square error as the fitting function of each IMF component signal; The fitting functions of all IMF component signals are obtained, and the fitting functions of all IMF component signals are superimposed to obtain the fitting function curve of the original signal time series. 3. The hemodialysis data intelligent processing system based on a smart phone application according to claim 1, characterized in that the obtaining of the observed error set comprises the following specific steps: The fitting function curve is used as the state equation of the Kalman filter, and the Kalman filter is used to traverse each original signal in the original signal time series to obtain the predicted amplitude of each original signal. The predicted amplitude of the original signal is obtained by using the state equation of the Kalman filter. The amplitude of the original signal at the next sampling point is predicted through the state of the original signal at each sampling point to obtain the predicted amplitude of the original signal at the next sampling point; the absolute value of the difference between the predicted amplitude of each original signal and the actual signal amplitude is recorded as the prediction error of each original signal to obtain a set of observation errors. 4. The hemodialysis data intelligent processing system based on a smart phone application according to claim 1, characterized in that the obtaining of the histogram of the local observation error variance comprises the following specific steps: The observation errors in the observation error set are organized into an observation error curve in chronological order, and the observation error curve is curve fitted to obtain an error fitting curve, which is composed of a number of error fitting values; a sliding window with a length equal to a preset length Y is slid on the error fitting curve, and the variance of all error fitting values in each sliding window is calculated, which is recorded as the local observation error variance; a histogram composed of the local observation error variances of all sliding windows is obtained, and the horizontal axis of the histogram is the local observation error variance, and the vertical axis is the frequency of each local observation error variance. 5. The hemodialysis data intelligent processing system based on a smart phone application according to claim 1, characterized in that the obtaining of the digital signal comprises the following specific steps: According to the minimum total number of sampling points E, equally spaced sampling is performed on the original signal timing sequence. According to the increased number of sampling points S, equally spaced sampling is performed on the local signal segments corresponding to all non-concentrated items on the original signal timing sequence. The sampling points obtained on the original signal timing sequence are encoded and converted using PCM to obtain a digital signal, which is transmitted to the mobile phone application after conventional preprocessing. 6. The hemodialysis data intelligent processing system based on a smart phone application according to claim 1, characterized in that the method for obtaining the minimum total number of sampling points is as follows: The original signal time series is transformed into the frequency domain using Fourier transform. The signal bandwidth is obtained according to the difference between the maximum and minimum frequencies of the signal in the frequency domain, and then the minimum sampling rate is obtained. The time width in the time domain and the frequency width in the frequency domain are reciprocals of each other. Therefore, the reciprocal of the minimum sampling rate is the minimum number of sampling points A of the single-peak signal in the original signal time series. According to all the extreme points in the original signal, the number of all single-peak signals G is obtained, and then the minimum total number of sampling points of the original signal time series is obtained as E=2A. G.