Medical remote monitoring method and system

Abstract

The present invention relates to the field of medical monitoring technology, specifically a medical remote monitoring method and system, including reading a first health type, a first sampling index and a first sampling frequency; collecting the first sampling index monitoring data of the first sample population; obtaining the first health index scale by improving the hierarchical scaling method, and training the first BP neural network; collecting the second sampling index monitoring data of the monitored object; obtaining the second health index scale of the monitored object by improving the hierarchical scaling method, inputting the first BP neural network, and obtaining the health type of the monitored object; repeatedly correcting the second sampling frequency by using the second sampling frequency correction method, and cyclically sampling until the number of the same health types obtained reaches a threshold value; transmitting the health type to the medical personnel. The present invention can reduce the interference of abnormal monitoring data, improve the robustness of medical monitoring data analysis and diagnosis, realize adaptive dynamic correction of sampling frequency, and improve data processing efficiency.

Description

Technical Field

[0001] The present invention relates to the technical field of medical monitoring, and in particular to a medical remote monitoring method and system. Background Art

[0002] Medical telemonitoring is the use of digital means to enable medical professionals to remotely monitor patients' physical indicators and remotely diagnose diseases. Medical telemonitoring includes three steps: physical indicator data collection, physical indicator data transmission, and physical indicator data analysis and diagnosis. However, at present, medical telemonitoring still has the following problems: first, the problem of abnormal data interference. The reasons for abnormal data include jumps in the sampling data of the monitoring equipment and abnormal monitoring data caused by external environmental factors. Existing medical telemonitoring methods often diagnose diseases through a single or a small number of interrelated indicators. These methods are easily interfered by abnormal monitoring data, affecting the robustness of diagnosis; second, the existing medical telemonitoring system often uses a fixed sampling frequency in the process of monitoring data collection, resulting in a lack of pertinence in the sampling data, and the amount of monitoring data at the source end is bloated, affecting the data processing efficiency. This system can be used for medical telemonitoring in the case of abnormal interference in the monitoring data, excessive amount of monitoring data, and lack of adaptive dynamic adjustment of the sampling frequency of monitoring data. Summary of the invention

[0003] (1) Technical issues to be solved

[0004] The purpose of the present invention is to provide a medical remote monitoring method and system to solve the problem of abnormal data collection at the medical monitoring source and the bloated data volume affecting the diagnosis effect.

[0005] (2) Technical solution

[0006] To achieve the above object, on the one hand, the present invention provides a medical remote monitoring method, the method comprising:

[0007] Step 1: read the first file data, the first file data includes a first health type, a first sampling indicator and a first sampling frequency; the first health type, the first sampling indicator and the first sampling frequency are all arrays, and the first sampling indicator and the first sampling frequency have a one-to-one mapping relationship; construct a first sample population, and establish a mapping relationship between the first sample population and the first health type; take the first sampling frequency as the sampling frequency, and collect a monitoring data sequence of the first sampling indicator of the first sample population within a first time range; hierarchically scale the monitoring data sequence of the first sampling indicator of the first sample population by improving the hierarchical scaling method, and obtain the first health indicator scale of the first sample population; according to the mapping relationship between the first sample population and the first health type and the first health indicator scale of the first sample population, obtain the mapping relationship between the first health type and the first health indicator scale;

[0008] Step 2: Merge the first health type and the first health indicator scale to obtain a first training sample; input the first training sample into a BP neural network, output a BP neural network weight parameter, and obtain a first BP neural network; initialize the second sampling frequency to the first sampling frequency, initialize the second sampling indicator to the first sampling indicator, and start medical remote monitoring;

[0009] Step 3: Using the second sampling frequency as the sampling frequency, collect the monitoring data sequence of the second sampling indicator of the monitored object within a preset time range; hierarchically scale the monitoring data sequence of the second sampling indicator of the monitored object by improving the hierarchical scaling method to obtain the second health indicator scale of the monitored object; input the second health indicator scale of the monitored object into the first BP neural network to obtain the health type of the monitored object; clone the health type of the monitored object into the health type library of the monitored object; use the second sampling frequency correction method to correct the second sampling frequency, and assign the second sampling frequency to the corrected second sampling frequency; check whether the number of the same health types contained in the health type library of the monitored object reaches the preset first health number threshold value, if not, repeat the step 3;

[0010] Step 4: Transmit the health type library of the monitored object to medical personnel via a communication network.

[0011] Furthermore, the method of performing hierarchical scaling on the monitoring data sequence of the first sampling indicator of the first sample population by improving the hierarchical scaling method to obtain the first health indicator scale of the first sample population includes:

[0012] Step 11: clone the monitoring data sequence of the first sampling indicator of the first sample population into the first monitoring cell matrix; each element in the first monitoring cell matrix is ​​the corresponding monitoring data sequence of the first sampling indicator of the first sample population; calculate the average value of all elements in the first monitoring cell matrix to obtain the first sampling matrix; divide the first sampling matrix into T first indicator mean arrays by column, where T is the number of columns of the first sampling matrix; establish a mapping relationship between the first indicator mean array and the first sampling indicator;

[0013] Step 12: Spread the T first indicator mean arrays to T independent two-dimensional coordinate planes respectively, and perform maximum likelihood estimation fitting respectively to obtain probability density curve clusters of the first indicator mean arrays; establish a mapping relationship between the first sampling indicator and the probability density curve cluster of the first indicator mean array;

[0014] Step 13: hierarchically separate each probability density curve in the probability density curve cluster of the first indicator mean array according to the cumulative distribution to obtain a hierarchical mapping relationship of the first sampling matrix; map the first sampling matrix according to the hierarchical mapping relationship to obtain the first health indicator scale of the first sample population.

[0015] Furthermore, the method of hierarchically separating each probability density curve in the probability density curve cluster of the first indicator mean array according to the cumulative distribution to obtain the hierarchical mapping relationship of the first sampling matrix includes:

[0016] Step 131: using an improved concentration evaluation method to obtain the classification value of each probability density curve in the probability density curve cluster of the first indicator mean array; taking a weighted average of the classification values ​​of each probability density curve in the probability density curve cluster of the first indicator mean array, and rounding up, to obtain an overall classification value X;

[0017] Step 132: For any probability density curve in the probability density curve cluster of the first indicator mean array, let the horizontal axis variable start from negative infinity and search the integral area of ​​the probability density curve in a positive direction. When the integral area is When , the corresponding horizontal axis variable value is recorded as the i-th interval point of the probability density curve, and i is traversed through all integers in the interval from 1 to 2X to obtain the first interval point to the 2X interval point of the probability density curve;

[0018] Step 133: traverse all probability density curves in the probability density curve cluster of the first indicator mean value array, and repeat the interval point construction process respectively;

[0019] Step 134: Establishing a mapping relationship between the probability density curve and the first sampling index; for all the probability density curves, dividing the abscissa of each probability density curve into 2X+1 intervals with all the interval points of the probability density curve as intervals, and sequentially naming them from the first interval to the 2X+1 interval of the first sampling index element corresponding to the probability density curve;

[0020] Step 135: Establish a hierarchical mapping relationship of the first sampling matrix; the input of the hierarchical mapping relationship of the first sampling matrix is ​​the first sampling matrix, and the output is a matrix composed of integers in the range of -X to X; when the value of the element of the first sampling matrix falls within the corresponding mth interval, the element at the same position in the corresponding output matrix is ​​-X+m-1, 1≤m≤2X+1.

[0021] Furthermore, the method of using the improved concentration evaluation method to obtain the classification amount of each probability density curve in the probability density curve cluster of the first indicator mean array includes:

[0022] Step 1311: Calculate the variable variance corresponding to each probability density curve in the probability density curve cluster of the first indicator mean array;

[0023] Step 1312: Input the variable variance into a preset variance logic mapping table to obtain the classification amount of each probability density curve.

[0024] Furthermore, the method of correcting the second sampling frequency using the second sampling frequency correction method includes:

[0025] Step 31: Read the number of corrections;

[0026] Step 32: Use the second sampling frequency gain formula Obtaining a frequency gain coefficient k, wherein l is the number of corrections, C1 is a preset gain speed constant, C2 is a preset gain amplitude constant, and C3 is a preset gain reference constant;

[0027] Step 33: multiply the first sampling frequency by the frequency gain coefficient k to obtain a corrected second sampling frequency.

[0028] Based on the same inventive concept, on the other hand, the present invention also provides a medical remote monitoring system, the system comprising:

[0029] A first health sample construction module is used to read first archive data, wherein the first archive data includes a first health type, a first sampling indicator and a first sampling frequency; the first health type, the first sampling indicator and the first sampling frequency are all arrays, and the first sampling indicator has a one-to-one mapping relationship with the first sampling frequency; construct a first sample population, and establish a mapping relationship between the first sample population and the first health type; take the first sampling frequency as the sampling frequency, and collect a monitoring data sequence of the first sampling indicator of the first sample population within a first time range; hierarchically scale the monitoring data sequence of the first sampling indicator of the first sample population by improving the hierarchical scaling method, and obtain a first health indicator scale of the first sample population; obtain a mapping relationship between the first health type and the first health indicator scale according to the mapping relationship between the first sample population and the first health type and the first health indicator scale of the first sample population;

[0030] The first BP neural network training module is connected to the first health sample construction module, and is used to merge the first health type and the first health indicator scale to obtain a first training sample; input the first training sample into the BP neural network, output the BP neural network weight parameter, and obtain the first BP neural network; initialize the second sampling frequency to the first sampling frequency, initialize the second sampling indicator to the first sampling indicator, and start medical remote monitoring;

[0031] The monitored object sampling diagnosis module is connected to the first BP neural network training module, and is used to collect the monitoring data sequence of the second sampling indicator of the monitored object within a preset time range with the second sampling frequency as the sampling frequency; the monitoring data sequence of the second sampling indicator of the monitored object is graded and scaled by an improved hierarchical scaling method to obtain the second health indicator scale of the monitored object; the second health indicator scale of the monitored object is input into the first BP neural network to obtain the health type of the monitored object; the health type of the monitored object is cloned into the health type library of the monitored object; the second sampling frequency is corrected by the second sampling frequency correction method, and the second sampling frequency is assigned to the corrected second sampling frequency; it is checked whether the number of the same health types contained in the health type library of the monitored object reaches the preset first health number threshold value, and if not, the monitored object sampling diagnosis module is entered again;

[0032] The monitoring data transmission module is connected to the sampling diagnosis module of the monitored object, and is used to transmit the health type library of the monitored object to medical personnel through the communication network.

[0033] Furthermore, the first healthy sample construction module includes:

[0034] A first indicator mean array construction module is used to clone the monitoring data sequence of the first sampling indicator of the first sample population into a first monitoring cell matrix; each element in the first monitoring cell matrix is ​​a corresponding monitoring data sequence of the first sampling indicator of the first sample population; average values ​​are calculated for all elements in the first monitoring cell matrix to obtain a first sampling matrix; the first sampling matrix is ​​divided into T first indicator mean arrays by column, where T is the number of columns of the first sampling matrix; a mapping relationship between the first indicator mean array and the first sampling indicator is established;

[0035] The probability density curve cluster construction module of the first indicator mean array is connected to the first indicator mean array construction module, and is used to spread the T first indicator mean arrays to T independent two-dimensional coordinate planes, and perform maximum likelihood estimation fitting respectively to obtain the probability density curve cluster of the first indicator mean array; establish a mapping relationship between the first sampling indicator and the probability density curve cluster of the first indicator mean array;

[0036] The first health indicator scale construction module of the first sample population is connected to the probability density curve cluster construction module of the first indicator mean array, and is used to grade and separate each probability density curve in the probability density curve cluster of the first indicator mean array according to the cumulative distribution to obtain the graded mapping relationship of the first sampling matrix; the first sampling matrix is ​​mapped according to the graded mapping relationship to obtain the first health indicator scale of the first sample population.

[0037] Furthermore, the first health indicator scale construction module of the first sample population includes:

[0038] An overall grading quantity construction module is used to obtain the grading quantity of each probability density curve in the probability density curve cluster of the first indicator mean array by adopting an improved concentration evaluation method; weighted average the grading quantity of each probability density curve in the probability density curve cluster of the first indicator mean array, and round up to obtain the overall grading quantity X;

[0039] The interval point construction module is connected to the overall classification quantity construction module, and is used to start the horizontal coordinate variable from negative infinity for any probability density curve in the probability density curve cluster of the first indicator mean array, and search the integral area of ​​the probability density curve in a positive direction. When the integral area is When , the corresponding horizontal axis variable value is recorded as the i-th interval point of the probability density curve, and i is traversed through all integers in the interval from 1 to 2X to obtain the first interval point to the 2X interval point of the probability density curve;

[0040] The interval point traversal module is connected to the interval point construction module, and is used to traverse all probability density curves in the probability density curve cluster of the first indicator mean value array, and repeat the interval point construction process respectively;

[0041] The interval construction module is connected to the interval point traversal module, and is used to establish a mapping relationship between the probability density curve and the first sampling index; for all the probability density curves, the horizontal coordinates of each probability density curve are divided into 2X+1 intervals with all the interval points of the probability density curve as intervals, and are sequentially named from the first interval to the 2X+1 interval of the first sampling index element corresponding to the probability density curve;

[0042] The mapping connection module is connected to the interval construction module to establish a hierarchical mapping relationship of the first sampling matrix; the input of the hierarchical mapping relationship of the first sampling matrix is ​​the first sampling matrix, and the output is a matrix composed of integers in the range of -X to X; when the value of the element of the first sampling matrix falls in the corresponding mth interval, the element at the same position in the corresponding output matrix is ​​-X+m-1, 1≤m≤2X+1.

[0043] Furthermore, the overall classification amount building module includes:

[0044] A variance calculation module, used to calculate the variable variance corresponding to each probability density curve in the probability density curve cluster of the first indicator mean array;

[0045] The variance logic table mapping module is connected to the variance calculation module and is used to input the variable variance into a preset variance logic mapping table to obtain the classification amount of each probability density curve.

[0046] Furthermore, the monitored object sampling diagnosis module includes:

[0047] A correction number reading module is used to read the correction number;

[0048] The frequency gain calculation module is connected to the correction number reading module to use the second sampling frequency gain formula Obtaining a frequency gain coefficient k, wherein l is the number of corrections, C1 is a preset gain speed constant, C2 is a preset gain amplitude constant, and C3 is a preset gain reference constant;

[0049] The second sampling frequency correction module is connected to the frequency gain calculation module, and is used to multiply the first sampling frequency by the frequency gain coefficient k to obtain a corrected second sampling frequency.

[0050] (3) Beneficial effects

[0051] Compared with the prior art, the present invention has the following beneficial effects:

[0052] 1. The medical remote monitoring sampling data is preprocessed according to the sample statistical laws through the improved hierarchical scaling method, which achieves the following beneficial effects: on the one hand, it can effectively extract key data features through distributed edge computing, reduce data processing volume, and improve system computing power; on the other hand, the scale value is obtained by estimating and fitting the sample data, which reduces the interference of subjective factors; in addition, data preprocessing through the hierarchical scaling method can reduce the interference of abnormal data and improve the robustness of data analysis and diagnosis.

[0053] 2. By adopting the second sampling frequency correction method to adaptively and dynamically correct the sampling frequency, the pertinence of system data processing can be improved and the data processing efficiency can be improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] Figure 1 This is a flowchart of a medical remote monitoring method according to Embodiment 1 of the present invention;

[0055] Figure 2 This is a module block diagram of a medical remote monitoring system according to Embodiment 2 of the present invention. DETAILED DESCRIPTION

[0056] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0057] Example 1: Figure 1 As shown, this embodiment provides a medical remote monitoring method, the method comprising:

[0058] Step 1: read the first file data, the first file data includes a first health type, a first sampling indicator and a first sampling frequency; the first health type, the first sampling indicator and the first sampling frequency are all arrays, and the first sampling indicator and the first sampling frequency have a one-to-one mapping relationship; construct a first sample population, and establish a mapping relationship between the first sample population and the first health type; take the first sampling frequency as the sampling frequency, and collect a monitoring data sequence of the first sampling indicator of the first sample population within a first time range; hierarchically scale the monitoring data sequence of the first sampling indicator of the first sample population by improving the hierarchical scaling method, and obtain the first health indicator scale of the first sample population; according to the mapping relationship between the first sample population and the first health type and the first health indicator scale of the first sample population, obtain the mapping relationship between the first health type and the first health indicator scale;

[0059] The first archival data needs to be pre-set. The first health type is an array, and the elements of the array should include "health", which may include any one or more of diseases such as "hypertension" and "diabetes". The first sampling indicator is an array, and the elements of the array may include physical indicators such as "blood pressure", "blood sugar", and "heartbeat". The first sampling frequency is an array, and each element of the array is the sampling frequency of the corresponding indicator data in the first sampling indicator. The mapping relationship between the first sample population and the first health type is pre-set. Exemplarily, if one person in the first sample population suffers from both hypertension and diabetes, it is necessary to connect the element corresponding to the person in the first sample population with the two elements of "hypertension" and "diabetes" in the first health type through a mapping relationship. The purpose of the improved hierarchical scaling method is to map and scale the massive monitoring data sequence of the first sampling indicator according to a statistical method to reduce the amount of data processing.

[0060] Step 2: Merge the first health type and the first health indicator scale to obtain a first training sample; input the first training sample into a BP neural network, output a BP neural network weight parameter, and obtain a first BP neural network; initialize the second sampling frequency to the first sampling frequency, initialize the second sampling indicator to the first sampling indicator, and start medical remote monitoring;

[0061] The first BP neural network used in this embodiment is a three-layer neural network, including an input layer, a hidden layer and an output layer. The number of neurons in the input layer is determined according to the number of output levels of the hierarchical scaling method. The number of neurons in the output layer is determined according to the number of elements of the first health type. The activation function in the first BP neural network adopts the Sigmoid function. By using the first training sample for training and learning, the weight parameters of the first BP neural network are optimized, thereby realizing the pattern recognition function.

[0062] Step 3: Using the second sampling frequency as the sampling frequency, collect the monitoring data sequence of the second sampling indicator of the monitored object within a preset time range; hierarchically scale the monitoring data sequence of the second sampling indicator of the monitored object by improving the hierarchical scaling method to obtain the second health indicator scale of the monitored object; input the second health indicator scale of the monitored object into the first BP neural network to obtain the health type of the monitored object; clone the health type of the monitored object into the health type library of the monitored object; use the second sampling frequency correction method to correct the second sampling frequency, and assign the second sampling frequency to the corrected second sampling frequency; check whether the number of the same health types contained in the health type library of the monitored object reaches the preset first health number threshold value, if not, repeat the step 3;

[0063] The time range needs to be pre-set according to actual conditions. If the time range is set too short, it may be difficult for the calculation to converge, and if the time range is set too long, it will affect the calculation efficiency. The preferred value of the time range can be obtained by simulation calculation and particle swarm optimization, and the preferred value is used as the pre-set time range. By way of example, the pre-set time range in this embodiment is 10 minutes. The first healthy number threshold value needs to be pre-set according to actual conditions. If the first healthy number threshold value is set too low, it may cause inaccurate calculations, and if the first healthy number threshold value is set too high, it will affect the calculation efficiency. The preferred value of the first healthy number threshold value can be obtained by simulation calculation and particle swarm optimization, and the preferred value is used as the pre-set first healthy number threshold value. By way of example, the pre-set first healthy number threshold value in this embodiment is 3.

[0064] Step 4: Transmit the health type library of the monitored object to medical personnel via a communication network.

[0065] Since the massive amount of monitoring data has been processed by edge computing, the bandwidth and latency performance of the communication network are not required to be high. However, since the health type library of the monitored object involves personal privacy, the security requirements of the communication network are relatively high. Exemplarily, this embodiment uses a 4G cellular network based on the DES data encryption algorithm as the communication network. Furthermore, the method of obtaining the first health indicator scale of the first sample population by graded scaling the monitoring data sequence of the first sampling indicator of the first sample population by improving the graded scaling method includes:

[0066] Step 11: clone the monitoring data sequence of the first sampling indicator of the first sample population into the first monitoring cell matrix; each element in the first monitoring cell matrix is ​​the corresponding monitoring data sequence of the first sampling indicator of the first sample population; calculate the average value of all elements in the first monitoring cell matrix to obtain the first sampling matrix; divide the first sampling matrix into T first indicator mean arrays by column, where T is the number of columns of the first sampling matrix; establish a mapping relationship between the first indicator mean array and the first sampling indicator;

[0067] Step 12: Spread the T first indicator mean arrays to T independent two-dimensional coordinate planes respectively, and perform maximum likelihood estimation fitting respectively to obtain probability density curve clusters of the first indicator mean arrays; establish a mapping relationship between the first sampling indicator and the probability density curve cluster of the first indicator mean array;

[0068] Step 13: hierarchically separate each probability density curve in the probability density curve cluster of the first indicator mean array according to the cumulative distribution to obtain a hierarchical mapping relationship of the first sampling matrix; map the first sampling matrix according to the hierarchical mapping relationship to obtain the first health indicator scale of the first sample population.

[0069] Furthermore, the method of hierarchically separating each probability density curve in the probability density curve cluster of the first indicator mean array according to the cumulative distribution to obtain the hierarchical mapping relationship of the first sampling matrix includes:

[0070] Step 131: using an improved concentration evaluation method to obtain the classification value of each probability density curve in the probability density curve cluster of the first indicator mean array; taking a weighted average of the classification values ​​of each probability density curve in the probability density curve cluster of the first indicator mean array, and rounding up, to obtain an overall classification value X;

[0071] If the overall classification amount is too large, it will increase the computing pressure and reduce the fault tolerance of abnormal data. If the overall classification amount is too small, it will affect the accuracy of health type diagnosis and may cause the first BP neural network to fail to converge. In order to select a reasonable overall classification amount, the overall classification amount is obtained based on the distribution characteristics of the sampled data by adopting an improved concentration evaluation method. Exemplarily, the overall classification amount obtained in this embodiment is X=2.

[0072] Step 132: For any probability density curve in the probability density curve cluster of the first indicator mean array, let the horizontal axis variable start from negative infinity and search the integral area of ​​the probability density curve in a positive direction. When the integral area is When , the corresponding horizontal axis variable value is recorded as the i-th interval point of the probability density curve, and i is traversed through all integers in the interval from 1 to 2X to obtain the first interval point to the 2X interval point of the probability density curve;

[0073] Exemplarily, in this embodiment, the horizontal axis variable is respectively started from negative infinity, and the integral area of ​​any probability density curve in the probability density curve cluster of the first indicator mean array is searched forward. When the integral area is 1 / 5, the corresponding horizontal axis variable value is recorded as the first interval point of the probability density curve; the horizontal axis variable is respectively started from negative infinity, and the integral area of ​​any probability density curve in the probability density curve cluster of the first indicator mean array is searched forward. When the integral area is 2 / 5, the corresponding horizontal axis variable value is recorded as the second interval point of the probability density curve; the horizontal axis variable is respectively started from negative infinity, and the integral area of ​​any probability density curve in the probability density curve cluster of the first indicator mean array is searched forward. When the integral area is 3 / 5, the corresponding horizontal axis variable value is recorded as the third interval point of the probability density curve; the horizontal axis variable is respectively started from negative infinity, and the integral area of ​​any probability density curve in the probability density curve cluster of the first indicator mean array is searched forward. When the integral area is 4 / 5, the corresponding horizontal axis variable value is recorded as the fourth interval point of the probability density curve.

[0074] Step 133: traverse all probability density curves in the probability density curve cluster of the first indicator mean value array, and repeat the interval point construction process respectively;

[0075] Step 134: Establishing a mapping relationship between the probability density curve and the first sampling index; for all the probability density curves, dividing the abscissa of each probability density curve into 2X+1 intervals with all the interval points of the probability density curve as intervals, and sequentially naming them from the first interval to the 2X+1 interval of the first sampling index element corresponding to the probability density curve;

[0076] Exemplarily, in this embodiment, the interval in which the horizontal coordinate variable range is less than or equal to the first interval point is defined as the first interval of the first sampling indicator element; the interval in which the horizontal coordinate variable range is greater than the first interval point and less than or equal to the second interval point is defined as the second interval of the first sampling indicator element; the interval in which the horizontal coordinate variable range is greater than the second interval point and less than or equal to the third interval point is defined as the third interval of the first sampling indicator element; the interval in which the horizontal coordinate variable range is greater than the third interval point and less than or equal to the fourth interval point is defined as the fourth interval of the first sampling indicator element; the interval in which the horizontal coordinate variable range is greater than the fourth interval point is defined as the fifth interval of the first sampling indicator element.

[0077] Step 135: Establish a hierarchical mapping relationship of the first sampling matrix; the input of the hierarchical mapping relationship of the first sampling matrix is ​​the first sampling matrix, and the output is a matrix composed of integers in the range of -X to X; when the value of the element of the first sampling matrix falls within the corresponding mth interval, the element at the same position in the corresponding output matrix is ​​-X+m-1, 1≤m≤2X+1.

[0078] Exemplarily, in this embodiment, the hierarchical mapping input of the first sampling matrix is ​​the first sampling matrix, and the output is a matrix composed of -2, -1, 0, 1, and 2; the method for determining the hierarchical mapping output of the first sampling matrix is ​​to scatter the first sampling indicator element on the horizontal axis of the corresponding probability density curve, and if it falls into the first interval of the first sampling indicator element, output -2; if it falls into the second interval of the first sampling indicator element, output -1; if it falls into the third interval of the first sampling indicator element, output 0; if it falls into the fourth interval of the first sampling indicator element, output 1; if it falls into the fifth interval of the first sampling indicator element, output 2.

[0079] Furthermore, the method of using the improved concentration evaluation method to obtain the classification amount of each probability density curve in the probability density curve cluster of the first indicator mean array includes:

[0080] Step 1311: Calculate the variable variance corresponding to each probability density curve in the probability density curve cluster of the first indicator mean array;

[0081] Step 1312: Input the variable variance into a preset variance logic mapping table to obtain the classification amount of each probability density curve.

[0082] The variance logic mapping table needs to be preset according to the system computing power.

[0083] Furthermore, the method of correcting the second sampling frequency using the second sampling frequency correction method includes:

[0084] Step 31: Read the number of corrections;

[0085] Step 32: Use the second sampling frequency gain formula Obtaining a frequency gain coefficient k, wherein l is the number of corrections, C1 is a preset gain speed constant, C2 is a preset gain amplitude constant, and C3 is a preset gain reference constant;

[0086] The gain speed constant, gain amplitude constant, and gain reference constant all need to be set in advance, and the preferred values ​​of the gain speed constant, gain amplitude constant, and gain reference constant can be obtained through simulation experiments. If the gain reference constant is lower, the lower limit of the sampling frequency is lower. If the gain speed constant is larger, the sampling frequency changes more slowly. If the gain amplitude constant is larger, the upper limit of the sampling frequency is higher. Exemplarily, this embodiment selects C1=20, C2=20, and C3=0.2. This method can achieve a step-by-step change of the sampling frequency from low frequency to high frequency in accordance with the law that meets actual needs.

[0087] Step 33: multiply the first sampling frequency by the frequency gain coefficient k to obtain a corrected second sampling frequency.

[0088] Embodiment 2: Based on the same inventive concept, Figure 2 As shown, this embodiment also provides a medical remote monitoring system, the system comprising:

[0089] A first health sample construction module is used to read first archive data, wherein the first archive data includes a first health type, a first sampling indicator and a first sampling frequency; the first health type, the first sampling indicator and the first sampling frequency are all arrays, and the first sampling indicator has a one-to-one mapping relationship with the first sampling frequency; construct a first sample population, and establish a mapping relationship between the first sample population and the first health type; take the first sampling frequency as the sampling frequency, and collect a monitoring data sequence of the first sampling indicator of the first sample population within a first time range; hierarchically scale the monitoring data sequence of the first sampling indicator of the first sample population by improving the hierarchical scaling method, and obtain a first health indicator scale of the first sample population; obtain a mapping relationship between the first health type and the first health indicator scale according to the mapping relationship between the first sample population and the first health type and the first health indicator scale of the first sample population;

[0090] The first BP neural network training module is connected to the first health sample construction module, and is used to merge the first health type and the first health indicator scale to obtain a first training sample; input the first training sample into the BP neural network, output the BP neural network weight parameter, and obtain the first BP neural network; initialize the second sampling frequency to the first sampling frequency, initialize the second sampling indicator to the first sampling indicator, and start medical remote monitoring;

[0091] The monitored object sampling diagnosis module is connected to the first BP neural network training module, and is used to collect the monitoring data sequence of the second sampling indicator of the monitored object within a preset time range with the second sampling frequency as the sampling frequency; the monitoring data sequence of the second sampling indicator of the monitored object is graded and scaled by an improved hierarchical scaling method to obtain the second health indicator scale of the monitored object; the second health indicator scale of the monitored object is input into the first BP neural network to obtain the health type of the monitored object; the health type of the monitored object is cloned into the health type library of the monitored object; the second sampling frequency is corrected by the second sampling frequency correction method, and the second sampling frequency is assigned to the corrected second sampling frequency; it is checked whether the number of the same health types contained in the health type library of the monitored object reaches the preset first health number threshold value, and if not, the monitored object sampling diagnosis module is entered again;

[0092] The monitoring data transmission module is connected to the sampling diagnosis module of the monitored object, and is used to transmit the health type library of the monitored object to medical personnel through the communication network.

[0093] Furthermore, the first healthy sample construction module includes:

[0094] A first indicator mean array construction module is used to clone the monitoring data sequence of the first sampling indicator of the first sample population into a first monitoring cell matrix; each element in the first monitoring cell matrix is ​​a corresponding monitoring data sequence of the first sampling indicator of the first sample population; average values ​​are calculated for all elements in the first monitoring cell matrix to obtain a first sampling matrix; the first sampling matrix is ​​divided into T first indicator mean arrays by column, where T is the number of columns of the first sampling matrix; a mapping relationship between the first indicator mean array and the first sampling indicator is established;

[0095] The probability density curve cluster construction module of the first indicator mean array is connected to the first indicator mean array construction module, and is used to spread the T first indicator mean arrays to T independent two-dimensional coordinate planes, and perform maximum likelihood estimation fitting respectively to obtain the probability density curve cluster of the first indicator mean array; establish a mapping relationship between the first sampling indicator and the probability density curve cluster of the first indicator mean array;

[0096] The first health indicator scale construction module of the first sample population is connected to the probability density curve cluster construction module of the first indicator mean array, and is used to grade and separate each probability density curve in the probability density curve cluster of the first indicator mean array according to the cumulative distribution to obtain the graded mapping relationship of the first sampling matrix; the first sampling matrix is ​​mapped according to the graded mapping relationship to obtain the first health indicator scale of the first sample population.

[0097] Furthermore, the first health indicator scale construction module of the first sample population includes:

[0098] An overall grading quantity construction module is used to obtain the grading quantity of each probability density curve in the probability density curve cluster of the first indicator mean array by adopting an improved concentration evaluation method; weighted average the grading quantity of each probability density curve in the probability density curve cluster of the first indicator mean array, and round up to obtain the overall grading quantity X;

[0099] The interval point construction module is connected to the overall classification quantity construction module, and is used to start the horizontal coordinate variable from negative infinity for any probability density curve in the probability density curve cluster of the first indicator mean array, and search the integral area of ​​the probability density curve in a positive direction. When the integral area is When , the corresponding horizontal axis variable value is recorded as the i-th interval point of the probability density curve, and i is traversed through all integers in the interval from 1 to 2X to obtain the first interval point to the 2X interval point of the probability density curve;

[0100] The interval point traversal module is connected to the interval point construction module, and is used to traverse all probability density curves in the probability density curve cluster of the first indicator mean value array, and repeat the interval point construction process respectively;

[0101] The interval construction module is connected to the interval point traversal module, and is used to establish a mapping relationship between the probability density curve and the first sampling index; for all the probability density curves, the horizontal coordinates of each probability density curve are divided into 2X+1 intervals with all the interval points of the probability density curve as intervals, and are sequentially named from the first interval to the 2X+1 interval of the first sampling index element corresponding to the probability density curve;

[0102] The mapping connection module is connected to the interval construction module to establish a hierarchical mapping relationship of the first sampling matrix; the input of the hierarchical mapping relationship of the first sampling matrix is ​​the first sampling matrix, and the output is a matrix composed of integers in the range of -X to X; when the value of the element of the first sampling matrix falls in the corresponding mth interval, the element at the same position in the corresponding output matrix is ​​-X+m-1, 1≤m≤2X+1.

[0103] Furthermore, the overall classification amount building module includes:

[0104] A variance calculation module, used to calculate the variable variance corresponding to each probability density curve in the probability density curve cluster of the first indicator mean array;

[0105] The variance logic table mapping module is connected to the variance calculation module and is used to input the variable variance into a preset variance logic mapping table to obtain the classification amount of each probability density curve.

[0106] Furthermore, the monitored object sampling diagnosis module includes:

[0107] A correction number reading module is used to read the correction number;

[0108] The frequency gain calculation module is connected to the correction number reading module to use the second sampling frequency gain formula Obtaining a frequency gain coefficient k, wherein l is the number of corrections, C1 is a preset gain speed constant, C2 is a preset gain amplitude constant, and C3 is a preset gain reference constant;

[0109] The second sampling frequency correction module is connected to the frequency gain calculation module, and is used to multiply the first sampling frequency by the frequency gain coefficient k to obtain a corrected second sampling frequency.

[0110] It should be noted that, regarding the system in the above embodiment, the specific manner in which each module performs operations has been described in detail in the embodiment of the method, and will not be elaborated here.

[0111] Finally, it should be noted that: Although the present invention has been described in detail with reference to the aforementioned embodiments, it is still possible for those skilled in the art to modify the technical solutions described in the aforementioned embodiments, or to make equivalent substitutions for some of the technical features therein. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A medical remote monitoring method, characterized in that: The method comprises: Step 1: read the first file data, the first file data includes a first health type, a first sampling indicator and a first sampling frequency; the first health type, the first sampling indicator and the first sampling frequency are all arrays, and the first sampling indicator and the first sampling frequency have a one-to-one mapping relationship; construct a first sample population, and establish a mapping relationship between the first sample population and the first health type; take the first sampling frequency as the sampling frequency, and collect a monitoring data sequence of the first sampling indicator of the first sample population within a first time range; hierarchically scale the monitoring data sequence of the first sampling indicator of the first sample population by improving the hierarchical scaling method, and obtain the first health indicator scale of the first sample population; according to the mapping relationship between the first sample population and the first health type and the first health indicator scale of the first sample population, obtain the mapping relationship between the first health type and the first health indicator scale; Step 2: Merge the first health type and the first health indicator scale to obtain a first training sample; input the first training sample into a BP neural network, output a BP neural network weight parameter, and obtain a first BP neural network; initialize the second sampling frequency to the first sampling frequency, initialize the second sampling indicator to the first sampling indicator, and start medical remote monitoring; Step 3: Using the second sampling frequency as the sampling frequency, collect the monitoring data sequence of the second sampling indicator of the monitored object within a preset time range; hierarchically scale the monitoring data sequence of the second sampling indicator of the monitored object by improving the hierarchical scaling method to obtain the second health indicator scale of the monitored object; input the second health indicator scale of the monitored object into the first BP neural network to obtain the health type of the monitored object; clone the health type of the monitored object into the health type library of the monitored object; use the second sampling frequency correction method to correct the second sampling frequency, and assign the second sampling frequency to the corrected second sampling frequency; check whether the number of the same health types contained in the health type library of the monitored object reaches the preset first health number threshold value, if not, repeat the step 3; Step 4: Transmitting the health type library of the monitored object to medical personnel via a communication network; The method of performing hierarchical scaling on the monitoring data sequence of the first sampling indicator of the first sample population by improving the hierarchical scaling method to obtain the first health indicator scale of the first sample population includes: Step 11: clone the monitoring data sequence of the first sampling indicator of the first sample population into the first monitoring cell matrix; each element in the first monitoring cell matrix is ​​the corresponding monitoring data sequence of the first sampling indicator of the first sample population; calculate the average value of all elements in the first monitoring cell matrix to obtain the first sampling matrix; divide the first sampling matrix into T first indicator mean arrays by column, where T is the number of columns of the first sampling matrix; establish a mapping relationship between the first indicator mean array and the first sampling indicator; Step 12: Spread the T first indicator mean arrays to T independent two-dimensional coordinate planes respectively, and perform maximum likelihood estimation fitting respectively to obtain probability density curve clusters of the first indicator mean arrays; establish a mapping relationship between the first sampling indicator and the probability density curve cluster of the first indicator mean array; Step 13: hierarchically separate each probability density curve in the probability density curve cluster of the first indicator mean array according to the cumulative distribution to obtain a hierarchical mapping relationship of the first sampling matrix; map the first sampling matrix according to the hierarchical mapping relationship to obtain a first health indicator scale of the first sample population; The method of hierarchically separating each probability density curve in the probability density curve cluster of the first indicator mean array according to the cumulative distribution to obtain the hierarchical mapping relationship of the first sampling matrix includes: Step 131: using an improved concentration evaluation method to obtain the classification value of each probability density curve in the probability density curve cluster of the first indicator mean array; taking a weighted average of the classification values ​​of each probability density curve in the probability density curve cluster of the first indicator mean array, and rounding up, to obtain an overall classification value X; Step 132: For any probability density curve in the probability density curve cluster of the first indicator mean array, let the horizontal axis variable start from negative infinity and search the integral area of ​​the probability density curve in a positive direction. When the integral area is When , the corresponding horizontal axis variable value is recorded as the probability density curve i Interval point, i Traverse all integers in the interval from 1 to 2X to obtain the first interval point to the 2X interval point of the probability density curve; Step 133: traverse all probability density curves in the probability density curve cluster of the first indicator mean value array, and repeat the interval point construction process respectively; Step 134: Establishing a mapping relationship between the probability density curve and the first sampling index; for all the probability density curves, dividing the abscissa of each probability density curve into 2X+1 intervals with all the interval points of the probability density curve as intervals, and sequentially naming them from the first interval to the 2X+1 interval of the first sampling index element corresponding to the probability density curve; Step 135: Establishing a hierarchical mapping relationship of the first sampling matrix; the input of the hierarchical mapping relationship of the first sampling matrix is ​​the first sampling matrix, and the output is a matrix composed of integers in the range of -X to X; when the value of the element of the first sampling matrix falls within the corresponding mth interval, the element at the same position in the corresponding output matrix is ​​-X+m-1, 1≤m≤2X+1; The method of using the improved concentration evaluation method to obtain the classification amount of each probability density curve in the probability density curve cluster of the first indicator mean array includes: Step 1311: Calculate the variable variance corresponding to each probability density curve in the probability density curve cluster of the first indicator mean array; Step 1312: inputting the variable variance into a preset variance logistic mapping table to obtain the classification amount of each probability density curve; The method for correcting the second sampling frequency by using the second sampling frequency correction method comprises: Step 31: Read the number of corrections; Step 32: Use the second sampling frequency gain formula Obtaining a frequency gain coefficient k, wherein l is the number of corrections, C1 is a preset gain speed constant, C2 is a preset gain amplitude constant, and C3 is a preset gain reference constant; Step 33: multiply the first sampling frequency by the frequency gain coefficient k to obtain a corrected second sampling frequency.

2. A medical remote monitoring system, characterized in that: The system comprises: A first health sample construction module is used to read first archive data, wherein the first archive data includes a first health type, a first sampling indicator and a first sampling frequency; the first health type, the first sampling indicator and the first sampling frequency are all arrays, and the first sampling indicator has a one-to-one mapping relationship with the first sampling frequency; construct a first sample population, and establish a mapping relationship between the first sample population and the first health type; take the first sampling frequency as the sampling frequency, and collect a monitoring data sequence of the first sampling indicator of the first sample population within a first time range; hierarchically scale the monitoring data sequence of the first sampling indicator of the first sample population by improving the hierarchical scaling method, and obtain a first health indicator scale of the first sample population; obtain a mapping relationship between the first health type and the first health indicator scale according to the mapping relationship between the first sample population and the first health type and the first health indicator scale of the first sample population; The first BP neural network training module is connected to the first health sample construction module, and is used to merge the first health type and the first health indicator scale to obtain a first training sample; input the first training sample into the BP neural network, output the BP neural network weight parameter, and obtain the first BP neural network; initialize the second sampling frequency to the first sampling frequency, initialize the second sampling indicator to the first sampling indicator, and start medical remote monitoring; The monitored object sampling diagnosis module is connected to the first BP neural network training module, and is used to collect the monitoring data sequence of the second sampling indicator of the monitored object within a preset time range with the second sampling frequency as the sampling frequency; the monitoring data sequence of the second sampling indicator of the monitored object is graded and scaled by an improved hierarchical scaling method to obtain the second health indicator scale of the monitored object; the second health indicator scale of the monitored object is input into the first BP neural network to obtain the health type of the monitored object; the health type of the monitored object is cloned into the health type library of the monitored object; the second sampling frequency is corrected by the second sampling frequency correction method, and the second sampling frequency is assigned to the corrected second sampling frequency; it is checked whether the number of the same health types contained in the health type library of the monitored object reaches the preset first health number threshold value, if not, the monitored object sampling diagnosis module is entered again; The monitoring data transmission module is connected to the monitored object sampling diagnosis module, and is used to transmit the health type library of the monitored object to the medical staff through the communication network; The first healthy sample construction module includes: A first indicator mean array construction module is used to clone the monitoring data sequence of the first sampling indicator of the first sample population into a first monitoring cell matrix; each element in the first monitoring cell matrix is ​​a corresponding monitoring data sequence of the first sampling indicator of the first sample population; average values ​​are calculated for all elements in the first monitoring cell matrix to obtain a first sampling matrix; the first sampling matrix is ​​divided into T first indicator mean arrays by column, where T is the number of columns of the first sampling matrix; a mapping relationship between the first indicator mean array and the first sampling indicator is established; The probability density curve cluster construction module of the first indicator mean array is connected to the first indicator mean array construction module, and is used to spread the T first indicator mean arrays to T independent two-dimensional coordinate planes, and perform maximum likelihood estimation fitting respectively to obtain the probability density curve cluster of the first indicator mean array; establish a mapping relationship between the first sampling indicator and the probability density curve cluster of the first indicator mean array; The first health indicator scale construction module of the first sample population is connected to the probability density curve cluster construction module of the first indicator mean array, and is used to hierarchically separate each probability density curve in the probability density curve cluster of the first indicator mean array according to the cumulative distribution to obtain the hierarchical mapping relationship of the first sampling matrix; the first sampling matrix is ​​mapped according to the hierarchical mapping relationship to obtain the first health indicator scale of the first sample population; The first health indicator scale construction module of the first sample population includes: An overall grading quantity construction module is used to obtain the grading quantity of each probability density curve in the probability density curve cluster of the first indicator mean array by adopting an improved concentration evaluation method; weighted average the grading quantity of each probability density curve in the probability density curve cluster of the first indicator mean array, and round up to obtain the overall grading quantity X; The interval point construction module is connected to the overall classification quantity construction module, and is used to start the horizontal coordinate variable from negative infinity for any probability density curve in the probability density curve cluster of the first indicator mean array, and search the integral area of ​​the probability density curve in a positive direction. When the integral area is When , the corresponding horizontal axis variable value is recorded as the probability density curve i Interval point, i Traverse all integers in the interval from 1 to 2X to obtain the first interval point to the 2X interval point of the probability density curve; The interval point traversal module is connected to the interval point construction module, and is used to traverse all probability density curves in the probability density curve cluster of the first indicator mean value array, and repeat the interval point construction process respectively; The interval construction module is connected to the interval point traversal module, and is used to establish a mapping relationship between the probability density curve and the first sampling index; for all the probability density curves, the horizontal coordinates of each probability density curve are divided into 2X+1 intervals with all the interval points of the probability density curve as intervals, and are sequentially named from the first interval to the 2X+1 interval of the first sampling index element corresponding to the probability density curve; The mapping connection module is connected to the interval construction module, and is used to establish a hierarchical mapping relationship of the first sampling matrix; the input of the hierarchical mapping relationship of the first sampling matrix is ​​the first sampling matrix, and the output is a matrix composed of integers in the range of -X to X; when the value of the element of the first sampling matrix falls in the corresponding mth interval, the element at the same position in the corresponding output matrix is ​​-X+m-1, 1≤m≤2X+1; The overall classification volume building block includes: A variance calculation module, used to calculate the variable variance corresponding to each probability density curve in the probability density curve cluster of the first indicator mean array; The variance logic table mapping module is connected to the variance calculation module and is used to input the variable variance into a preset variance logic mapping table to obtain the classification amount of each probability density curve; The monitored object sampling diagnosis module comprises: A correction number reading module is used to read the correction number; The frequency gain calculation module is connected to the correction number reading module to use the second sampling frequency gain formula Obtaining a frequency gain coefficient k, wherein l is the number of corrections, C1 is a preset gain speed constant, C2 is a preset gain amplitude constant, and C3 is a preset gain reference constant; The second sampling frequency correction module is connected to the frequency gain calculation module, and is used to multiply the first sampling frequency by the frequency gain coefficient k to obtain a corrected second sampling frequency.

Images