Remote moxibustion monitoring and health data management system based on cloud platform

Abstract

The present invention relates to the field of data processing technology, and in particular to a remote moxibustion monitoring and health data management system based on a cloud platform. The system comprises a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, the following steps are implemented: respectively obtaining a monitoring data distribution feature sequence of each moxibustion patient with the same disease condition; in the process of performing density peak clustering on all monitoring data distribution feature sequences, obtaining the number of pre-assigned data points of each cluster center, obtaining an initial cluster cluster and unassigned data points, respectively optimizing the metric distance between the unassigned data points and each cluster center, so as to assign all the unassigned data points to the initial cluster cluster, obtaining a final cluster cluster, and classifying and managing the moxibustion patients according to the final cluster cluster, thereby improving the accuracy of density peak clustering and ensuring accurate classification and management of the moxibustion patients.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular to a remote moxibustion monitoring and health data management system based on a cloud platform. Background Art

[0002] The cloud platform is a computing resource provision method based on the Internet. It can provide a variety of services such as storage, computing, networking and applications. Through this platform, the relevant monitoring data of the patient's moxibustion process can be transmitted to the cloud analysis system through the network. Through the analysis and calculation of the system, the classification management and analysis of moxibustion patients can be realized, thereby providing doctors or health managers with personalized assessment and guidance on the health status of moxibustion patients, which can effectively prevent or promptly discover potential health problems of moxibustion patients.

[0003] In the prior art, a density peaks clustering (DPC) algorithm is used to classify and manage the monitoring data of moxibustion patients. Specifically, after the DPC algorithm selects the cluster center point by means of local density peaks, the remaining data points in the sample points will be divided into different cluster centers. However, in the division process, the farther the data points are from the cluster center, the more likely they are to be divided incorrectly. For example, the division between the edge points of a cluster and the cluster clusters is most likely to be divided incorrectly, resulting in shape differences between the cluster clusters corresponding to each cluster center, which ultimately affects the accuracy of the classification management of moxibustion patients.

[0004] Therefore, how to improve the accuracy of classification management of moxibustion patients' monitoring data using the DPC algorithm has become an urgent problem to be solved. Summary of the invention

[0005] In view of this, an embodiment of the present invention provides a remote moxibustion monitoring and health data management system based on a cloud platform to solve the problem of how to improve the classification management of moxibustion patient monitoring data using a DPC algorithm.

[0006] In an embodiment of the present invention, a remote moxibustion monitoring and health data management system based on a cloud platform is provided, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the processor implements the following steps when executing the computer program:

[0007] The multidimensional monitoring data of each moxibustion patient with the same condition in a treatment process are obtained respectively, and the median corresponding to each dimension is obtained in all the multidimensional monitoring data, and the monitoring data distribution feature sequence of each moxibustion patient is obtained according to the difference between the multidimensional monitoring data of each moxibustion patient and the median corresponding to each dimension;

[0008] In the process of performing density peak clustering on all monitoring data distribution feature sequences, the metric distance between every two data points is obtained, and at least two cluster centers are obtained according to all metric distances and a preset cutoff distance, and the number of previously allocated data points of each cluster center is obtained respectively;

[0009] Based on the number of previously assigned data points of each cluster center, initial clusters and unassigned data points are obtained, and for any unassigned data point, the metric distance between the unassigned data point and each cluster center is optimized to obtain a final metric distance between the unassigned data point and each cluster center;

[0010] According to the final metric distance between each of the unassigned data points and each of the cluster centers, all the unassigned data points are assigned to the initial clusters to obtain final clusters, and moxibustion patients are classified and managed according to the final clusters.

[0011] Preferably, the step of obtaining a distribution characteristic sequence of monitoring data of each moxibustion patient includes:

[0012] For any moxibustion patient, the ratio between the monitoring data of each dimension of the moxibustion patient and the median of the corresponding dimension is obtained to form a distribution feature sequence of the monitoring data of the moxibustion patient.

[0013] Preferably, obtaining the metric distance between every two data points includes:

[0014] Calculate the standard deviation of all elements in the monitoring data distribution feature sequence corresponding to each data point respectively; for any two data points, calculate the absolute value of the difference between the standard deviations corresponding to the two data points, and calculate the addition result between the absolute value of the difference and a preset value; calculate the absolute value of the difference between the elements corresponding to the two data points in each dimension respectively, and obtain the average absolute value of the difference; obtain the metric distance between the two data points according to the product between the average absolute value of the difference and the addition result.

[0015] Preferably, the obtaining of the number of pre-assigned data points of each cluster center respectively includes:

[0016] For any cluster center, other data points within the cutoff distance are taken as neighboring data points of the cluster center, and the average metric distance is obtained according to the metric distance between each of the neighboring data points and the cluster center. The average local density is obtained according to the local density of each of the neighboring data points, and the product of the reciprocal of the average metric distance and the average local density is weighted normalized to obtain the local data distribution compactness of the cluster center;

[0017] Obtain the ratio between the total number of cluster centers and the total number of all data points, obtain the multiplication result between the ratio and the preset hyperparameter, round the product of the multiplication result and the local data distribution density to obtain the number of pre-allocated data points of the cluster center.

[0018] Preferably, respectively optimizing the metric distance between the unassigned data point and each of the cluster centers to obtain the final metric distance between the unassigned data point and each of the cluster centers comprises:

[0019] For any cluster center, the outermost data point of the initial cluster cluster where the cluster center is located is obtained, and the metric distance between each of the outermost data points and the cluster center is used to form a reference distance sequence; after adding the unassigned data point to the initial cluster cluster where the cluster center is located, the outermost data point of the initial cluster cluster is obtained, and the metric distance between each of the outermost data points and the cluster center is used to form a target distance sequence, and the DTW distance between the reference distance sequence and the target distance sequence is obtained;

[0020] In the initial clustering cluster where the cluster center is located, a data point closest to the unassigned data point is obtained as a first reference point, an absolute value of a local density difference and a metric distance between the unassigned data point and the first reference point are obtained, and a first ratio of an absolute value of a local density difference and a metric distance between the unassigned data point and the first reference point is obtained;

[0021] Acquire the data point closest to the first reference point as the second reference point, obtain the absolute value of the local density difference and the metric distance between the first reference point and the second reference point, and obtain a second ratio of the absolute value of the local density difference and the metric distance between the first reference point and the second reference point;

[0022] Obtaining an absolute value of a difference between the first ratio and the second ratio, normalizing the product of the absolute value of the difference and the DTW distance to obtain a normalized value, and obtaining a confidence level of adding the unassigned data point to the initial clustering cluster where the cluster center is located according to a difference between a constant 1 and the normalized value;

[0023] The confidence level is used to optimize the metric distance between the unassigned data point and the cluster center, and a final metric distance is obtained accordingly.

[0024] Preferably, the optimizing the metric distance between the unassigned data point and the cluster center by using the confidence level to obtain the final metric distance includes:

[0025] The difference between the constant 1 and the confidence level is obtained, and the product of the metric distance between the unassigned data point and the cluster center and the difference is used as the final metric distance.

[0026] Compared with the prior art, the embodiments of the present invention have the following beneficial effects:

[0027] When the present invention classifies and manages moxibustion patients with the same condition through a density peak clustering algorithm, a difference analysis is performed through multiple monitoring data of moxibustion patients to obtain the metric distance between any two moxibustion patients. In order to avoid inaccurate division caused by differences in cluster shapes in traditional density peak clustering, after determining the cluster center, the number of pre-assigned data points is first determined for each cluster center, and an initial cluster cluster is obtained according to the number of pre-assigned data points. Then, for the remaining unassigned data points, the metric distance between each unassigned data point and the cluster center is optimized according to the shape of each initial cluster cluster and the local density regularity change of the added unassigned data points. Based on the optimized metric distance, the unassigned data points are added to the initial cluster cluster, which improves the accuracy of density peak clustering, and further improves the accuracy of classified management of moxibustion patients according to the final cluster cluster. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, 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 labor.

[0029] Figure 1 This is a flow chart of a remote moxibustion monitoring and health data management method based on a cloud platform provided in Embodiment 1 of the present invention;

[0030] Figure 2 A schematic diagram of allocating other data points to a cluster center provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0031] Embodiments of the present disclosure are described in detail below, and examples of the embodiments are shown in the accompanying drawings. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present disclosure, but should not be understood as limiting the present disclosure.

[0032] It should be noted that the terms "first", "second", etc. in the specification of the present disclosure and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present disclosure described herein can be implemented in an order other than those illustrated or described herein. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present disclosure. Instead, they are merely examples of devices and methods consistent with some aspects of the present disclosure.

[0033] In order to illustrate the technical solution of the present invention, specific embodiments are provided below for illustration.

[0034] The embodiment of the present invention provides a remote moxibustion monitoring and health data management system based on a cloud platform, comprising a processor and a memory, wherein the processor executes a computer program stored in the memory to implement a remote moxibustion monitoring and health data management method based on a cloud platform, such as Figure 1 As shown, the remote moxibustion monitoring and health data management method based on the cloud platform includes the following steps:

[0035] Step S101, respectively obtain the multidimensional monitoring data of each moxibustion patient with the same disease during a treatment process, obtain the median corresponding to each dimension in all the multidimensional monitoring data, and obtain the monitoring data distribution feature sequence of each moxibustion patient based on the difference between the multidimensional monitoring data of each moxibustion patient and the median corresponding to each dimension.

[0036] Moxibustion is a therapy that uses the heat of moxibustion fire to give the human body warm stimulation and prevent and treat diseases through the action of meridians and acupoints. Among them, scar moxibustion can treat asthma, chronic gastrointestinal diseases, rheumatism, etc.; scarless moxibustion is used to treat cold diseases such as asthma, chronic diarrhea, wind-cold dampness, etc.; ginger moxibustion can treat vomiting, diarrhea, diarrhea, rheumatism, etc.; garlic moxibustion can treat surgical sores, yin carbuncle, etc.; salt moxibustion is used to treat urinary problems, acute abdominal pain, and stroke. Since the symptoms of different moxibustion patients may be different, and the hospital resources and doctors' service hours are limited, in the existing technology, relevant medical equipment and instruments are used to collect relevant monitoring data of each moxibustion patient during the moxibustion process, and the relevant monitoring data of all moxibustion patients during the moxibustion process are transmitted to the cloud platform through the network. The cloud platform analyzes the received data and classifies moxibustion patients with similar physical condition attributes into one category to discover potential similarities in the treatment process of moxibustion patients, so as to facilitate subsequent doctors to uniformly manage moxibustion patients of the same category, thereby optimizing resource allocation, improving the medical experience of moxibustion patients, and so on.

[0037] However, in the process of classifying and managing moxibustion patients using the density peak clustering algorithm, the farther the data points are from the cluster center, the more likely they are to be incorrectly divided. For example, the division between the edge points of clusters and clusters is most likely to be incorrectly divided, resulting in shape differences between the clusters corresponding to each cluster center, ultimately affecting the accuracy of the classification and management of moxibustion patients. Therefore, in an embodiment of the present invention, taking one disease condition as an example, the classification of all moxibustion patients under the disease condition is optimized. Please refer to the following for details.

[0038] First, the temperature data, body temperature data, heart rate value and skin resistance data of the moxibustion part of the moxibustion patient during the treatment are collected once per second. There is no restriction here and it can be set according to the needs, corresponding to a temperature data sequence, body temperature data sequence, heart rate value sequence and skin resistance data sequence in the treatment process. Then, the temperature mean of the temperature data sequence, the body temperature mean of the body temperature data sequence, the heart rate mean of the heart rate value sequence and the skin resistance mean of the skin resistance data sequence are calculated, and then the temperature mean, body temperature mean, heart rate mean and skin resistance mean are combined to form the corresponding multi-dimensional monitoring data of the moxibustion patient during a treatment process.

[0039] The multi-dimensional monitoring data of each moxibustion patient can reflect the attribute information of its own state. Therefore, when classifying and managing moxibustion patients, moxibustion patients with similar physical state attributes should be classified into one category as much as possible, so as to facilitate the unified management of moxibustion patients of the same category in the future. However, since the monitoring data of moxibustion patients in different dimensions have different meanings, in order to accurately measure the differences in physical attributes between moxibustion patients, it is necessary to analyze the distribution characteristics of the monitoring data of each dimension of moxibustion patients compared with the overall distribution. Specifically, taking the temperature dimension as an example, in the multi-dimensional monitoring data of all moxibustion patients, all temperature means are obtained and arranged in order (either ascending or descending order is possible), and the median is obtained from the sorted temperature means, which belongs to the middle level of all moxibustion patients in the temperature dimension. Similarly, the corresponding median of each dimension is obtained.

[0040] When the monitoring data of all dimensions of any two moxibustion patients are relatively close to the median corresponding to each dimension, it means that the physical state attributes of the two moxibustion patients are relatively close. Therefore, in the embodiment of the present invention, according to the difference between the multidimensional monitoring data of each moxibustion patient and the median corresponding to each dimension, the monitoring data distribution feature sequence of each moxibustion patient is obtained to characterize the distribution characteristics of each moxibustion patient compared to the intermediate level. Among them, obtaining the monitoring data distribution feature sequence of each moxibustion patient includes:

[0041] For any moxibustion patient, the ratio between the monitoring data of each dimension of the moxibustion patient and the median of the corresponding dimension is obtained to form a distribution feature sequence of the monitoring data of the moxibustion patient.

[0042] In one embodiment, the ratio corresponding to the i-th dimension of any moxibustion patient is used to represent the distribution characteristics of the monitoring data, and the calculation expression of the corresponding ratio is:

[0043]

[0044] in, represents the ratio corresponding to the i-th dimension of the a-th moxibustion patient, represents the monitoring data corresponding to the i-th dimension of the a-th moxibustion patient, Represents the median corresponding to the i-th dimension.

[0045] At this point, the monitoring data distribution feature sequence of each moxibustion patient is obtained.

[0046] Step S102, in the process of performing density peak clustering on all monitoring data distribution feature sequences, obtain the metric distance between every two data points, obtain at least two cluster centers based on all metric distances and a preset cutoff distance, and obtain the number of pre-assigned data points for each cluster center.

[0047] The general process of known density peak clustering is: calculate the Euclidean distance between all sample points in the data set and select the cutoff distance; calculate the local density and relative distance of each sample point in the data set based on the cutoff distance and combined with the local density and relative distance formula; draw a decision diagram based on the local density and relative distance; select the cluster center based on the decision diagram; distribute the remaining sample points, and distribute each remaining sample point to the cluster cluster where its nearest neighbor and the cluster center with a larger local density are located. Therefore, in the embodiment of the present invention, the monitoring data distribution feature sequence of each moxibustion patient is used as the data set for density peak clustering, and the classification management of all moxibustion patients is achieved by performing density peak clustering on all monitoring data distribution feature sequences.

[0048] The specific process of performing density peak clustering on all monitoring data distribution feature sequences is as follows:

[0049] First, the metric distance between every two data points is obtained, wherein the metric distance is obtained by: respectively calculating the standard deviation of all elements in the monitoring data distribution feature sequence corresponding to each data point; for any two data points, calculating the absolute value of the difference between the standard deviations corresponding to the two data points, and calculating the addition result between the absolute value of the difference and a preset value; respectively calculating the absolute value of the difference between the elements corresponding to the two data points in each dimension, and obtaining the average absolute value of the difference; and obtaining the metric distance between the two data points according to the product between the average absolute value of the difference and the addition result.

[0050] In one embodiment, the calculation formula for the metric distance is:

[0051]

[0052] in, represents the metric distance between the two data points corresponding to the a-th moxibustion patient and the b-th moxibustion patient, represents the standard deviation of the distribution characteristic sequence of the monitoring data corresponding to the a-th moxibustion patient, represents the standard deviation of the distribution characteristic sequence of the monitoring data corresponding to the bth moxibustion patient, c represents the preset value to prevent the calculation result of the metric distance from being 0, and c=0.01 is preferably set, and m represents the number of dimensions of the multidimensional monitoring data. represents the sth element in the distribution feature sequence of the monitoring data corresponding to the ath moxibustion patient, represents the sth element in the distribution feature sequence of the monitoring data corresponding to the bth moxibustion patient, and || represents the absolute value symbol.

[0053] It should be noted that It is used to characterize the difference in the distribution sequence of monitoring data between the a-th moxibustion patient and the b-th moxibustion patient. The smaller the value of is, the smaller the difference in monitoring data in each dimension between the a-th moxibustion patient and the b-th moxibustion patient is, the closer the physical state attributes between the two are, and the smaller the metric distance between the two data points corresponding to the a-th moxibustion patient and the b-th moxibustion patient is. The smaller the value is, the smaller the difference in monitoring data in each dimension between the a-th moxibustion patient and the b-th moxibustion patient is, the closer the physical state attributes between the two are, and the smaller the metric distance between the two data points corresponding to the a-th moxibustion patient and the b-th moxibustion patient is.

[0054] Then, all the metric distances are sorted in descending order, and the top 2% of the distances are selected as the cutoff distances. Then, based on all the metric distances and the cutoff distances, the local density and relative distance of each data point are obtained, and the cluster center is determined. It is worth noting that obtaining the local density, relative distance, and cluster center are all existing technologies in the density peak clustering algorithm, and will not be described in detail here.

[0055] After the cluster center is determined, the allocation process of the remaining points in the traditional density peak clustering algorithm is optimized to improve the classification accuracy of the density peak clustering algorithm. The specific optimization method is: according to the selected cluster center and the density of the surrounding data points, a certain number of data points are first allocated to each cluster center, and then, according to the changes in the cluster clusters when the remaining data points are allocated to each cluster cluster, and combined with the metric distance between the remaining data points and the cluster center, the cluster center that is more suitable for the remaining data points is found to complete the final allocation of all data points. That is, the optimization process is divided into two parts. The first part is to obtain the number of first allocated data points for each cluster center respectively, and the second part is to optimize the metric distance between the unallocated data points and each cluster center to divide the unallocated data points.

[0056] Among them, for the first part, the method of obtaining the number of first assigned data points for each cluster center is:

[0057] For any cluster center, other data points within the cutoff distance are taken as neighboring data points of the cluster center, and the average metric distance is obtained according to the metric distance between each of the neighboring data points and the cluster center. The average local density is obtained according to the local density of each of the neighboring data points, and the product of the reciprocal of the average metric distance and the average local density is weighted normalized to obtain the local data distribution compactness of the cluster center;

[0058] Obtain the ratio between the total number of cluster centers and the total number of all data points, obtain the multiplication result between the ratio and the preset hyperparameter, round the product of the multiplication result and the local data distribution density to obtain the number of pre-allocated data points of the cluster center.

[0059] In one implementation, taking the jth cluster center as an example, it is known that the local density of the jth cluster center is obtained based on the number of other data points within the cutoff distance. Therefore, the other data points within the cutoff distance are used as neighboring data points of the jth cluster center. According to the neighboring data points and the cluster center, the number of previously allocated data points of the jth cluster center is obtained:

[0060]

[0061] in, represents the number of data points first assigned to the jth cluster center, wnorm() represents the weight normalization function, represents the mean of the local density of all neighboring data points of the jth cluster center, represents the number of neighboring data points of the jth cluster center, represents the metric distance between the jth cluster center and the rth nearest neighbor data point of the jth cluster center, M represents the total number of data points, that is, the number of moxibustion patients, represents the preset hyperparameter, the value is 10, w represents the number of cluster centers, Indicates the round-up symbol.

[0062] It should be noted that The larger the value of , the more compact the clusters formed by the j-th cluster center in the future may be, and more data points can be allocated to the j-th cluster center, that is, the larger the number of data points allocated to the j-th cluster center in advance; The smaller the value is, the closer the neighboring data points are to the jth cluster center, the tighter the clusters formed by the jth cluster center in the future may be, the more data points will be assigned to it, and the larger the number of data points assigned to the jth cluster center in advance.

[0063] At this point, the number of pre-assigned data points for each cluster center can be obtained, which is used to divide each cluster center into initial cluster clusters in advance.

[0064] Step S103, based on the number of pre-assigned data points of each cluster center, initial clusters and unassigned data points are obtained, and for any unassigned data point, the metric distance between the unassigned data point and each cluster center is optimized to obtain the final metric distance between the unassigned data point and each cluster center.

[0065] After determining the number of data points to be allocated to each cluster center, a corresponding number of data points are allocated to each cluster center to form an initial cluster cluster. One cluster center corresponds to one initial cluster cluster. Specifically: Figure 2 As shown, starting from cluster center 1, follow the counterclockwise direction and start from the horizontal right direction (that is, Figure 2 ), and then select the data points closest to the cluster center in turn (e.g. Figure 2 2) until the number of selected data points is the number of previously assigned data points of the cluster center, and the initial cluster cluster where the cluster center is located is obtained.

[0066] After obtaining multiple initial clusters according to the number of data points assigned to each cluster center, the second part mentioned above can be implemented, that is, optimizing the metric distance between the unassigned data points and each cluster center to divide the unassigned data points, and adaptively adding each unassigned data point to the initial cluster, thereby completing the purpose of classifying all moxibustion patients by the density peak clustering algorithm.

[0067] Take an unassigned data point as an example. When this unassigned data point is added to any initial cluster, if the shape of this initial cluster changes significantly, it means that the assigned data point may have destroyed the original shape of this initial cluster and is not suitable to be added to this initial cluster. At the same time, if the local distribution relationship of the unassigned data point is significantly different from the local distribution relationship of its adjacent data points after the unassigned data point is added to this initial cluster, it indicates that the unassigned data point does not conform to the density change law of the initial cluster and is not suitable to be added to this initial cluster. Therefore, according to the changes of the unassigned data point before and after adding any initial cluster, the confidence of the unassigned data point added to the initial cluster is obtained, and the confidence is used to optimize the metric distance between the unassigned data point and each cluster center. The specific process is as follows:

[0068] For any cluster center, the outermost data point of the initial cluster cluster where the cluster center is located is obtained, and the metric distance between each of the outermost data points and the cluster center is used to form a reference distance sequence; after adding the unassigned data point to the initial cluster cluster where the cluster center is located, the outermost data point of the initial cluster cluster is obtained, and the metric distance between each of the outermost data points and the cluster center is used to form a target distance sequence, and the DTW distance between the reference distance sequence and the target distance sequence is obtained;

[0069] In the initial clustering cluster where the cluster center is located, a data point closest to the unassigned data point is obtained as a first reference point, an absolute value of a local density difference and a metric distance between the unassigned data point and the first reference point are obtained, and a first ratio of an absolute value of a local density difference and a metric distance between the unassigned data point and the first reference point is obtained;

[0070] Acquire the data point closest to the first reference point as the second reference point, obtain the absolute value of the local density difference and the metric distance between the first reference point and the second reference point, and obtain a second ratio of the absolute value of the local density difference and the metric distance between the first reference point and the second reference point;

[0071] Obtaining an absolute value of a difference between the first ratio and the second ratio, normalizing the product of the absolute value of the difference and the DTW distance to obtain a normalized value, and obtaining a confidence level of adding the unassigned data point to the initial clustering cluster where the cluster center is located according to a difference between a constant 1 and the normalized value;

[0072] The confidence level is used to optimize the metric distance between the unassigned data point and the cluster center, and a final metric distance is obtained accordingly.

[0073] In one implementation, taking the jth cluster center as an example, the unassigned data point is denoted as y. First, before the unassigned data point is added to the initial cluster cluster where the jth cluster center is located, in the initial cluster cluster where the jth cluster center is located, the data point farthest from the jth cluster center in each direction (that is, a circle of the jth cluster center) is obtained as the outermost data point, and the metric distance between the jth cluster center and each outermost data point is used to form a reference distance sequence. Then, after the unassigned data points are added to the initial clustering cluster where the jth cluster center is located, in the initial clustering cluster where the jth cluster center is located, the data point farthest from the jth cluster center in each direction is obtained as the outermost data point, and the metric distance between the jth cluster center and each outermost data point is used to form the target distance sequence , the DTW algorithm is used to obtain the DTW distance between the reference distance sequence and the target distance sequence, wherein the DTW algorithm belongs to the prior art and will not be described in detail here.

[0074] The data point closest to the unassigned data point in the initial cluster where the jth cluster center is located is recorded as the first reference point. According to the local density corresponding to each first reference point, the average local density is calculated as the local density of the first reference point. , based on the metric distance between each first reference point and the unassigned data point, calculate the average metric distance as the metric distance of the first reference point Similarly, in the initial cluster where the jth cluster center is located, the data point closest to the first reference point is obtained and recorded as the second reference point. According to the local density corresponding to each second reference point, the average local density is calculated as the local density of the second reference point. , based on the metric distance between each first reference point and each second reference point, calculate the average metric distance as the metric distance of the second reference point .

[0075] Based on the above, the confidence calculation expression of the unassigned data point added to the initial clustering cluster where the jth cluster center is located is:

[0076]

[0077] in, It represents the confidence of adding the unassigned data point y to the initial cluster where the jth cluster center is located. 1 represents a constant, and norm() represents a normalization function. represents the DTW distance between the reference distance sequence and the target distance sequence, || represents the absolute value symbol, represents the local density of unassigned data points y, represents the local density of the first reference point, represents the metric distance of the first reference point, represents the local density of the second reference point, Indicates the metric distance of the second reference point.

[0078] It should be noted that It is used to characterize the difference between the unassigned data points before and after they are added to the initial cluster where the j-th cluster center is located. The smaller the value of , the smaller the influence of the unassigned data point added to the initial cluster where the j-th cluster center is located on the cluster shape, and the greater the confidence of the unassigned data point y added to the initial cluster where the j-th cluster center is located; It is used to characterize the local density variation between the unassigned data points and the first reference point. Used for the local density variation between the first reference point and the second reference point, and use it as a reference. The larger the value of , the greater the change law of the local density after the unassigned data point is added to the initial cluster where the j-th cluster center is located, the greater the influence of the unassigned data point on the initial cluster where the j-th cluster center is located, and the smaller the confidence of the corresponding unassigned data point y added to the initial cluster where the j-th cluster center is located.

[0079] Similarly, the confidence of each unassigned data point added to the initial clustering cluster where each clustering center is located can be obtained, and then the metric distance between each unassigned data point and each clustering center can be optimized according to the confidence, so that all unassigned data points can be assigned to the initial clustering cluster according to the optimized metric distance, wherein the method for optimizing the metric distance between each unassigned data point and each clustering center according to the confidence is: obtaining the difference between a constant 1 and the confidence, and taking the product of the metric distance between the unassigned data point and the clustering center and the difference as the final metric distance.

[0080] In one embodiment, the calculation expression of the final metric distance is:

[0081]

[0082] in, represents the final metric distance between the unassigned data point y and the jth cluster center, which is also the optimized metric distance. represents the metric distance between the unassigned data point y and the jth cluster center, which is also the metric distance before optimization. It represents the confidence of adding the unassigned data point y to the initial cluster where the j-th cluster center is located, and 1 represents a constant.

[0083] It should be noted that the greater the confidence, the more the corresponding unassigned data point should be assigned to the initial cluster where the j-th cluster center is located, and the smaller the metric distance between the corresponding unassigned data point y and the j-th cluster center should be.

[0084] Step S104, allocating all unassigned data points to initial clusters according to the final metric distance between each unassigned data point and each cluster center to obtain final clusters, and classifying and managing moxibustion patients according to the final clusters.

[0085] After determining the final metric distance between each unassigned data point and each cluster center, all unassigned data points are divided into each initial cluster according to the allocation method of allocating the remaining data points to the cluster center in the density peak clustering. At this time, the clustering of all data points is completed, and multiple final clusters are obtained. A final cluster represents a class of moxibustion patients with similar physical state attributes, which improves the accuracy of moxibustion patient classification management. At this point, all moxibustion patients with the same condition are divided into different categories, which helps to discover the potential similarities in the treatment process of moxibustion patients, allowing doctors to develop personalized treatment plans based on the specific conditions of patients and improve treatment effects. At the same time, this analysis can also optimize resource allocation, improve the overall quality of care, and promote the rehabilitation and satisfaction of moxibustion patients.

[0086] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention, and should be included in the protection scope of the present invention.

Claims

1. A remote moxibustion monitoring and health data management system based on a cloud platform, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that: When the processor executes the computer program, the following steps are implemented: The multidimensional monitoring data of each moxibustion patient with the same condition in a treatment process are obtained respectively, and the median corresponding to each dimension is obtained in all the multidimensional monitoring data, and the monitoring data distribution feature sequence of each moxibustion patient is obtained according to the difference between the multidimensional monitoring data of each moxibustion patient and the median corresponding to each dimension; In the process of performing density peak clustering on all monitoring data distribution feature sequences, the metric distance between every two data points is obtained, and at least two cluster centers are obtained according to all metric distances and a preset cutoff distance, and the number of previously allocated data points of each cluster center is obtained respectively; Based on the number of previously assigned data points of each cluster center, initial clusters and unassigned data points are obtained, and for any unassigned data point, the metric distance between the unassigned data point and each cluster center is optimized to obtain a final metric distance between the unassigned data point and each cluster center; According to the final metric distance between each of the unassigned data points and each of the cluster centers, all the unassigned data points are assigned to the initial clusters to obtain final clusters, and moxibustion patients are classified and managed according to the final clusters; The respectively optimizing the metric distance between the unassigned data point and each of the cluster centers to obtain the final metric distance between the unassigned data point and each of the cluster centers includes: For any cluster center, the outermost data point of the initial cluster cluster where the cluster center is located is obtained, and the metric distance between each of the outermost data points and the cluster center is used to form a reference distance sequence; after adding the unassigned data point to the initial cluster cluster where the cluster center is located, the outermost data point of the initial cluster cluster is obtained, and the metric distance between each of the outermost data points and the cluster center is used to form a target distance sequence, and the DTW distance between the reference distance sequence and the target distance sequence is obtained; In the initial clustering cluster where the cluster center is located, a data point closest to the unassigned data point is obtained as a first reference point, an absolute value of a local density difference and a metric distance between the unassigned data point and the first reference point are obtained, and a first ratio of an absolute value of a local density difference and a metric distance between the unassigned data point and the first reference point is obtained; Acquire the data point closest to the first reference point as the second reference point, obtain the absolute value of the local density difference and the metric distance between the first reference point and the second reference point, and obtain a second ratio of the absolute value of the local density difference and the metric distance between the first reference point and the second reference point; Obtaining an absolute value of a difference between the first ratio and the second ratio, normalizing the product of the absolute value of the difference and the DTW distance to obtain a normalized value, and obtaining a confidence level of adding the unassigned data point to the initial clustering cluster where the cluster center is located according to a difference between a constant 1 and the normalized value; The confidence level is used to optimize the metric distance between the unassigned data point and the cluster center, and a final metric distance is obtained accordingly.

2. The remote moxibustion monitoring and health data management system based on a cloud platform according to claim 1, characterized in that: The step of obtaining a distribution characteristic sequence of monitoring data of each moxibustion patient comprises: For any moxibustion patient, the ratio between the monitoring data of each dimension of the moxibustion patient and the median of the corresponding dimension is obtained to form a distribution feature sequence of the monitoring data of the moxibustion patient.

3. The cloud platform-based remote moxibustion monitoring and health data management system according to claim 1, characterized in that: The obtaining of the metric distance between every two data points comprises: Calculate the standard deviation of all elements in the monitoring data distribution feature sequence corresponding to each data point respectively; for any two data points, calculate the absolute value of the difference between the standard deviations corresponding to the two data points, and calculate the addition result between the absolute value of the difference and a preset value; calculate the absolute value of the difference between the elements corresponding to the two data points in each dimension respectively, and obtain the average absolute value of the difference; obtain the metric distance between the two data points according to the product between the average absolute value of the difference and the addition result.

4. The cloud platform-based remote moxibustion monitoring and health data management system according to claim 1, characterized in that: The step of respectively obtaining the number of pre-assigned data points of each cluster center comprises: For any cluster center, other data points within the cutoff distance are taken as neighboring data points of the cluster center, and the average metric distance is obtained according to the metric distance between each of the neighboring data points and the cluster center. The average local density is obtained according to the local density of each of the neighboring data points, and the product of the reciprocal of the average metric distance and the average local density is weighted normalized to obtain the local data distribution compactness of the cluster center; Obtain the ratio between the total number of cluster centers and the total number of all data points, obtain the multiplication result between the ratio and the preset hyperparameter, round the product of the multiplication result and the local data distribution density to obtain the number of pre-allocated data points of the cluster center.

5. The cloud platform-based remote moxibustion monitoring and health data management system according to claim 1, characterized in that: The optimizing the metric distance between the unassigned data point and the cluster center by using the confidence level to obtain a final metric distance includes: The difference between the constant 1 and the confidence level is obtained, and the product of the metric distance between the unassigned data point and the cluster center and the difference is used as the final metric distance.

Images