An intelligent management method for hospital pharmacy prescription information

Abstract

The present invention relates to the technical field of medical information processing, and particularly relates to an intelligent management method for hospital pharmacy prescription information; obtaining the medication frequency characteristic value and the medication characteristic representation value of a patient according to the prescription information of the patient's visit, and obtaining the prescription characteristic points of the patient; obtaining the average difference distance according to the distance characteristics of the prescription characteristic points, and obtaining the adjustment coefficient according to the relevant characteristics of the medication characteristic representation value; obtaining the basic neighborhood radius according to the average difference distance and the adjustment coefficient and performing clustering; obtaining the outlier degree and the optimized cluster according to the medication characteristic representation value of the clustering cluster and the distance characteristics of the clustering cluster; obtaining the compression level according to the quantity characteristics of the prescription characteristic points within the optimized cluster and compressing the prescription information. The present invention performs clustering according to the characteristics of the prescription information and optimizes the clustering result, and performs different degrees of compression according to the clustering result, improving the accuracy of the compression degree of the prescription information.

Description

Technical Field

[0001] The present invention relates to the technical field of medical information processing, and particularly relates to an intelligent management method for prescription information in a hospital pharmacy. Background Art

[0002] With the development of medical digitization, the information systems of hospitals have been continuously improved; for the convenience of doctors' diagnosis and treatment, information such as prescribing prescriptions for patients during medical treatment is stored in the hospital database. Due to the excessive amount of prescription information data of patients, in order to facilitate the efficient storage and management of prescription information, it is necessary to compress and store the prescription information.

[0003] When compressing prescription information with the same compression degree, the differences in the sizes and contents of prescription information of different patients are not considered, resulting in poor management of the compressed data; therefore, it is necessary to distinguish different prescription information; usually, a clustering algorithm is used to cluster different types of prescription information to obtain different clustering clusters, but different settings of clustering parameters during the clustering process are likely to cause errors in the clustering results of prescription information, and then lead to errors in the compression degree of different prescription information by the compression algorithm, affecting the accuracy of the compression degree. Summary of the Invention

[0004] In order to solve the technical problem that the clustering algorithm for prescription information is inaccurate and affects the accuracy of the compression degree, the purpose of the present invention is to provide an intelligent management method for prescription information in a hospital pharmacy, and the specific technical solution adopted is as follows:

[0005] Obtain the prescription information of patients during medical treatment, and obtain the medication frequency characteristic value and medication characteristic representation value of the patients according to the types of medications, the number of medication times, the dosage of medications, and the number of follow-up visits in the prescription information;

[0006] Construct a three-dimensional space according to the medication characteristic representation value, the number of follow-up visits, and the medication frequency characteristic value to obtain the prescription characteristic points of different patients, and obtain the average difference distance according to the distance characteristics between the prescription characteristic points; obtain the adjustment coefficient according to the correlation characteristics between the medication characteristic representation values; obtain the basic neighborhood radius for clustering according to the average difference distance and the adjustment coefficient;

[0007] Cluster the prescription characteristic points according to the basic neighborhood radius to obtain clustering clusters; obtain the inter-cluster difference value according to the difference characteristics of the medication characteristic representation values between the clustering clusters; obtain the outlier degree between the clustering clusters according to the distance characteristics between the clustering clusters and the inter-cluster difference value;

[0008] Merge the clustering clusters according to the outlier degree to obtain optimized clusters; obtain the compression level according to the quantity characteristics of the prescription characteristic points within the optimized clusters and compress the prescription information.

[0009] Further, the step of obtaining the medication frequency characteristic value and the medication characteristic representation value of the patient according to the types of medications, the number of medication administrations, the dosage of medications, and the number of follow-up visits in the prescription information includes:

[0010] Calculate the ratio of the number of medication administrations to the number of follow-up visits to obtain the average number of medications; calculate the average value of the average number of medications for all types of medications of the same patient to obtain the medication frequency representation value of the patient's prescription information;

[0011] Calculate the product of the average number of medications and the dosage of medications to obtain the medication characteristic value, calculate the sum value of the medication characteristic values of all types of medications in the prescription information and perform a positive correlation mapping to obtain the medication characteristic representation value of the patient's prescription information.

[0012] Further, the step of constructing a three-dimensional space according to the medication characteristic representation value, the number of follow-up visits, and the medication frequency characteristic value to obtain the prescription characteristic points of different patients, and obtaining the average difference distance according to the distance characteristics between the prescription characteristic points includes:

[0013] Take the medication characteristic representation value as the X-axis, the number of follow-up visits as the Y-axis, and the medication frequency characteristic value as the Z-axis, and determine the prescription characteristic points according to the positions of the medication characteristic representation value, the number of follow-up visits, and the medication frequency characteristic value of each patient in the three-dimensional rectangular coordinate system; calculate the Euclidean distance between any two of the prescription characteristic points in the three-dimensional space to obtain the difference distance; calculate the average value of the difference distances to obtain the average difference distance of the prescription characteristic points of the patient in the three-dimensional space.

[0014] Further, the step of obtaining the adjustment coefficient according to the correlation characteristics between the medication characteristic representation values includes:

[0015] Calculate the absolute value of the difference between any two of the medication characteristic representation values to obtain the medication characteristic difference value; calculate the average value of the medication characteristic difference values and normalize it to obtain the adjustment coefficient of the neighborhood radius.

[0016] Further, the step of obtaining the basic neighborhood radius according to the average difference distance and the adjustment coefficient includes:

[0017] Calculate the product of the adjustment coefficient and the average difference distance to obtain the basic neighborhood radius of the clustering algorithm.

[0018] Further, the step of clustering the medication characteristic representation values according to the basic neighborhood radius to obtain clustering clusters includes:

[0019] Perform clustering on the basis of the basic neighborhood radius and the prescription characteristic points in the three-dimensional space through the DBSCAN clustering algorithm to obtain different clustering clusters.

[0020] Further, the step of obtaining the inter-cluster difference value according to the difference characteristics of the medication feature representation values between the clustering clusters includes:

[0021] Calculate the average value of the medication feature representation values corresponding to the prescription feature points within the clustering cluster to obtain the intra-cluster medication feature representation mean value; calculate the absolute value of the difference between the intra-cluster medication feature representation mean values of any two clustering clusters to obtain the inter-cluster difference value between the clustering clusters.

[0022] Further, the step of obtaining the dissimilarity between the clustering clusters according to the distance characteristics between the clustering clusters and the inter-cluster difference value includes:

[0023] Calculate the Euclidean distance between the centroids of any two of the clustering clusters to obtain the inter-cluster distance; calculate the product of the inter-cluster distance and the corresponding inter-cluster difference value and normalize it to obtain the dissimilarity between the clustering clusters.

[0024] Further, the step of merging the clustering clusters according to the dissimilarity to obtain the optimized clusters includes:

[0025] Merge the clustering clusters with dissimilarity less than the preset dissimilarity threshold until the dissimilarity between the merged clustering clusters is not less than the preset dissimilarity threshold. All the clustering clusters after the merging are called optimized clusters.

[0026] Further, the step of obtaining the compression level according to the quantity characteristics of the prescription feature points within the optimized clusters and compressing the prescription information includes:

[0027] Calculate the quantity ratio of the prescription feature points in any one of the optimized clusters to all the prescription feature points and map it positively to a preset range to obtain the compression level of the optimized cluster; compress the prescription information corresponding to the optimized cluster by the Deflate algorithm according to the compression level.

[0028] The present invention has the following beneficial effects:

[0029] In the embodiments of the present invention, obtaining the medication frequency eigenvalue and the medication feature representation value can reflect the degree of dependence of patients on drugs in different prescription information. Furthermore, three-dimensional spaces can be constructed through the medication feature representation value, the number of follow-up visits, and the medication frequency eigenvalue to obtain the prescription feature points of different patients. Constructing a three-dimensional space regarding the prescription feature points facilitates clustering. The average difference distance can reflect the distance characteristics between prescription feature points, and thus the neighborhood radius in clustering can be determined according to the average difference distance; obtaining the adjustment coefficient can determine the value of the basic neighborhood radius according to the difference characteristics of the medication feature representation value; based on the basic neighborhood radius, the clustering clusters of prescription feature points can be more accurately obtained through a clustering algorithm. Calculating the dissimilarity degree can reflect the difference characteristics between different clustering clusters and whether they can be merged; obtaining the optimized clusters can more accurately distinguish the difference characteristics between different prescription information, making the clustering result more accurate. Finally, different degrees of compression are performed on the prescription information according to the quantity characteristics of the prescription feature points within the optimized clusters, thereby improving the accuracy of the compression degree of different prescription information. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the following will briefly introduce the accompanying drawings required for the description of the embodiments or the prior art. Obviously, the accompanying drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0031] Figure 1 It is a flowchart of an intelligent management method for hospital pharmacy prescription information provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0032] In order to further elaborate on the technical means and effects adopted by the present invention to achieve the intended invention purpose, the following, in combination with the accompanying drawings and preferred embodiments, details the specific implementation manner, structure, characteristics, and effects of an intelligent management method for hospital pharmacy prescription information proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. In addition, the specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0033] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those of ordinary skill in the technical field to which the present invention belongs.

[0034] The following specifically describes the specific solution of an intelligent management method for hospital pharmacy prescription information provided by the present invention with reference to the accompanying drawings.

[0035] Please refer to Figure 1, which shows a flowchart of an intelligent management method for hospital pharmacy prescription information provided by an embodiment of the present invention. The method includes the following steps:

[0036] Step S1, obtain the prescription information of the patient's visit. According to the types of medications, the number of medication times, the dosage of medications, and the number of follow-up visits in the prescription information, obtain the medication frequency characteristic value and the medication characteristic representation value of the patient.

[0037] In an embodiment of the present invention, the implementation scenario is to compress and store the prescription information of hospital patients. First, obtain the prescription information of all patients' visits. The prescription information includes the number of follow-up visits of the patient, the number of medication times for different types of medications, and the dosage of each medication. It should be noted that since different medications may be solid or liquid, the dosage of each medication is unified to milligram units; subsequently, perform a dimensionless calculation on all the data in the prescription information. The obtained prescription information can reflect the medication characteristics of different patients. For example, patients with similar medication times, dosages, types of medications, and number of follow-up visits can be regarded as having the same type of medication characteristics; therefore, when compressing and storing the prescription information of all patients, those with similar medication characteristics can be grouped together and the compression level can be unified. Therefore, it is necessary to classify different prescription information. First, according to the types of medications, the number of medication times, the dosage of medications, and the number of follow-up visits in the prescription information, obtain the medication frequency characteristic value and the medication characteristic representation value of the patient.

[0038] Preferably, in an embodiment of the present invention, obtaining the medication frequency characteristic value and the medication characteristic representation value includes: calculating the ratio of the number of medication times to the number of follow-up visits to obtain the average number of medications. The larger the value of the average number of medications, the more serious the patient's dependence on the drug. Calculate the average value of the average number of medications for all types of medications of the same patient to obtain the medication frequency representation value of the patient's prescription information; this medication frequency representation value can reflect the overall dependence degree of the patient on different drugs. Calculate the product of the average number of medications and the dosage of medications to obtain the medication characteristic value, calculate the sum value of the medication characteristic values of all types of medications in the prescription information and perform a positive correlation mapping to obtain the medication characteristic representation value of the patient's prescription information; the larger this medication characteristic representation value, the more serious the patient's condition may be and the higher the dependence on the drug. It should be noted that this medication characteristic representation value can only reflect the dependence degree of different patients on drugs and is convenient for subsequent classification, and cannot reflect the type of the patient's condition; patients with more similar medication characteristic representation values have closer dependence degrees on drugs and more similar sizes of prescription information, and can be grouped together for compression. The formula for obtaining the medication characteristic representation value includes:

[0039]

[0040] In the formula, represents the medication characteristic representation value of the patient, represents the number of types of medications of this patient, represents the number of follow-up visits of the patient, represents the number of times the th drug is taken, represents the dosage of the th drug, represents the average number of times the th drug is taken, represents the medication characteristic value of the th drug;

[0041] Step S2, construct a three-dimensional space based on the medication characteristic representation value, the number of follow-up visits, and the medication frequency characteristic value to obtain the prescription characteristic points of different patients, and obtain the average difference distance according to the distance characteristics between the prescription characteristic points; obtain the adjustment coefficient according to the correlation characteristics between the medication characteristic representation values; obtain the basic neighborhood radius for clustering according to the average difference distance and the adjustment coefficient.

[0042] After obtaining the medication characteristic values of the patients, the prescription information of different patients can be clustered; currently, common clustering algorithms include density clustering algorithms such as the DBSCAN clustering algorithm and the OPTICS clustering algorithm. These algorithms cluster based on the density between different data points, clustering data points that are close together into one cluster. It should be noted that in the embodiments of the present invention, the DBSCAN clustering algorithm with better clustering effect is used for analysis. In this algorithm, two hyperparameters, namely the neighborhood radius and the minimum number of samples, need to be set. The setting size of the neighborhood radius will affect the final clustering effect; if the neighborhood radius is too large, prescription information with large characteristic differences will be clustered into the same cluster, and if the neighborhood radius is too small, prescription information with similar characteristics will be clustered into different clusters, ultimately affecting both the compression efficiency and accuracy. Therefore, the neighborhood radius needs to be set according to the difference characteristics between different prescription information.

[0043] Furthermore, the DBSCAN clustering algorithm clusters based on the density of data points in space. Therefore, it is necessary to construct a three-dimensional space based on the medication characteristic representation value, the number of follow-up visits, and the medication frequency characteristic value to obtain the prescription characteristic points of different patients, and obtain the average difference distance according to the distance characteristics between the prescription characteristic points.

[0044] Preferably, in an embodiment of the present invention, obtaining the prescription feature points and the average difference distance includes: constructing a three-dimensional rectangular coordinate system according to the medication feature characterization value, the number of follow-up visits, and the medication frequency feature value; wherein the medication feature characterization value is the X-axis, the number of follow-up visits is the Y-axis, and the medication frequency feature value is the Z-axis; determining the prescription feature points according to the positions of the medication feature characterization value, the number of follow-up visits, and the medication frequency feature value of each patient in the three-dimensional rectangular coordinate system; the position where each prescription feature point is located can reflect the patient's dependence on the drug, and the closer the prescription feature points are, the closer the prescription information is. Calculate the Euclidean distance between any two prescription feature points in the three-dimensional space to obtain the difference distance; calculate the average value of the difference distance to obtain the average difference distance of the patient's prescription feature points in the three-dimensional space. The larger the average difference distance, the greater the characteristic difference of the prescription information between different patients, and the longer the neighborhood radius needs to be increased during the clustering process; when the average difference distance is smaller, it means that the characteristic difference of the prescription information between different patients is smaller, and a smaller neighborhood radius is required during the clustering process to ensure the clustering effect.

[0045] After obtaining the average difference distance, it is necessary to analyze the neighborhood radius in the DBSCAN clustering algorithm. The medication feature characterization values of the patients are calculated in step S1. When the difference between the medication feature characterization values is larger, it means that the characteristic difference of the prescription information is larger, and a larger neighborhood radius is required; therefore, the adjustment coefficient can be obtained according to the relevant characteristics between the medication feature characterization values.

[0046] Preferably, in an embodiment of the present invention, obtaining the adjustment coefficient includes: calculating the absolute value of the difference between any two medication feature characterization values to obtain the medication feature difference value; calculating the average value of the medication feature difference value and normalizing it to obtain the adjustment coefficient of the neighborhood radius; when the medication feature difference value is larger, the adjustment coefficient is larger, which means that the characteristic difference of the prescription information is larger, and a larger neighborhood radius is required to ensure the clustering effect; when the medication feature difference value is smaller, the adjustment coefficient is smaller, which means that the characteristic difference of the prescription information is smaller, and a smaller neighborhood radius is required to ensure the clustering effect.

[0047] Furthermore, after obtaining the average difference distance of the prescription feature points in the three-dimensional space and the adjustment coefficient of the neighborhood radius, the basic neighborhood radius can be obtained according to the average difference distance and the adjustment coefficient; preferably, calculate the product of the adjustment coefficient and the average difference distance to obtain the basic neighborhood radius of the DBSCAN clustering algorithm. When the adjustment coefficient and the average difference distance are larger, it means that the overall distribution of the prescription feature points in the three-dimensional space is relatively discrete and the characteristic difference of the prescription information is larger, and a larger neighborhood radius is required for clustering; when the adjustment coefficient and the average difference distance are smaller, it means that the overall distribution of the prescription feature points in the three-dimensional space is relatively dense and the characteristic difference of the prescription information is smaller, and a smaller neighborhood radius is required for clustering. The formula for obtaining the basic neighborhood radius includes:

[0048]

[0049] In the formula, represents the basic neighborhood radius, represents the number of medication characteristic difference values, represents the th medication characteristic difference value, represents the normalization function, represents the adjustment coefficient of the neighborhood radius, represents the average difference distance.

[0050] Step S3: Cluster the prescription feature points according to the basic neighborhood radius to obtain clustering clusters; obtain the between-cluster difference values according to the difference characteristics of the medication characteristic representation values between the clustering clusters; obtain the dissimilarity between the clustering clusters according to the distance characteristics between the clustering clusters and the between-cluster difference values.

[0051] After obtaining the basic neighborhood radius in the DBSCAN clustering algorithm, the prescription feature points in the three-dimensional space can be clustered by the DBSCAN clustering algorithm according to the basic neighborhood radius to obtain different clustering clusters; it should be noted that the DBSCAN clustering algorithm belongs to the prior art, and the specific clustering steps will not be elaborated here; the minimum number of samples is set to 10, and the implementer can determine it according to the implementation scenario.

[0052] After clustering, the prescription information corresponding to the prescription feature points within each clustering cluster is relatively similar, and the degree of dependence of the patients on the drugs is close; however, this clustering algorithm may misidentify non-spherical or other complex-shaped clusters during the clustering process, resulting in poor clustering effects. Therefore, it is necessary to analyze the difference characteristics between the obtained clustering clusters, merge the clustering clusters with small differences, and improve the compression accuracy; first, obtain the between-cluster difference values according to the difference characteristics of the medication characteristic representation values between the clustering clusters.

[0053] Preferably, in the embodiment of the present invention, obtaining the between-cluster difference values includes: calculating the average value of the medication characteristic representation values corresponding to the prescription feature points within the clustering cluster to obtain the within-cluster medication characteristic representation mean value; calculating the absolute value of the difference between the within-cluster medication characteristic representation mean values of any two clustering clusters to obtain the between-cluster difference value between the clustering clusters. When the between-cluster difference value between two clustering clusters is larger, it means that the characteristic differences of the prescription information within the two clusters are larger, and the prescription information is significantly different; when the between-cluster difference value between two clustering clusters is smaller, it means that the characteristic differences of the prescription information within the two clusters are smaller, and the probability of merging is greater.

[0054] Further, the closer the distances between two clusters in three-dimensional space are, the greater the probability of cluster merging. Therefore, the dissimilarity between clusters can be obtained based on the distance characteristics between clusters and the inter-cluster difference value. Preferably, obtaining the dissimilarity includes: calculating the Euclidean distance between the centroids of any two clusters to obtain the cluster distance. The larger the cluster distance is, the greater the distance between the two clusters is; conversely, the smaller the cluster distance is, the smaller the distance between the two clusters is. Calculate the product of the cluster distance and the corresponding inter-cluster difference value and normalize it to obtain the dissimilarity between clusters. When the inter-cluster difference value and the cluster distance are smaller, it means that the two clusters are closer, and the characteristic difference of the prescription information within the two clusters is smaller, and the dissimilarity is smaller, then the probability that the two clusters can be merged is greater; conversely, when the inter-cluster difference value and the cluster distance of the two clusters are larger, it means that the two clusters are farther apart, and the characteristic difference of the prescription information within the two clusters is larger, and the dissimilarity is larger, and the probability that the two clusters can be merged is smaller.

[0055] Step S4: Merge the clusters according to the dissimilarity to obtain optimized clusters; obtain the compression level based on the quantity characteristics of the prescription feature points within the optimized clusters and compress the prescription information.

[0056] After obtaining the dissimilarity between different clusters, the clusters can be merged according to the dissimilarity to obtain optimized clusters, which specifically includes: merging the clusters with dissimilarity less than the preset dissimilarity threshold until the dissimilarity between the merged clusters is not less than the preset dissimilarity threshold. All the merged clusters are called optimized clusters. In the embodiment of the present invention, the preset dissimilarity threshold is 0.3, and the implementer can determine it according to the implementation scenario; and if the dissimilarity between the merged cluster and other clusters, further merging is required until the dissimilarity between all the merged clusters is not less than the preset dissimilarity threshold. All the final clusters are called optimized clusters; the characteristic differences of the prescription information between each optimized cluster are obvious.

[0057] After obtaining different optimized clusters, it is necessary to compress all prescription information to varying degrees. Therefore, the compression level is obtained based on the quantity characteristics of the prescription feature points within the optimized cluster, and the prescription information is compressed accordingly. Preferably, in an embodiment of the present invention, obtaining the compression level and compressing the prescription information includes: calculating the ratio of the number of prescription feature points in any one optimized cluster to the number of all prescription feature points and positively correlating and mapping it to a preset range. The larger this ratio, the more corresponding prescription information there is in the optimized cluster, indicating that this type of prescription information is common and does not need to be frequently called, so a larger compression level is required and the compression degree is greater. The smaller this ratio, the less corresponding prescription information there is in the optimized cluster, indicating that this type of prescription information is rare, so the compression degree needs to be reduced. In the embodiment of the present invention, the preset range is from 0 to 9, corresponding to 10 compression levels of the Deflate algorithm. The larger the ratio, the larger the compression level and the greater the compression degree. The specific mapping steps are to calculate the ratio of the number of all optimized clusters, positively correlate and map the ratio to ten integers from 0 to 9, and obtain the compression level of the optimized cluster. According to the compression level, the prescription information corresponding to the optimized cluster is compressed by the Deflate algorithm. It should be noted that the Deflate compression algorithm belongs to the prior art, and the specific compression steps will not be elaborated. Thus, the prescription information of all patients is classified and compressed to varying degrees according to the quantity characteristics of the optimized cluster, improving the accuracy of the compression degree of the prescription information.

[0058] In summary, the embodiment of the present invention provides an intelligent management method for hospital pharmacy prescription information; obtaining the medication frequency characteristic value and medication characteristic representation value of a patient based on the prescription information of the patient's visit, and obtaining the prescription feature points of the patient; obtaining the average difference distance based on the distance characteristics of the prescription feature points, and obtaining the adjustment coefficient based on the relevant characteristics of the medication characteristic representation value; obtaining the basic neighborhood radius based on the average difference distance and the adjustment coefficient and performing clustering; obtaining the dissimilarity degree and the optimized cluster based on the medication characteristic representation value of the clustering cluster and the distance characteristics of the clustering cluster; obtaining the compression level based on the quantity characteristics of the prescription feature points within the optimized cluster and compressing the prescription information. The present invention clusters according to the characteristics of the prescription information and optimizes the clustering result, and performs different degrees of compression according to the clustering result, improving the accuracy of the compression degree of the prescription information.

[0059] It should be noted that: the above sequence of the embodiments of the present invention is only for description and does not represent the advantages or disadvantages of the embodiments. The processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0060] Each embodiment in this specification is described in a progressive manner. For the same or similar parts among the embodiments, reference can be made to each other, and the difference between each embodiment and other embodiments is emphasized.

Claims

1. An intelligent management method for hospital pharmacy prescription information, characterized in that, The method includes the following steps: Obtain the prescription information of the patient's visit. Based on the types of medications, the number of medication administrations, the dosage of medications, and the number of follow-up visits in the prescription information, obtain the medication frequency characteristic value and the medication characteristic representation value of the patient; Construct a three-dimensional space based on the medication characteristic representation value, the number of follow-up visits, and the medication frequency characteristic value to obtain the prescription characteristic points of different patients. Obtain the average difference distance based on the distance characteristics between the prescription characteristic points; Obtain the adjustment coefficient based on the correlation characteristics between the medication characteristic representation values; Obtain the basic neighborhood radius for clustering based on the average difference distance and the adjustment coefficient; Cluster the prescription characteristic points according to the basic neighborhood radius to obtain clustering clusters; Obtain the between-cluster difference value based on the difference characteristics of the medication characteristic representation values between the clustering clusters; Obtain the dissimilarity between the clustering clusters based on the distance characteristics between the clustering clusters and the between-cluster difference value; Merge the clustering clusters according to the dissimilarity to obtain optimized clusters; Obtain the compression level based on the quantity characteristics of the prescription characteristic points within the optimized clusters and compress the prescription information; The step of obtaining the medication frequency characteristic value and the medication characteristic representation value of the patient based on the types of medications, the number of medication administrations, the dosage of medications, and the number of follow-up visits in the prescription information includes: Calculate the ratio of the number of medication administrations to the number of follow-up visits to obtain the average number of medication administrations; Calculate the average value of the average number of medication administrations for all types of medications of the same patient to obtain the medication frequency representation value of the patient's prescription information; Calculate the product of the average number of medication administrations and the dosage of medications to obtain the medication characteristic value. Calculate the sum value of the medication characteristic values of all types of medications in the prescription information and perform a positive correlation mapping to obtain the medication characteristic representation value of the patient's prescription information; The step of obtaining the basic neighborhood radius for clustering based on the average difference distance and the adjustment coefficient includes: Calculate the product of the adjustment coefficient and the average difference distance to obtain the basic neighborhood radius of the clustering algorithm; The step of obtaining the dissimilarity between the clustering clusters based on the distance characteristics between the clustering clusters and the between-cluster difference value includes: Calculate the Euclidean distance between the centroids of any two clustering clusters to obtain the cluster distance; Calculate the product of the cluster distance and the corresponding between-cluster difference value and normalize it to obtain the dissimilarity between the clustering clusters; The step of merging the clustering clusters according to the dissimilarity to obtain optimized clusters includes: Merge the clustering clusters with a dissimilarity less than the preset dissimilarity threshold until the dissimilarity between the merged clustering clusters is not less than the preset dissimilarity threshold. All the clustering clusters after the merging is completed are called optimized clusters.

2. The intelligent management method for hospital pharmacy prescription information according to claim 1, characterized in that, The step of constructing a three-dimensional space based on the medication characteristic representation value, the number of follow-up visits, and the medication frequency characteristic value to obtain the prescription characteristic points of different patients and obtaining the average difference distance based on the distance characteristics between the prescription characteristic points includes: Taking the medication characteristic representation value as the X-axis, the number of follow-up visits as the Y-axis, and the medication frequency characteristic value as the Z-axis, determine the prescription characteristic points according to the positions of the medication characteristic representation value, the number of follow-up visits, and the medication frequency characteristic value of each patient in a three-dimensional rectangular coordinate system; calculate the Euclidean distance between any two of the prescription characteristic points in three-dimensional space to obtain the difference distance; calculate the average value of the difference distances to obtain the average difference distance of the prescription characteristic points of the patient in three-dimensional space.

3. The intelligent management method for hospital pharmacy prescription information according to claim 1, wherein The step of obtaining the adjustment coefficient according to the correlation characteristics between the medication characteristic representation values includes: Calculate the absolute value of the difference between any two of the medication characteristic representation values to obtain the medication characteristic difference value; calculate the average value of the medication characteristic difference values and normalize it to obtain the adjustment coefficient of the neighborhood radius.

4. The intelligent management method for hospital pharmacy prescription information according to claim 1, wherein The step of clustering the prescription characteristic points according to the basic neighborhood radius to obtain clustering clusters includes: Cluster according to the basic neighborhood radius and the prescription characteristic points in three-dimensional space by the DBSCAN clustering algorithm to obtain different clustering clusters.

5. The intelligent management method for hospital pharmacy prescription information according to claim 1, characterized in that The step of obtaining the between-cluster difference value according to the difference characteristics of the medication characteristic representation values between the clustering clusters includes: Calculate the average value of the medication characteristic representation values corresponding to the prescription characteristic points within the clustering cluster to obtain the within-cluster medication characteristic representation mean value; calculate the absolute value of the difference between the within-cluster medication characteristic representation mean values of any two clustering clusters to obtain the between-cluster difference value between the clustering clusters.

6. The intelligent management method for hospital pharmacy prescription information according to claim 1, characterized in that, The step of obtaining the compression level according to the quantity characteristics of the prescription characteristic points within the optimized cluster and compressing the prescription information includes: Calculate the ratio of the number of prescription characteristic points in any one optimized cluster to the number of all prescription characteristic points and map it positively to a preset range to obtain the compression level of the optimized cluster; compress the prescription information corresponding to the optimized cluster by the Deflate algorithm according to the compression level.

Images