A urological diagnosis and treatment data intelligent processing method and system
By calculating the important evaluation factors and similarity of patient diagnosis and treatment data, iterative clustering is performed, which solves the problem of inaccurate patient clustering in the existing technology and improves the classification and storage effect and personalized management of urological diagnosis and treatment data.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- BEIJING SHIJIUYI INFORMATION TECHNOLOGY CO LTD
- Filing Date
- 2025-12-04
- Publication Date
- 2026-05-12
AI Technical Summary
Existing clustering algorithms cannot effectively consider the mutual influence between different dimensions of urological diagnosis and treatment data, resulting in inaccurate patient clustering and reduced effectiveness of classification and storage of diagnosis and treatment data.
By acquiring different dimensions of diagnosis and treatment data for each patient, important evaluation factors are calculated, and iterative clustering is performed based on similarity. The best clustering results are then selected and stored in the database.
It improved the classification and storage effectiveness of patient group diagnosis and treatment data, and enhanced the personalized management and storage efficiency of diagnosis and treatment data.
Smart Images
Figure CN121687543B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of diagnostic and treatment data processing, and specifically to an intelligent processing method and system for urological diagnostic and treatment data. Background Technology
[0002] Urological diagnosis and treatment data can be used to assess and analyze patients' urinary system diseases. Because patients' urological diagnosis and treatment data involve multiple types and dimensions, and the data volume is large and complex, the accurate classification and storage of patients' urological diagnosis and treatment data can not only help doctors develop personalized treatment plans for patients, but also improve the efficiency of diagnosis and treatment, which is of great significance to the urological diagnosis and treatment process.
[0003] In related technologies, patients are usually clustered based on the differences in the same dimension of their medical data, and the medical data of patients in the same cluster are stored as the same class. However, since conventional clustering algorithms do not take into account the mutual influence between medical data of different dimensions during the clustering process, the existing clustering methods cannot accurately cluster patients, thus reducing the effectiveness of classifying and storing the medical data of patient groups. Summary of the Invention
[0004] To address the technical problem that existing clustering methods cannot accurately cluster patients, thus reducing the effectiveness of classifying and storing patient diagnostic and treatment data, the present invention aims to provide an intelligent processing method and system for urological diagnostic and treatment data. The specific technical solution adopted is as follows:
[0005] This invention proposes an intelligent processing method for urological diagnostic and treatment data, the method comprising:
[0006] To acquire diagnostic and treatment data from different dimensions for each patient in the urology department;
[0007] Using any patient as the target patient and any dimension as the target dimension, the important evaluation factors of the target dimension of the target patient are obtained based on the differences in the diagnostic and treatment data of the target dimension between the target patient and other patients, as well as the correlation between the diagnostic and treatment data of the target dimension and other dimensions. The similarity between any two patients is obtained based on the differences in the diagnostic and treatment data of the same dimension between any two patients and the differences in the important evaluation factors.
[0008] Based on the similarity, iterative clustering is performed on all patients to obtain multiple clustering results, each containing multiple clusters. Any clustering result is taken as the target clustering result. The optimization degree of the target clustering result is obtained based on the similarity between each patient in each cluster and the patient corresponding to the cluster center, as well as the important evaluation factors of the same dimension for each patient in each cluster.
[0009] Based on the optimization of each clustering result, the best clustering result is selected from all clustering results, and the diagnosis and treatment data of patients in the same cluster of the best clustering result are stored in the database.
[0010] Furthermore, the key evaluation factors for obtaining the target dimensions of the target patient include:
[0011] The average value of the diagnostic and treatment data of all patients in the target dimension is taken as the first overall diagnostic and treatment data of the target dimension;
[0012] The absolute value of the difference between the target patient's target dimension diagnosis and treatment data and the first overall diagnosis and treatment data is used as the first data deviation value of the target patient's target dimension. The first data deviation value of the target patient's target dimension is used as the numerator, and the sum of the first data deviation values of all dimensions of the target patient is used as the denominator. The ratio is used as the first importance of the target patient's target dimension.
[0013] The sequence of diagnostic and treatment data for all patients in the same dimension is taken as the diagnostic and treatment data sequence for each dimension.
[0014] Using dimensions other than the target dimension as reference dimensions, the correlation between the diagnosis and treatment data sequences of the target dimension and the diagnosis and treatment data sequences of each reference dimension is analyzed to obtain the second importance of the target dimension.
[0015] The product of the first importance and the second importance of the target dimension for the target patient is used as an important evaluation factor for the target dimension of the target patient.
[0016] Furthermore, the second importance of obtaining the target dimension includes:
[0017] The absolute value of the Pearson correlation coefficient between the diagnostic and treatment data sequence of the target dimension and the diagnostic and treatment data sequence of each reference dimension is taken as the data correlation between the target dimension and each reference dimension.
[0018] The average of the data relevance between the target dimension and all reference dimensions is used as the second importance of the target dimension.
[0019] Furthermore, obtaining the degree of similarity between any two patients includes:
[0020] The important evaluation factors of the target dimension for each patient are used as the numerator, the cumulative value of the important evaluation factors of all dimensions for each patient is used as the denominator, and the ratio is used as the importance percentage of the target dimension for each patient.
[0021] Two patients are randomly selected as the first patient and the second patient. The absolute value of the difference between the importance ratio of the target dimension between the first patient and the second patient is negatively correlated to obtain the first similarity parameter of the target dimension between the first patient and the second patient.
[0022] A negative correlation mapping is performed on the absolute values of the differences in the diagnostic and treatment data of the first patient and the second patient in the target dimension to obtain a second similarity parameter between the first patient and the second patient in the target dimension;
[0023] The first similarity parameter and the second similarity parameter are combined to obtain the comprehensive similarity parameter of the target dimension between the first patient and the second patient;
[0024] The sum of the comprehensive similarity parameters across all dimensions between the first and second patients is normalized to obtain the degree of similarity between the first and second patients.
[0025] Furthermore, obtaining multiple clustering results includes:
[0026] Using the iterative self-organizing clustering algorithm, based on the similarity between patients, all patients are first clustered to obtain the first clustering result, and the first clustering result is used as the current clustering result;
[0027] Take any cluster in the current clustering result as the cluster to be tested. Based on the diagnosis and treatment data of patients in the cluster to be tested in each dimension, select the cluster center patient of the cluster to be tested from all patients in the cluster to be tested. Take any patient in the cluster to be tested as the patient to be tested. Based on the similarity between the patient to be tested and the cluster center patient of the cluster to be tested, and the difference of the important evaluation factors of the same dimension of each patient in the cluster to be tested, obtain the weight parameters of the patient to be tested in the current clustering result.
[0028] Using the iterative self-organizing clustering algorithm, based on the weight parameters of each patient in the current clustering result and the similarity between patients, the next clustering is performed on all patients to obtain the next clustering result. If the termination condition of iterative clustering is not met, the next clustering result is used as the current clustering result and iterative clustering continues; otherwise, iterative clustering is stopped.
[0029] Furthermore, the step of selecting the cluster center patient of the cluster to be tested from all patients in the cluster to be tested includes:
[0030] The average value of the diagnostic and treatment data of the same dimension of all patients in the cluster to be tested is used as the second overall diagnostic and treatment data of each dimension in the cluster to be tested.
[0031] The absolute value of the difference between the diagnostic and treatment data of the target dimension and the second overall diagnostic and treatment data of the target dimension for each patient in the cluster to be tested is used as the second data deviation value of the target dimension for each patient in the cluster to be tested.
[0032] The average of the second data deviation values of all dimensions for each patient in the cluster to be tested is used as the center deviation value for each patient in the cluster to be tested;
[0033] In the cluster to be tested, the patient corresponding to the minimum value of the center deviation is taken as the cluster center patient of the cluster to be tested.
[0034] Furthermore, the weight parameters for obtaining the patients to be tested in the current clustering results include:
[0035] The average value of the important evaluation factors of the same dimension for all patients in the cluster to be tested is used as the overall important evaluation factor for each dimension of the cluster to be tested.
[0036] The absolute value of the difference between the important evaluation factor of each dimension of the patient under test and the overall important evaluation factor of each dimension of the cluster under test is used as the importance deviation value of each dimension of the patient under test; the average value of the importance deviation values of all dimensions of the patient under test is used as the classification necessity of the patient under test.
[0037] The similarity between the patient to be tested and the patients at the cluster centers of the clusters to be tested, as well as the classification necessity, are comprehensively processed to obtain the weight parameters of the patient to be tested in the current clustering results. The weight parameters of the patient to be tested in the current clustering results are positively correlated with the similarity between the patient to be tested and the patients at the cluster centers of the clusters to be tested, and negatively correlated with the classification necessity.
[0038] Furthermore, the degree of preference for obtaining the target clustering results includes:
[0039] Take any cluster in the target clustering results as the target cluster, and take the product of the similarity between each patient in the target cluster and the cluster center patient of the target cluster and the weight parameter of each patient as the update stability of each patient in the target cluster.
[0040] The average of the update stability of all patients in the target cluster is used as the evaluation score of the clustering effect of the target cluster.
[0041] The average value of the clustering effect evaluation of all clusters in the target clustering result is taken as the degree of preference of the target clustering result.
[0042] Furthermore, the step of selecting the best clustering result from all clustering results includes:
[0043] The clustering result corresponding to the maximum value of the preferred degree is taken as the best clustering result.
[0044] The present invention also proposes an intelligent processing system for urological diagnosis and treatment data. The system includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements any of the steps of an intelligent processing method for urological diagnosis and treatment data.
[0045] The present invention has the following beneficial effects:
[0046] This invention addresses the limitation of existing clustering methods in accurately clustering patients, which reduces the effectiveness of classifying and storing patient data. Therefore, it first acquires the multi-dimensional medical data of each patient. Since conventional clustering algorithms do not consider the impact of multi-dimensional medical data, resulting in poor clustering performance, this invention uses important evaluation factors to reflect the importance of each dimension of medical data for each patient. These important evaluation factors are then incorporated into the similarity analysis between patients to improve the final clustering effect. Iterative clustering is performed on all patients using their similarity, resulting in multiple clustering results. Because different clustering results have varying effects, the optimality of each clustering result is used to reflect its effectiveness. The best clustering result is then selected, and the multi-dimensional medical data of patients within the same cluster from the best clustering result is stored in the database, thereby improving the effectiveness of classifying and storing patient data. Attached Figure Description
[0047] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0048] Figure 1The flowchart illustrates an intelligent processing method for urological diagnostic and treatment data, as provided in one embodiment of the present invention. Detailed Implementation
[0049] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of an intelligent processing method and system for urological diagnostic and treatment data proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.
[0050] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.
[0051] The following description, in conjunction with the accompanying drawings, details the specific scheme of the intelligent processing method and system for urological diagnostic and treatment data provided by the present invention.
[0052] Please see Figure 1 The diagram illustrates a flowchart of an intelligent processing method for urological diagnostic and treatment data according to an embodiment of the present invention. The method includes:
[0053] Step S1: Obtain diagnostic and treatment data for each patient in the urology department from different dimensions.
[0054] During the diagnosis of urinary system diseases, the hospital will test various diagnostic data of patients with urinary system diseases, such as urobilinogen concentration, pH, creatinine, urea nitrogen, and uric acid in the blood, and save these data in the hospital's data management database for subsequent analysis and processing.
[0055] Therefore, in this embodiment of the invention, the diagnostic and treatment data of each patient in the urology department are first obtained from the hospital’s data management database. It should be noted that since the dimensions of the diagnostic and treatment data of different dimensions are different, in order to eliminate the influence of the dimensions, it is also necessary to perform data standardization processing on the diagnostic and treatment data of each patient in different dimensions. Data standardization processing is a technical means well known to those skilled in the art, and will not be described in detail here.
[0056] Step S2: Select any patient as the target patient and any dimension as the target dimension. Based on the differences in the diagnostic and treatment data of the target dimension between the target patient and other patients, as well as the correlation between the diagnostic and treatment data of the target dimension and other dimensions, obtain the important evaluation factors of the target dimension for the target patient. Based on the differences in the diagnostic and treatment data of the same dimension and the differences in important evaluation factors between any two patients, obtain the degree of similarity between any two patients.
[0057] In the subsequent classification of patients' medical data, conventional clustering algorithms fail to consider the interrelationships between different dimensions of medical data, leading to inaccurate clustering results and reduced effectiveness in classifying and storing patient medical data. For example, if a patient's medical data differs from that of most other patients in certain dimensions, they may be incorrectly clustered together because their medical data in other dimensions is similar to other patients, resulting in the loss of important information in the patient's medical data and hindering personalized management and storage of the data. Therefore, this invention first uses any patient as the target patient and any dimension as the target dimension. It then analyzes the differences in the target dimension's medical data between the target patient and other patients, as well as the correlation between the target dimension and other dimensions. The obtained important evaluation factors reflect the importance of each dimension's medical data for each patient. Subsequently, these important evaluation factors can be added along with the medical data to the similarity analysis between patients, thereby improving the final clustering effect.
[0058] Preferably, in one embodiment of the present invention, the method for obtaining important evaluation factors of the target dimension of the target patient specifically includes:
[0059] First, the greater the difference in the target dimension's diagnostic and treatment data between the target patient and other patients, the more important the target dimension's diagnostic and treatment data is to the target patient. Therefore, the average value of the target dimension's diagnostic and treatment data of all patients can be used as the first overall diagnostic and treatment data for the target dimension. This first overall diagnostic and treatment data can reflect the overall level of the target dimension's diagnostic and treatment data for all patients.
[0060] The absolute value of the difference between the target patient's treatment data in the target dimension and the overall treatment data is taken as the first data deviation value of the target patient's target dimension. The larger the first data deviation value, the greater the difference in treatment data in the target dimension between the target patient and other patients. The first data deviation value of each dimension of the target patient can be obtained through the same method. Then, the first data deviation value of the target patient's target dimension is used as the numerator, and the sum of the first data deviation values of all dimensions of the target patient is used as the denominator. The ratio is taken as the first importance of the target patient's target dimension. The larger the first importance value, the greater the difference in treatment data in the target dimension between the target patient and other patients is compared with the difference in treatment data in other dimensions between the target patient and other patients. This indicates that the treatment data in the target dimension is more important to the target patient.
[0061] As an example, in one embodiment of the present invention, the expression for the first importance of the target dimension of the target patient can be specifically as follows:
[0062]
[0063]
[0064] in, Indicates the first importance of the target dimension for the target patient; The first data deviation value representing the target dimension for the target patient; Indicates the first [patient's] [number] The first data deviation value in each dimension; Indicates the number of all dimensions; Diagnostic and treatment data representing the target dimensions of the target patient; Indicates the first Treatment data for each patient across target dimensions; This indicates the total number of patients. This represents the first overall diagnostic and treatment data representing the target dimension; Equivalent to Normalization processing. It should be understood that, in extreme cases... If the sum of the first data deviation values of all dimensions of the target patient is 0, it means that the first data deviation value of each dimension of the target patient is 0, which means that there is no data deviation in each dimension of the target patient. Therefore, the first importance of each dimension of the target patient is 0, the data of the target patient has no analytical value, and the target patient will not be analyzed further.
[0065] Then, the sequence of diagnostic and treatment data of all patients in the same dimension is taken as the diagnostic and treatment data sequence of each dimension. Other dimensions besides the target dimension are taken as reference dimensions. The stronger the correlation between the diagnostic and treatment data sequence of the target dimension and the diagnostic and treatment data sequence of the reference dimension, the more important the diagnostic and treatment data of the target dimension is. Therefore, the correlation between the diagnostic and treatment data sequence of the target dimension and the diagnostic and treatment data sequence of each reference dimension can be analyzed to obtain the second importance of the target dimension.
[0066] Preferably, in one embodiment of the present invention, the method for obtaining the second importance of the target dimension specifically includes:
[0067] The absolute value of the Pearson correlation coefficient between the diagnostic and treatment data sequences of the target dimension and the diagnostic and treatment data sequences of each reference dimension is used as the data correlation between the target dimension and each reference dimension. The higher the data correlation, the stronger the correlation between the target dimension and each reference dimension, and thus the more important the target dimension is. Therefore, the average value of the data correlation between the target dimension and all reference dimensions can be used as the second importance of the target dimension. In other embodiments of the present invention, methods such as Spearman's rank correlation coefficient or consistency correlation coefficient can also be used to calculate the data correlation between the target dimension and each reference dimension, which is not limited here.
[0068] As an example, in one embodiment of the present invention, the expression for the second importance of the target dimension can be specifically as follows:
[0069]
[0070] in, This indicates the second most important dimension of the objective. Represents the sequence of diagnostic and treatment data in the target dimension and the first Pearson correlation coefficients between the diagnostic and treatment data sequences of each reference dimension; Representing the target dimension and the first The data correlation between the reference dimensions; To represent the number of all dimensions, then This indicates the number of all reference dimensions for the target dimension.
[0071] Finally, the product of the first importance and the second importance of the target dimension for the target patient is used as an important evaluation factor for the target dimension of the target patient.
[0072] As an example, in one embodiment of the present invention, the expression for the important evaluation factors of the target dimension of the target patient can be specifically as follows:
[0073]
[0074] in, Key evaluation factors representing the target dimensions of the target patient; Indicates the first importance of the target dimension for the target patient; This indicates the second most important dimension of the objective.
[0075] By using the same method described above, we can obtain the important evaluation factors for each dimension of the target patient, as well as the important evaluation factors for each dimension of each patient.
[0076] Existing clustering algorithms typically cluster patients based solely on differences or similarities in medical data within the same dimension, without considering the interrelationships between medical data across different dimensions. In other words, they fail to account for the varying importance of different dimensions of medical data to patients. Therefore, this invention analyzes the similarity of medical data between patients based on differences in medical data within the same dimension between any two patients, combined with differences in important evaluation factors within the same dimension. This yields the degree of similarity between any two patients, allowing for iterative clustering of patients based on this similarity. This improves the clustering effect and consequently enhances the effectiveness of subsequent classification and storage of patient medical data.
[0077] Preferably, in one embodiment of the present invention, the method for obtaining the similarity between any two patients specifically includes:
[0078] The important evaluation factors of each patient's target dimension are used as the numerator, and the sum of the important evaluation factors of all dimensions for each patient is used as the denominator. The ratio is used as the importance percentage of each patient's target dimension. The higher the importance percentage, the more important the diagnostic and treatment data of a certain patient's target dimension is to that patient compared with other dimensions.
[0079] Two patients are randomly selected as the first patient and the second patient. The absolute value of the difference in the importance ratio of the target dimension between the first patient and the second patient is negatively correlated to obtain the first similarity parameter of the target dimension between the first patient and the second patient. The larger the first similarity parameter, the closer the importance of the target dimension between the first patient and the second patient is.
[0080] By performing a negative correlation mapping on the absolute value of the difference in the target dimension of the diagnosis and treatment data between the first patient and the second patient, a second similarity parameter in the target dimension between the first patient and the second patient is obtained. The larger the second similarity parameter, the closer the diagnosis and treatment data in the target dimension between the first patient and the second patient are.
[0081] The first similarity parameter and the second similarity parameter are combined to obtain the comprehensive similarity parameter of the target dimension between the first patient and the second patient.
[0082] In embodiments of the present invention, the sum or product of the first similarity parameter and the second similarity parameter can be used as a comprehensive similarity parameter of the target dimension between the first patient and the second patient to achieve a comprehensive understanding of the two, without limitation.
[0083] Using the same method described above, the comprehensive similarity parameters for each dimension between the first and second patients can be obtained. Then, the sum of the comprehensive similarity parameters for all dimensions between the first and second patients can be normalized, limiting the calculation results to... Within this range, the similarity between the first and second patients can be obtained.
[0084] In one embodiment of the present invention, the normalization process can be specifically, for example, maximum and minimum value normalization. Furthermore, the normalization in subsequent steps can all adopt maximum and minimum value normalization. In other embodiments of the present invention, other normalization methods can be selected according to the specific range of the numerical values, which will not be elaborated further.
[0085] As an example, in one embodiment of the present invention, the expression for the degree of similarity between the first patient and the second patient may specifically be as follows:
[0086]
[0087] in, Indicates the degree of similarity between the first patient and the second patient; Indicates the first patient's... The importance percentage of each dimension; Indicating the second patient's... The importance percentage of each dimension; Indicates the relationship between the first patient and the second patient. The first similarity parameter in each dimension; Indicates the first patient's... Diagnostic and treatment data from multiple dimensions; Indicating the second patient's... Treatment data from multiple dimensions; Indicates the relationship between the first patient and the second patient. The second similarity parameter in each dimension; Indicates the number of all dimensions; Represents the normalization function; and These represent the first preset adjustment parameter and the second preset adjustment parameter, respectively, to prevent the denominator from being 0. and The range of values is In one embodiment of the present invention, the following is used: and All were set to 0.01. and The specific values can be set by the implementer according to the specific implementation scenario, and are not limited here.
[0088] It should be noted that negative correlation mapping can also be achieved through other basic mathematical operations in other embodiments of the present invention, which will not be elaborated here.
[0089] The similarity between any two patients can be obtained using the same method described above.
[0090] Step S3: Based on the similarity, perform iterative clustering on all patients to obtain multiple clustering results, each containing multiple clusters; take any one of the clustering results as the target clustering result, and obtain the optimization degree of the target clustering result based on the similarity between each patient in each cluster and the patient corresponding to the cluster center, as well as the important evaluation factors of the same dimension for each patient in each cluster.
[0091] To improve the clustering effect for patients, this embodiment of the invention performs iterative clustering on all patients based on similarity, obtaining multiple clustering results. The clustering effect of each clustering result can then be analyzed to select the clustering result with the best effect, thereby maximizing the clustering effect for patients and improving the effectiveness of subsequent personalized processing of patients' diagnosis and treatment data.
[0092] Preferably, in one embodiment of the present invention, the method for obtaining multiple clustering results specifically includes:
[0093] Using an iterative self-organizing clustering algorithm, all patients are clustered for the first time based on the similarity between any two patients, and the first clustering result is used as the current clustering result. The iterative self-organizing clustering algorithm is a well-known technique in the art and will not be described in detail here.
[0094] Any cluster in the current clustering result is taken as the cluster to be tested. Based on the diagnosis and treatment data of patients in the cluster to be tested in various dimensions, the cluster center patient of the cluster to be tested is selected from all patients in the cluster to be tested. Subsequently, the weight parameters of each patient in the current clustering result and the clustering effect of the current clustering result can be analyzed based on the similarity between the patients in the cluster to be tested and the cluster center patient. Since the patients corresponding to the cluster centers of the clusters will change during the iterative clustering process, the error will accumulate and the iterative clustering effect will decrease as the iteration progresses. Therefore, firstly, any patient in the cluster to be tested is taken as the test patient. Based on the similarity between the test patient and the cluster center patient of the cluster to be tested, as well as the differences in the important evaluation factors of the same dimension of each patient in the cluster to be tested, the weight parameters of the test patient in the current clustering result are obtained. The larger the weight parameter of the test patient, the more the test patient needs to be retained in the next iteration of clustering, and thus the higher the weight of the test patient in the next iteration of clustering.
[0095] Preferably, in one embodiment of the present invention, the method for obtaining the cluster center patients of the cluster to be tested specifically includes:
[0096] First, the average value of the same dimension of the diagnosis and treatment data of all patients in the cluster to be tested is used as the second overall diagnosis and treatment data of each dimension in the cluster to be tested. The second overall diagnosis and treatment data reflects the overall level of the same dimension of the diagnosis and treatment data of all patients in the cluster to be tested.
[0097] Then, the absolute value of the difference between the target dimension diagnosis and treatment data and the second overall diagnosis and treatment data of the target dimension for each patient in the cluster to be tested is taken as the second data deviation value of the target dimension for each patient in the cluster to be tested. The second data deviation value of each dimension for each patient in the cluster to be tested can be obtained by the same method described above. Then, the average value of the second data deviation values of all dimensions for each patient in the cluster to be tested can be taken as the center deviation value of each patient in the cluster to be tested. The smaller the center deviation value of a patient in the cluster to be tested, the more likely the patient is to be in the center position of the cluster to be tested. Therefore, the patient corresponding to the minimum center deviation value can be taken as the cluster center patient of the cluster to be tested.
[0098] The same method described above can be used to obtain the cluster center of each cluster in the current clustering results.
[0099] Preferably, in one embodiment of the present invention, the method for obtaining the weight parameters of the patients to be tested in the current clustering results specifically includes:
[0100] First, determine the similarity between the patient to be tested and the patients at the cluster centers of the clusters to be tested. The smaller the similarity, the greater the distance between the patient to be tested and the patients at the cluster centers of the clusters to which it belongs, and thus the less likely the patient to be tested is to be retained in the next iteration of clustering.
[0101] Then, the average value of the important evaluation factors of the same dimension for all patients in the cluster to be tested is taken as the overall important evaluation factor of each dimension of the cluster to be tested. The overall important evaluation factor reflects the overall level of the important evaluation factors of the same dimension for all patients in the cluster to be tested.
[0102] The absolute value of the difference between the important evaluation factors of each dimension of the patient under test and the overall important evaluation factors of each dimension of the cluster under test is used as the importance deviation value of each dimension of the patient under test. The larger the importance deviation value, the greater the difference between the important evaluation factors of each dimension of the patient under test and other patients. This means that the patient under test is less likely to be retained in the next iteration of clustering. Therefore, the average of the importance deviation values of all dimensions of the patient under test can be used as the classification necessity of the patient under test.
[0103] Finally, the similarity between the patient to be tested and the patients at the cluster centers of the clusters to be tested, as well as the classification necessity, are comprehensively processed to obtain the weight parameters of the patient to be tested in the current clustering results. Among them, the weight parameters of the patient to be tested in the current clustering results are positively correlated with the similarity between the patient to be tested and the patients at the cluster centers of the clusters to be tested, and negatively correlated with the classification necessity.
[0104] As an example, in one embodiment of the present invention, the expression for the weight parameter of the patient to be tested in the current clustering result can be specifically as follows:
[0105]
[0106] in, This represents the weight parameter of the patient to be tested in the current clustering results; This indicates the degree of similarity between the patient being tested and the patients at the cluster centers of the tested clusters; Indicates the first [number] of the patient to be tested Key evaluation factors in each dimension; The first cluster to be tested represents the cluster of the target cluster. The overall important evaluation factors of each dimension; Indicates the number of all dimensions; Indicates the first [number] of the patient to be tested The importance deviation values of each dimension; This indicates the necessity of classifying the patients to be tested.
[0107] The weight parameters for each patient in the current clustering result can be obtained using the same method described above.
[0108] After obtaining the weight parameters of each patient, the average of the weight parameters of any two patients is calculated as the overall weight of those two patients. This overall weight is then multiplied by the similarity between the two patients to obtain a weighted similarity. The iterative self-organizing clustering algorithm is then used again to perform the next clustering of all patients based on the weighted similarity between any two patients, obtaining the next clustering result. If the termination condition of iterative clustering is not met, the next clustering result is used as the current clustering result, and iterative clustering continues. Otherwise, iterative clustering stops. After the iteration ends, multiple clustering results are obtained. The termination condition of iterative clustering includes, for example, whether the cluster center patients have changed (i.e., when all cluster center patients no longer change, meaning the cluster center patients remain essentially unchanged between two iterations), or the maximum number of iterations (when the maximum number of iterations of the iterative self-organizing clustering algorithm is reached, the iteration terminates). These conditions are not limited here, as the iterative self-organizing clustering algorithm and its termination conditions are well-known techniques to those skilled in the art and will not be elaborated upon here.
[0109] It should be noted that during the overall iterative clustering process, after obtaining a new clustering result, it is necessary to extract and update the cluster center patients of each cluster in the new clustering result, and calculate and update the weight parameters of each patient in the new clustering result.
[0110] The iterative clustering process described above yielded multiple clustering results, each with varying effectiveness. Therefore, to select the optimal clustering result, it is necessary to evaluate the effectiveness of each clustering result. Thus, any single clustering result can be used as the target clustering result. The similarity between each patient in each cluster and the patient corresponding to the cluster center, as well as the important evaluation factors of the same dimensions for each patient in each cluster, can be analyzed. The degree of optimization reflects the effectiveness of the target clustering result; a higher degree of optimization indicates a better result. Subsequently, the optimal clustering result can be selected based on the degree of optimization, improving the effectiveness of classifying and storing patient diagnosis and treatment data.
[0111] Preferably, in one embodiment of the present invention, the method for obtaining the degree of preference of the target clustering result specifically includes:
[0112] First, any cluster in the target clustering results is taken as the target cluster. The product of the similarity between each patient in the target cluster and the patient at the cluster center of the target cluster and the weight parameter of each patient is taken as the update stability of each patient in the target cluster. The higher the update stability of each patient in the target cluster, the more similar the patients in the target cluster are to the patients at the cluster center, and the more they need to be retained. This indicates that the clustering structure of the target cluster is more stable, and the clustering effect of the target cluster is better. Therefore, the average update stability of all patients in the target cluster can be taken as the clustering effect evaluation score of the target cluster. The higher the clustering effect evaluation score, the better the clustering effect of the target cluster.
[0113] By using the same method described above, the clustering effect evaluation score of each cluster in the target clustering result can be obtained. Then, the average value of the clustering effect evaluation scores of all clusters in the target clustering result can be used as the degree of preference of the target clustering result.
[0114] As an example, in one embodiment of the present invention, the expression for the degree of preference of the target clustering result can be specifically as follows:
[0115]
[0116]
[0117] in, Indicates the degree of preference for the target clustering results; Indicating the first cluster in the target clustering results Evaluation of the clustering effect of each cluster; This indicates the number of clusters in the target clustering result; This indicates the evaluation score of the clustering effect of the target cluster; Represents the first in the target cluster The degree of similarity between individual patients and the patients who are the cluster centers of the target cluster; Represents the first in the target cluster Weight parameters for each patient; Represents the first in the target cluster Update stability of individual patients; This indicates the number of patients in the target cluster.
[0118] The same method described above can be used to determine the optimality of each clustering result.
[0119] Step S4: Based on the optimization of each clustering result, select the best clustering result from all clustering results, and store the diagnosis and treatment data of patients in the same cluster in the best clustering result in the database.
[0120] The higher the degree of optimization of a clustering result, the better the effect of that clustering. Therefore, in one embodiment of the present invention, the clustering result corresponding to the maximum degree of optimization is taken as the best clustering result.
[0121] Then, the diagnostic and treatment data of patients in the same cluster in the best clustering results can be stored in the database to realize the classified storage of the diagnostic and treatment data of each patient, thereby improving the effect of personalized processing of the diagnostic and treatment data of each patient.
[0122] One embodiment of the present invention provides an intelligent processing system for urological diagnosis and treatment data. The system includes a memory, a processor, and a computer program. The memory is used to store the corresponding computer program, and the processor is used to run the corresponding computer program. When the computer program runs in the processor, it can implement the methods described in steps S1 to S4.
[0123] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.
[0124] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.
Claims
1. A method for intelligent processing of urological diagnostic and treatment data, characterized in that, The method includes: To acquire diagnostic and treatment data from different dimensions for each patient in the urology department; Using any patient as the target patient and any dimension as the target dimension, the important evaluation factors of the target dimension of the target patient are obtained based on the differences in the diagnostic and treatment data of the target dimension between the target patient and other patients, as well as the correlation between the diagnostic and treatment data of the target dimension and other dimensions. The similarity between any two patients is obtained based on the differences in the diagnostic and treatment data of the same dimension between any two patients and the differences in the important evaluation factors. Based on the similarity, iterative clustering is performed on all patients to obtain multiple clustering results, each containing multiple clusters. Any clustering result is taken as the target clustering result. The optimization degree of the target clustering result is obtained based on the similarity between each patient in each cluster and the patient corresponding to the cluster center, as well as the important evaluation factors of the same dimension for each patient in each cluster. Based on the optimization of each clustering result, the best clustering result is selected from all clustering results, and the diagnosis and treatment data of patients in the same cluster in the best clustering result are stored in the database. Methods for determining key evaluation factors include: The average value of the diagnostic and treatment data for all patients in the target dimension is taken as the first overall diagnostic and treatment data for the target dimension; The absolute value of the difference between the target patient's target dimension diagnosis and treatment data and the first overall diagnosis and treatment data is used as the first data deviation value of the target patient's target dimension. The first data deviation value of the target patient's target dimension is used as the numerator, and the sum of the first data deviation values of all dimensions of the target patient is used as the denominator. The ratio is used as the first importance of the target patient's target dimension. The sequence of diagnostic and treatment data for all patients in the same dimension is taken as the diagnostic and treatment data sequence for each dimension. Using dimensions other than the target dimension as reference dimensions, the correlation between the diagnosis and treatment data sequences of the target dimension and the diagnosis and treatment data sequences of each reference dimension is analyzed to obtain the second importance of the target dimension. The product of the first importance and the second importance of the target dimension of the target patient is used as an important evaluation factor for the target dimension of the target patient. Methods for determining the second degree of importance include: The absolute value of the Pearson correlation coefficient between the diagnostic and treatment data sequence of the target dimension and the diagnostic and treatment data sequence of each reference dimension is taken as the data correlation between the target dimension and each reference dimension. The average of the data relevance between the target dimension and all reference dimensions is used as the second importance of the target dimension.
2. The intelligent processing method for urological diagnostic and treatment data according to claim 1, characterized in that, The method of obtaining the similarity between any two patients includes: The important evaluation factors of the target dimension for each patient are used as the numerator, the cumulative value of the important evaluation factors of all dimensions for each patient is used as the denominator, and the ratio is used as the importance percentage of the target dimension for each patient. Two patients are randomly selected as the first patient and the second patient. The absolute value of the difference between the importance ratio of the target dimension between the first patient and the second patient is negatively correlated to obtain the first similarity parameter of the target dimension between the first patient and the second patient. A negative correlation mapping is performed on the absolute values of the differences in the diagnostic and treatment data of the first patient and the second patient in the target dimension to obtain a second similarity parameter between the first patient and the second patient in the target dimension; The first similarity parameter and the second similarity parameter are combined to obtain the comprehensive similarity parameter of the target dimension between the first patient and the second patient; The sum of the comprehensive similarity parameters across all dimensions between the first and second patients is normalized to obtain the degree of similarity between the first and second patients.
3. The intelligent processing method for urological diagnostic and treatment data according to claim 1, characterized in that, The process of obtaining multiple clustering results includes: Using the iterative self-organizing clustering algorithm, based on the similarity between patients, all patients are first clustered to obtain the first clustering result, and the first clustering result is used as the current clustering result; Take any cluster in the current clustering result as the cluster to be tested. Based on the diagnosis and treatment data of patients in the cluster to be tested in each dimension, select the cluster center patient of the cluster to be tested from all patients in the cluster to be tested. Take any patient in the cluster to be tested as the patient to be tested. Based on the similarity between the patient to be tested and the cluster center patient of the cluster to be tested, and the difference of the important evaluation factors of the same dimension of each patient in the cluster to be tested, obtain the weight parameters of the patient to be tested in the current clustering result. Using the iterative self-organizing clustering algorithm, based on the weight parameters of each patient in the current clustering result and the similarity between patients, the next clustering is performed on all patients to obtain the next clustering result. If the termination condition of iterative clustering is not met, the next clustering result is used as the current clustering result and iterative clustering continues; otherwise, iterative clustering is stopped.
4. The intelligent processing method for urological diagnostic and treatment data according to claim 3, characterized in that, The patients selected as cluster center patients from all patients in the cluster to be tested include: The average value of the diagnostic and treatment data of the same dimension of all patients in the cluster to be tested is used as the second overall diagnostic and treatment data of each dimension in the cluster to be tested. The absolute value of the difference between the diagnostic and treatment data of the target dimension and the second overall diagnostic and treatment data of the target dimension for each patient in the cluster to be tested is used as the second data deviation value of the target dimension for each patient in the cluster to be tested. The average of the second data deviation values of all dimensions for each patient in the cluster to be tested is used as the center deviation value for each patient in the cluster to be tested; In the cluster to be tested, the patient corresponding to the minimum value of the center deviation is taken as the cluster center patient of the cluster to be tested.
5. The intelligent processing method for urological diagnostic and treatment data according to claim 3, characterized in that, The weight parameters for obtaining the patients to be tested in the current clustering results include: The average value of the important evaluation factors of the same dimension for all patients in the cluster to be tested is used as the overall important evaluation factor for each dimension of the cluster to be tested. The absolute value of the difference between the important evaluation factor of each dimension of the patient under test and the overall important evaluation factor of each dimension of the cluster under test is used as the importance deviation value of each dimension of the patient under test; the average value of the importance deviation values of all dimensions of the patient under test is used as the classification necessity of the patient under test. The similarity between the patient to be tested and the patients at the cluster centers of the clusters to be tested, as well as the classification necessity, are comprehensively processed to obtain the weight parameters of the patient to be tested in the current clustering results. The weight parameters of the patient to be tested in the current clustering results are positively correlated with the similarity between the patient to be tested and the patients at the cluster centers of the clusters to be tested, and negatively correlated with the classification necessity.
6. The intelligent processing method for urological diagnostic and treatment data according to claim 3, characterized in that, The degree of preference for obtaining the target clustering results includes: Take any cluster in the target clustering results as the target cluster, and take the product of the similarity between each patient in the target cluster and the cluster center patient of the target cluster and the weight parameter of each patient as the update stability of each patient in the target cluster. The average of the update stability of all patients in the target cluster is used as the evaluation score of the clustering effect of the target cluster. The average value of the clustering effect evaluation of all clusters in the target clustering result is taken as the degree of preference of the target clustering result.
7. The intelligent processing method for urological diagnostic and treatment data according to claim 1, characterized in that, The process of selecting the best clustering result from all clustering results includes: The clustering result corresponding to the maximum value of the preferred degree is taken as the best clustering result.
8. A urological diagnostic and treatment data intelligent processing system, the system comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the method as described in any one of claims 1 to 7.