Long-term intelligent follow-up system for patients with chronic airway diseases

Abstract

The present invention relates to the field of chronic airway disease follow-up, and specifically to a long-term intelligent follow-up system for patients with chronic airway diseases. The system first obtains the scores of each question in different test items of patients of different patient types during the current follow-up and multiple historical follow-ups, as well as the index data of multiple detection indicators of each patient during the current follow-up, and obtains the first individual parameter and the second individual parameter of the patient based on the scores of the patient in the test items during the current follow-up and historical follow-up, obtains multiple clusters of detection indicators, analyzes the distribution of indicator data of the detection indicators of patients of each patient type in the clustering cluster, and the difference in the number of patients of the same patient type between different clustering clusters, and combines the first individual parameter and the second individual parameter to obtain the patient discrimination of the detection indicator, and adjusts and clusters the distance measurement between patients. The present invention can improve the accuracy of the follow-up system for chronic airway diseases and cluster patients.

Description

Technical Field

[0001] The present invention relates to the field of chronic airway disease follow-up, and in particular to a long-term intelligent follow-up system for patients with chronic airway diseases. Background Art

[0002] Chronic airway diseases are characterized by a long course of illness, repeated attacks, and long treatment cycles, which have a serious impact on the patient's quality of life. At the same time, chronic airway diseases include many different types such as chronic obstructive pulmonary disease, asthma, bronchiectasis, etc., and their types are relatively complex and patients are highly heterogeneous. Regular follow-up of chronic airway diseases helps control the disease and improve prognosis. Therefore, long-term follow-up of patients with chronic airway diseases is required in order to conduct in-depth analysis of the characteristics of the patient's chronic airway diseases and guide treatment.

[0003] In the related art, during long-term follow-up, patients with chronic airway diseases are usually indiscriminately clustered based on their various test index data, so as to achieve targeted and personalized management of patients with similar disease manifestations in the same cluster. However, since there are many types of chronic airway diseases and they are often accompanied by a variety of hidden secondary diseases, some test indicators are universal in various types of chronic airway diseases and are less targeted to specific types of chronic airway diseases. As a result, these test indicators have a dilution effect on important test indicators during the clustering process, which in turn makes it impossible to accurately cluster patients using existing clustering algorithms in the long-term follow-up system for patients with chronic airway diseases. Summary of the invention

[0004] In order to solve the technical problem that the existing clustering algorithm cannot accurately cluster patients in the long-term follow-up system of patients with chronic airway diseases, the purpose of the present invention is to provide a long-term intelligent follow-up system for patients with chronic airway diseases. The technical solution adopted is as follows:

[0005] The present invention proposes a long-term intelligent follow-up system for patients with chronic airway diseases, the system comprising:

[0006] The data collection module is used to obtain the scores of each question in different test items of patients of different types of patients in the current follow-up and multiple historical follow-ups within a preset time period, and to obtain the index data of multiple test indicators of each patient in the current follow-up;

[0007] The individual analysis module is used to take any patient as a target patient, and obtain a first individual parameter of the target patient according to the scores of all questions in each test item of the target patient at the current follow-up, and the difference in scores of the target patient and other patients except the target patient in the same test item at the current follow-up; obtain a second individual parameter of the target patient according to the scores of all questions in the same test item of the target patient at each historical follow-up, the time sequence of the historical follow-ups, and the difference in scores of questions in the same test item between each historical follow-up and the current follow-up;

[0008] A cluster analysis module, for taking any detection index as a target detection index, clustering all patients according to the difference of the index data of the target detection index between different patients, and obtaining a plurality of clustering clusters about the target detection index; obtaining the patient discrimination of the target detection index according to the distribution of the index data of the target detection index of each patient of each patient type in each clustering cluster, the first individual parameter and the second individual parameter of each patient of each patient type, and the difference in the number of patients of the same patient type between different clustering clusters;

[0009] The clustering adjustment module is used to adjust the difference in indicator data of the same detection indicator between patients according to the patient discrimination of each detection indicator to obtain an adjusted distance metric between patients; based on the adjusted distance metric between patients, all patients are clustered to obtain multiple optimized clustering clusters.

[0010] Further, obtaining a first physical parameter of the target patient includes:

[0011] The sum of the scores of all questions in each test item of the target patient at the current follow-up is used as the total score of each test item of the target patient at the current follow-up; the average of the total scores of all test items of the target patient at the current follow-up is used as the disease severity of the target patient at the current follow-up;

[0012] Obtaining the degree of individual difference of the target patient at the current follow-up according to the difference in scores of the same questions in the same test items between the target patient and other patients at the current follow-up, and the difference in the total scores of the same test items between the target patient and other patients at the current follow-up;

[0013] The severity of the disease and the individual difference are integrated and normalized to obtain the first individual parameter of the target patient.

[0014] Furthermore, obtaining the individual difference degree of the target patient at the current follow-up includes:

[0015] Any other patient except the target patient is used as the other patient to be tested, and any test item is used as the target test item;

[0016] The cumulative value of the absolute value of the difference between the scores of all the same questions in the target test items between the target patient and other patients to be tested at the current follow-up is used as the patient characteristic difference degree of the target test items between the target patient and other patients to be tested at the current follow-up;

[0017] Performing negative correlation normalization processing on the absolute value of the difference in the total score of the target test item between the target patient and other patients to be tested at the current follow-up, to obtain the reference weight of the target test item between the target patient and other patients to be tested at the current follow-up;

[0018] Using the reference weight of each test item between the target patient and other patients to be tested at the current follow-up, weighted summing the patient characteristic difference degree of each test item between the target patient and other patients to be tested at the current follow-up is performed to obtain the initial difference degree between the target patient and other patients to be tested;

[0019] The cumulative value of the initial difference degrees between the target patient and all other patients is taken as the individual difference degree of the target patient at the current follow-up.

[0020] Further, obtaining the second individual parameter of the target patient includes:

[0021] Based on the calculation method of the total score of each test item of the target patient at the current follow-up, the total score of each test item of the target patient at each historical follow-up is obtained;

[0022] Using non-zero natural numbers, numbering each historical follow-up of the target patient in time sequence to obtain the serial number of each historical follow-up of the target patient; using the serial number of each historical follow-up of the target patient and the total score of the target test item of the target patient at each historical follow-up as the symptom manifestation sequence of the target patient with respect to the target test item at each historical follow-up;

[0023] Based on the calculation method of the patient characteristic difference degree of the target test item between the target patient and other patients to be tested at the current follow-up, according to the difference in the score of the target patient in the same question in the target test item between each historical follow-up and the current follow-up, the patient characteristic difference parameter of the target test item between each historical follow-up and the current follow-up is obtained, and the patient characteristic difference parameter is normalized to obtain the weight parameter of the target patient with respect to the target test item at each historical follow-up;

[0024] Input the symptom manifestation sequence and the weight parameter of the target patient in all historical follow-ups with respect to the target test items into the weighted PCA algorithm to obtain the principal component direction of the target patient with respect to the target test items; take the average of the principal component directions of the target patient with respect to all test items as the overall principal component direction of the target patient;

[0025] The slope of the overall principal component direction of the target patient is normalized to obtain a second individual parameter of the target patient.

[0026] Furthermore, the step of obtaining a plurality of clusters of target detection indicators includes:

[0027] The absolute value of the difference between the target detection index data of any two patients is used as the index distance measurement between any two patients with respect to the target detection index;

[0028] Using the K-means clustering algorithm, all patients are clustered based on the indicator distance metric regarding the target detection indicator between any two patients to obtain multiple clusters regarding the target detection indicator.

[0029] Furthermore, the patient discrimination of the target detection index is obtained including:

[0030] Taking any clustering cluster about the target detection index as the target clustering cluster, taking the average value of the index data of the target detection index of all patients in the target clustering cluster as the cluster center value of the target clustering cluster; taking the average value of the index data of the target detection index of all patients of each patient type in the target clustering cluster as the category center value of each patient type in the target clustering cluster;

[0031] Obtaining a first discrimination degree of the target detection indicator in the target cluster according to the distribution of the indicator data of the target detection indicator of each patient of each patient type in the target cluster, the first individual parameter and the second individual parameter of each patient of each patient type, and the difference between the category center value of each patient type in the target cluster and the cluster center value of the target cluster;

[0032] According to the difference in the number of patients of the same patient type between the target cluster and other clusters except the target cluster, a second discrimination degree of the target detection index in the target cluster is obtained;

[0033] Combining the first discrimination and the second discrimination to obtain a comprehensive discrimination of the target detection index in the target cluster;

[0034] The average value of the comprehensive discrimination of the target detection index in all clusters is normalized to obtain the patient discrimination of the target detection index.

[0035] Furthermore, obtaining a first discrimination degree of the target detection index in the target cluster includes:

[0036] The first individual parameter and the second individual parameter of each patient in the target cluster are integrated and normalized to obtain a weight coefficient of each patient in the target cluster;

[0037] Based on the calculation formula of the first discrimination, the first discrimination of the target detection index in the target cluster is obtained. The calculation formula of the first discrimination is:

[0038] ;

[0039] in, Indicates the first discrimination of the target detection index in the target cluster; Indicates the first The category center value of each patient type; Indicates the cluster center value of the target cluster; Indicates the first Patient type The target detection index data of each patient; Indicates the first Patient type Weight coefficient for each patient; Indicates the first the number of patients with each patient type; Indicates the number of patient types in the target cluster; represents the normalization function; Represents the adjustment parameter, the value range is .

[0040] Furthermore, obtaining a second discrimination degree of the target detection index in the target cluster includes:

[0041] The number of patients of each patient type in each cluster is taken as the numerator, the number of patients in each cluster is taken as the denominator, and the ratio is taken as the proportion of the number of each patient type in each cluster;

[0042] Based on the calculation formula of the second discrimination, the second discrimination of the target detection index in the target cluster is obtained. The calculation formula of the second discrimination is:

[0043] ;

[0044] in, Indicates the second discrimination of the target detection index in the target cluster; Indicates the first The proportion of each patient type; Indicates the clusters other than the target cluster The other clusters The proportion of each patient type; Indicates the number of clusters other than the target cluster; Indicates the number of patient types in the target cluster; Represents the normalization function.

[0045] Further, obtaining the adjusted distance metric between patients includes:

[0046] Based on the calculation formula of the adjusted distance metric, the adjusted distance metric between patients is obtained, and the calculation formula of the adjusted distance metric is:

[0047] ;

[0048] in, Indicates Patients and The adjusted distance measure between patients, ; Indicates The patient's The indicator data of each detection indicator; Indicates The patient's The indicator data of each detection indicator; Indicates The patient discrimination of each test indicator; Indicates the number of detection indicators.

[0049] Furthermore, clustering all patients based on the adjusted distance metric between patients to obtain multiple optimized clusters includes:

[0050] Using the K-means clustering algorithm, all patients are clustered based on the adjusted distance metric between patients to obtain multiple optimized clustering clusters.

[0051] The present invention has the following beneficial effects:

[0052] The present invention takes into account that in the long-term follow-up system of patients with chronic airway diseases, the existing clustering algorithm cannot accurately cluster patients. First, the scores of multiple questions in multiple test items of patients with different types of patients in the current follow-up and multiple historical follow-ups, as well as the index data of multiple detection indicators of each patient in the current follow-up, are obtained. Since there are individual differences in symptom manifestations among different patients and the severity of symptom manifestations of different patients is different, the first individual parameter is first used to reflect the individual differences in symptom manifestations of target patients and the severity of symptom manifestations of target patients, so as to improve the accuracy of subsequent cluster analysis. Considering that the symptom manifestations of patients may improve or worsen in the long-term process, and may even cause complications, the second individual parameter can be used to reflect the individual differences in symptom manifestations of target patients and the severity of symptom manifestations of target patients. The body parameters reflect the possibility of long-term deterioration and complications of the target patient's symptoms, which is convenient for improving the accuracy of subsequent clustering analysis. Taking into account that different detection indicators have different degrees of discrimination for different types of chronic airway diseases, some detection indicators are universal in various types of chronic airway diseases, and the discrimination of such detection indicators is low. The reference of such detection indicators should be reduced in the subsequent patient clustering process. Therefore, the present invention first obtains multiple clustering clusters about the target detection indicator, and reflects the discrimination of the target detection indicator for different patient types through the obtained patient discrimination, and then adjusts the distance measurement between patients through the patient discrimination of each detection indicator, thereby improving the reference of the detection indicator with higher discrimination in the clustering process, thereby improving the accuracy of the final clustering result. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0054] Figure 1 A block diagram of a long-term intelligent follow-up system for patients with chronic airway diseases provided by one embodiment of the present invention;

[0055] Figure 2 A flow chart of a method for obtaining patient discrimination of target detection indicators provided by one embodiment of the present invention. DETAILED DESCRIPTION

[0056] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the long-term intelligent follow-up system for patients with chronic airway diseases proposed by the present invention, its specific implementation method, structure, characteristics and effects are described in detail as follows in combination with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.

[0057] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0058] The specific scheme of the long-term intelligent follow-up system for patients with chronic airway diseases provided by the present invention is described in detail below with reference to the accompanying drawings.

[0059] See also Figure 1 , which shows a block diagram of a long-term intelligent follow-up system for patients with chronic airway diseases provided by an embodiment of the present invention. The system includes: a data acquisition module 101, an individual analysis module 102, a cluster analysis module 103, and a cluster adjustment module 104.

[0060] The data collection module 101 is used to obtain the scores of each question in different test items of patients of different patient types in the current follow-up and multiple historical follow-ups within a preset time period, and to obtain the indicator data of multiple detection indicators of each patient in the current follow-up.

[0061] Chronic airway diseases are characterized by long course, repeated attacks, and long treatment cycles. Since chronic airway diseases include many types of patients, such as COPD, asthma, and bronchiectasis, and may also be accompanied by subtle secondary diseases such as lung disease, pulmonary vascular disease, and sleep breathing disease, in the long-term follow-up of patients with chronic airway diseases in hospitals, it is usually necessary to use test items such as CAT questionnaires, mMRC questionnaires, or SGRQ questionnaires to score the patient's symptoms, and then record the patient's score results for different test items at each follow-up and save them in the database, where each test item contains multiple questions about symptom manifestations.

[0062] Therefore, the embodiment of the present invention first extracts the scores of each question in different test items of patients of different patient types in the current follow-up and multiple historical follow-ups within a preset time period from the hospital database, wherein the current follow-up can be considered as the most recent follow-up, and the multiple historical follow-ups can be considered as the multiple follow-ups before the most recent follow-up. The preset time period is set to 1 year, and the specific value of the preset time period can also be set by the implementer according to the specific implementation scenario, which is not limited here.

[0063] It should be noted that in actual follow-up, patients are generally asked to fill in scores for various test items through a website, and the types and numbers of test items on the website are generally fixed. Therefore, the types and numbers of test items are the same for different patients, and the types and numbers of test items for the same patient at different follow-up visits are also the same.

[0064] At the same time, during the follow-up of patients in the hospital, indicator data of various test indicators of the patients will also be collected and recorded and saved in the database. The test indicators include FEV1, FVC, FEV1 / FVC ratio, PEF variation rate, serum total IgE level, blood eosinophil count, etc. in lung function. Therefore, the embodiment of the present invention also needs to extract the indicator data of multiple test indicators of each patient at the current follow-up from the hospital database.

[0065] At this point, the patient's score and indicator data have been obtained. Subsequently, the score and indicator data can be integrated to accurately analyze the discrimination of each detection indicator for different types of chronic airway diseases and improve the accuracy of clustering.

[0066] The individual analysis module 102 is used to take any patient as a target patient, and obtain a first individual parameter of the target patient according to the scores of all questions in each test item of the target patient at the current follow-up, and the difference in scores of questions in the same test item between the target patient and other patients except the target patient at the current follow-up; and obtain a second individual parameter of the target patient according to the scores of all questions in the same test item of the target patient at each historical follow-up, the time sequence of the historical follow-ups, and the difference in scores of questions in the same test item between each historical follow-up and the current follow-up.

[0067] Since there are individual differences in symptom manifestations among different patients, and the severity of symptom manifestations among different patients is different, and the symptom manifestations of patients can be reflected by the scores of questions in different test items, the embodiment of the present invention first takes any patient as the target patient, and analyzes the scores of all questions in each test item of the target patient at the current follow-up, as well as the differences in scores of questions in the same test item between the target patient and other patients except the target patient at the current follow-up, and reflects the individual differences in symptom manifestations of the target patient and the severity of the symptom manifestations of the target patient by obtaining the first individual parameter. In the subsequent cluster analysis process, the discrimination analysis of the detection index can be weighted and adjusted based on the first individual parameter to improve the calculation accuracy of the discrimination of the detection index.

[0068] Preferably, in one embodiment of the present invention, the method for acquiring the first individual parameter of the target patient specifically includes:

[0069] First, since the larger the score of each question in the test item, the more severe the symptoms of the target patient, the sum of the scores of all questions in each test item of the target patient at the current follow-up can be used as the total score of each test item of the target patient at the current follow-up, and the average of the total scores of all test items of the target patient at the current follow-up can be used as the disease severity of the target patient at the current follow-up. The larger the patient severity, the more severe the symptoms of the chronic airway disease suffered by the target patient.

[0070] As an example, in one embodiment of the present invention, the expression of the severity of the target patient's illness at the current follow-up visit may be specifically, for example, as follows:

[0071] ;

[0072] in, Indicates the severity of the target patient's illness at the current follow-up; Indicates the target patient’s first The total score of the test items; Indicates the number of test items.

[0073] Then, based on the difference in scores of the same questions in the same test items between the target patient and other patients at the current follow-up, as well as the difference in total scores of the same test items between the target patient and other patients at the current follow-up, the individual difference degree of the target patient at the current follow-up is obtained. The greater the individual difference degree, the greater the difference in the symptom manifestation of the target patient from that of other patients. Subsequently, the first individual parameter of the target patient can be accurately calculated in combination with the severity of the disease and the individual difference degree.

[0074] Preferably, in one embodiment of the present invention, the method for obtaining the individual difference degree of the target patient at the current follow-up specifically includes:

[0075] Any other patient except the target patient is taken as the other patient to be tested, any test item is taken as the target test item, and the accumulated value of the absolute value of the difference in scores of all the same questions in the target test items at the current follow-up between the target patient and the other patients to be tested is taken as the patient characteristic difference between the target patient and the other patients to be tested in the target test items at the current follow-up. The greater the patient characteristic difference, the greater the difference in the manifestation of the same symptom between the target patient and the other patients to be tested in the target test items at the current follow-up, and further the greater the individual differences shown by the target patients.

[0076] The absolute value of the difference in the total score of the target test item between the target patient and other patients to be tested at the current follow-up is negatively normalized to obtain the reference weight of the target test item between the target patient and other patients to be tested at the current follow-up. The larger the reference weight, the closer the severity of the symptom manifestations of the target test item between the target patient and other patients to be tested at the current follow-up, and the greater the reference value of the patient characteristic difference obtained above, and the reference weight can be used subsequently to improve the accuracy of the calculation of the individual difference degree.

[0077] In one embodiment of the present invention, the natural constant e The negative exponential function with the base value is used to realize the normalization of negative correlation. In addition, the normalization of negative correlation in subsequent steps can be carried out by using the natural constant e The negative exponential function with base is processed.

[0078] The same method as above can be used to obtain the patient characteristic difference degree of each test item between the target patient and other patients to be tested at the current follow-up, as well as the reference weight of each test item between the target patient and other patients to be tested at the current follow-up. Then, the reference weight of each test item between the target patient and other patients to be tested at the current follow-up can be used to perform weighted summation of the patient characteristic difference degree of each test item between the target patient and other patients to be tested at the current follow-up to obtain the initial difference degree between the target patient and other patients to be tested. The greater the initial difference degree, the greater the difference in symptom manifestations between the target patient and other patients to be tested.

[0079] The initial degree of difference between the target patient and each other patient can be obtained by the same method as described above, and then the cumulative value of the initial degree of difference between the target patient and all other patients can be used as the individual degree of difference of the target patient at the current follow-up.

[0080] As an example, in one embodiment of the present invention, the expression of the degree of individual difference of the target patients can be specifically, for example, as follows:

[0081] ;

[0082] ;

[0083] in, Indicates the degree of individual differences among target patients at the current follow-up; Indicates the target patient and the initial degree of difference between the other patients; represents the number of patients other than the target patient; represents the initial degree of difference between the target patient and the other patients to be tested; Indicates the difference between the target patient and other patients to be tested at the current follow-up. The difference of patient characteristics among the test items; Indicates the target patient’s first The total score of the test items; Indicates the other patients to be tested at the time of the current follow-up. The total score of the test items; Indicates the difference between the target patient and other patients to be tested at the current follow-up. Reference weights of various test items; Indicates the number of test items; Indicated by natural constant An exponential function with base .

[0084] Finally, the severity of the disease and the degree of individual differences are combined and normalized to obtain the first individual parameter of the target patient.

[0085] In the embodiment of the present invention, the combination of the severity of the disease and the degree of individual difference can be achieved by calculating the sum or product of the two, which is not limited here.

[0086] In one embodiment of the present invention, the normalization processing can be specifically, for example, maximum and minimum value normalization processing, and the normalization in subsequent steps can all adopt maximum and minimum value normalization processing. In other embodiments of the present invention, other normalization methods can be selected according to the specific range of numerical values, which will not be described in detail.

[0087] As an example, in one embodiment of the present invention, the expression of the first physical parameter of the target patient may be specifically, for example, as follows:

[0088] ;

[0089] in, represents the first body parameter of the target patient; Indicates the severity of the target patient's illness at the current follow-up; Indicates the degree of individual differences among target patients; Represents the normalization function.

[0090] Since the symptoms of the target patient may improve or worsen in the long term, and may even cause complications, and the changes in the scores of the target patient in the same test items in multiple historical follow-ups over time can reflect whether the chronic airway disease of the target patient has improved or worsened, the scores of all questions in the same test items of the target patient in each historical follow-up and the time series of the historical follow-up can be analyzed. At the same time, combined with the differences in the scores of the questions in the same test items between each historical follow-up and the current follow-up, the possibility of the deterioration of the symptoms of the target patient and complications in the long term can be reflected through the second individual parameter obtained. In the subsequent cluster analysis process, the second individual parameter and the first individual parameter obtained above can be combined to perform weighted adjustment on the discrimination analysis of the detection index to improve the calculation accuracy of the discrimination of the detection index.

[0091] Preferably, in one embodiment of the present invention, the method for acquiring the second individual parameter of the target patient specifically includes:

[0092] First, based on the calculation method of the total score of each test item of the target patient at the current follow-up, the total score of each test item of the target patient at each historical follow-up is obtained. The specific calculation process is: the sum of the scores of all questions in each test item of the target patient at each historical follow-up is used as the total score of each test item of the target patient at each historical follow-up.

[0093] Then, a non-zero natural number is used to label each historical follow-up of the target patient in chronological order to obtain the serial number of each historical follow-up of the target patient. For example, the serial number of the first historical follow-up is 1, and the serial number of the second historical follow-up is 2. The two-dimensional sequence composed of the serial number of each historical follow-up of the target patient and the total score of the target test item of the target patient at each historical follow-up is used as the symptom manifestation sequence of the target patient with respect to the target test item at each historical follow-up. For example, the total score of the target test item at the first historical follow-up is 20, then the symptom manifestation sequence of the target patient with respect to the target test item at the first historical follow-up is .

[0094] Furthermore, based on the calculation method of the patient characteristic difference degree of the target test items between the target patient and other patients to be tested at the current follow-up, according to the difference in the scores of the same questions in the target test items of the target patient between each historical follow-up and the current follow-up, the patient characteristic difference parameters of the target test items of the target patient between each historical follow-up and the current follow-up are obtained. The specific calculation process is: the cumulative value of the absolute value of the difference in the scores of all the same questions in the target test items of the target patient between each historical follow-up and the current follow-up is used as the patient characteristic difference parameter of the target test items of the target patient between each historical follow-up and the current follow-up. The larger the patient characteristic difference parameter is, the greater the difference in the symptom manifestation of the target patient between each historical follow-up and the current follow-up in the target test item, and the greater the reference value in the subsequent principal component direction analysis. Therefore, the patient characteristic difference parameter can be normalized to obtain the weight parameter of the target patient regarding the target test item at each historical follow-up. The accuracy of the principal component direction analysis can be improved based on the weight parameter subsequently.

[0095] Finally, the symptom manifestation sequence and weight parameters of the target patient for the target test items in all historical follow-ups are input into the weighted PCA algorithm to obtain the principal component direction of the target patient for the target test items. The principal component direction of the target patient for each test item can be obtained by the same method as above, and then the average value of the principal component directions of the target patient for all test items can be used as the overall principal component direction of the target patient. The larger the slope of the overall principal component direction, the more severe the symptoms of the target patient become with multiple follow-ups and may be accompanied by the emergence of secondary complications. Therefore, the slope of the overall principal component direction of the target patient can be normalized to obtain the second individual parameter of the target patient. The weighted PCA algorithm is a technical means well known to those skilled in the art and will not be elaborated here. The slope of the overall principal component direction can be determined by the angle between the overall principal component direction and the standard coordinate axis, which is a technical means well known to those skilled in the art and will not be elaborated here.

[0096] As an example, in one embodiment of the present invention, the expression of the second individual parameter of the target patient may be specifically, for example, as follows:

[0097] ;

[0098] in, A second individual parameter representing the target patient; The slope representing the direction of the overall principal component of the target patient; Represents the normalization function.

[0099] At this point, the first body parameter and the second body parameter of the target patient are obtained, and the first body parameter and the second body parameter of each patient can be obtained by the same method as above.

[0100] The cluster analysis module 103 is used to take any detection indicator as the target detection indicator, cluster all patients according to the difference in indicator data of the target detection indicator between different patients, and obtain multiple cluster clusters about the target detection indicator; according to the distribution of indicator data of the target detection indicator of each patient of each patient type in each cluster cluster, the first individual parameter and the second individual parameter of each patient of each patient type, and the difference in the number of patients of the same patient type between different cluster clusters, the patient discrimination of the target detection indicator is obtained.

[0101] Since different detection indicators have different degrees of discrimination for different types of chronic airway diseases, some detection indicators are universal in various types of chronic airway diseases. These detection indicators have low discrimination for different types of chronic airway diseases. In the subsequent patient clustering process, the reference value of these detection indicators should be reduced to avoid poor clustering effect. Therefore, the embodiment of the present invention first analyzes any detection indicator, takes any detection indicator as the target detection indicator, and then clusters all patients according to the differences in indicator data of the target detection indicators between different patients to obtain multiple clustering clusters about the target detection indicators. Subsequently, based on the number of patients of the same patient type in different clustering clusters and the indicator data of the target detection indicators of each patient, the patient discrimination of the target detection indicator can be accurately calculated and analyzed.

[0102] Preferably, in one embodiment of the present invention, the method for obtaining multiple clusters of target detection indicators specifically includes:

[0103] The absolute value of the difference in index data of the target detection index between any two patients is used as the index distance metric for the target detection index between any two patients, and then the K-means clustering algorithm is used to cluster all patients based on the index distance metric for the target detection index between any two patients to obtain multiple clustering clusters for the target detection index, wherein the number of clustering clusters can be determined using the existing elbow method. In other embodiments of the present invention, other clustering algorithms based on distance metrics may also be used for clustering, which is not limited here.

[0104] Since the number of patients analyzed in the embodiment of the present invention is large, after the above-mentioned clustering operation, each cluster contains patients of various patient types, and there are multiple patients of the same patient type in the cluster. At the same time, the more concentrated the distribution of the indicator data of the target detection indicator of the patients of the same patient type in each cluster, and the greater the difference in the number of patients of the same patient type between different clusters, it means that the target detection indicator has a greater degree of discrimination for different patient types. Therefore, the distribution of the indicator data of the target detection indicator of each patient of each patient type in each cluster, and the difference in the number of patients of the same patient type between different clusters can be analyzed, and the first individual parameter and the second individual parameter of each patient of each patient type in the cluster are combined. The discrimination of the target detection indicator for different patient types is reflected by the obtained patient discrimination. Subsequently, based on the patient discrimination, the reference value of each detection indicator in the patient clustering can be adjusted to improve the accuracy of patient clustering.

[0105] Preferably, in one embodiment of the present invention, the method for obtaining the patient discrimination degree of the target detection index specifically includes:

[0106] See also Figure 2 , which shows a flow chart of a method for obtaining patient discrimination of target detection indicators provided by an embodiment of the present invention.

[0107] Step S301: Take any cluster cluster regarding the target detection index as the target cluster cluster, and take the average value of the index data of the target detection index of all patients in the target cluster cluster as the cluster center value of the target cluster cluster; take the average value of the index data of the target detection index of all patients of each patient type in the target cluster cluster as the category center value of each patient type in the target cluster cluster.

[0108] Subsequently, the difference between the category center value of each patient type in the target cluster and the cluster center value of the target cluster can be analyzed to calculate the first discrimination of the target detection index in the target cluster.

[0109] Step S302: Obtain a first discrimination degree of the target detection index in the target cluster according to the distribution of the index data of the target detection index of each patient of each patient type in the target cluster, the first individual parameter and the second individual parameter of each patient of each patient type, and the difference between the category center value of each patient type in the target cluster and the cluster center value of the target cluster.

[0110] The more concentrated the distribution of the indicator data of the target detection index of each patient of each patient type in the target clustering cluster, and the greater the difference between the category center value and the cluster center value of each patient type, the greater the discrimination of the target detection index for different patient types in the target clustering cluster. Therefore, the distribution of the indicator data of the target detection index of each patient of each patient type in the target clustering cluster, and the difference between the category center value of each patient type in the target clustering cluster and the cluster center value of the target clustering cluster can be analyzed. At the same time, the larger the first individual parameter and the second individual parameter of each patient of each patient type in the target clustering cluster, the greater the individual difference in the patient's symptom manifestation, and the more severe the patient's symptom manifestation, then the greater the weight when analyzing the distribution of the indicator data of the target detection index of each patient of each patient type in the target clustering cluster. Therefore, the first individual parameter and the second individual parameter of the patient are added in the analysis to improve the calculation accuracy of the first discrimination.

[0111] Preferably, in one embodiment of the present invention, the method for obtaining the first discrimination degree of the target detection index in the target cluster specifically includes:

[0112] Firstly, the first individual parameter and the second individual parameter of each patient in the target cluster are integrated and normalized to obtain the weight coefficient of each patient in the target cluster.

[0113] In the embodiment of the present invention, the integration of the first individual parameter and the second individual parameter of each patient in the target cluster can be achieved by calculating the sum or product of the first individual parameter and the second individual parameter, which is not limited here.

[0114] As an example, in one embodiment of the present invention, the expression of the weight coefficient of each patient in the target cluster may be specifically, for example, as follows:

[0115] ;

[0116] in, Indicates the first Weight coefficient for each patient; Indicates the first The first individual parameter of each patient; Indicates the first The second individual parameter of each patient; Represents the normalization function.

[0117] Then, based on the calculation formula of the first discrimination, the first discrimination of the target detection index in the target cluster is obtained. The calculation formula of the first discrimination is:

[0118] ;

[0119] in, Indicates the first discrimination of the target detection index in the target cluster; Indicates the first The category center value of each patient type; Indicates the cluster center value of the target cluster; Indicates the first

[0120] Patient type The target detection index data of each patient; Indicates the first Patient type Weight coefficient for each patient; Indicates the first the number of patients with each patient type; Indicates the number of patient types in the target cluster; represents the normalization function; Represents a tuning parameter, used to prevent the denominator from being 0. The value range is In one embodiment of the present invention, Set to 0.01, The specific value of can also be set by the implementer according to the specific implementation scenario and is not limited here.

[0121] in, It is the first cluster in the target cluster. In the process of calculating the standard deviation of the target detection index data of patients of different patient types, the patient's weight coefficient is added , which improves the accuracy of its distribution analysis. The smaller the value, the more The more concentrated the distribution of the target detection index data of the patient type, the greater the discrimination degree of the target detection index for the patient type in the target cluster. The bigger it is, The larger the value is, the more The greater the difference between the overall level of the target detection index data of the patient of the target cluster and the overall level of the target detection index data of the patient in the target cluster, the greater the discrimination degree of the target detection index for the patient type in the target cluster, and the higher the first discrimination degree. The bigger it is.

[0122] Step S303: obtaining a second discrimination degree of the target detection index in the target cluster according to the difference in the number of patients of the same patient type between the target cluster and other clusters except the target cluster.

[0123] The greater the difference in the number of patients of the same patient type between the target cluster and other clusters, the greater the discrimination of the target detection index for the patient type in the target cluster. Therefore, the difference in the number of patients of the same patient type between the target cluster and other clusters except the target cluster can be analyzed to obtain the second discrimination of the target detection index in the target cluster.

[0124] Preferably, in one embodiment of the present invention, the method for obtaining the second discrimination degree of the target detection index in the target cluster specifically includes:

[0125] First, the number of patients of each patient type in each cluster is taken as the numerator, the number of patients in each cluster is taken as the denominator, and the ratio is taken as the proportion of the number of each patient type in each cluster.

[0126] Then, based on the calculation formula of the second discrimination, the second discrimination of the target detection index in the target cluster is obtained. The calculation formula of the second discrimination is:

[0127] ;

[0128] in, Indicates the second discrimination of the target detection index in the target cluster; Indicates the first The proportion of each patient type; Indicates the clusters other than the target cluster The other clusters The proportion of each patient type; Indicates the number of clusters other than the target cluster; Indicates the number of patient types in the target cluster; Represents the normalization function.

[0129] Step S304: The first discrimination and the second discrimination are integrated to obtain the comprehensive discrimination of the target detection indicator in the target cluster; the average value of the comprehensive discrimination of the target detection indicator in all clusters is used as the patient discrimination of the target detection indicator.

[0130] The greater the first discrimination and the second discrimination of the target detection index in the target cluster, the greater the discrimination of the target detection index for the patient type in the target cluster. Therefore, the first discrimination and the second discrimination can be combined to obtain the comprehensive discrimination of the target detection index in the target cluster. The same method as above can be used to obtain the comprehensive discrimination of the target detection index in each cluster, and then the average value of the comprehensive discrimination of the target detection index in all clusters is normalized to obtain the patient discrimination of the target detection index. The greater the patient discrimination, the greater the discrimination of the target detection index for chronic airway diseases of different patient types. Conversely, it means that the target detection indicator is universal or common in chronic airway diseases of different patient types.

[0131] In the embodiment of the present invention, the sum or product of the first discrimination and the second discrimination may be used as the comprehensive discrimination of the target detection index in the target cluster to achieve the integration of the two, which is not limited here.

[0132] As an example, in one embodiment of the present invention, the expression of the patient discrimination of the target detection index may be specifically, for example, as follows:

[0133] ;

[0134] ;

[0135] in, represents the patient discrimination of the target detection indicator; Indicates that the target detection index is The comprehensive discrimination among the clusters; Indicates the number of clusters; Represents the comprehensive discrimination of target detection indicators in target clusters; Indicates the first discrimination of the target detection index in the target cluster; Indicates the second discrimination of the target detection index in the target cluster; Represents the normalization function.

[0136] The patient differentiation of each detection indicator can be obtained by the same method as above.

[0137] The clustering adjustment module 104 is used to adjust the difference in indicator data of the same detection indicator between patients according to the patient discrimination of each detection indicator to obtain an adjusted distance measurement between patients; based on the adjusted distance measurement between patients, all patients are clustered to obtain multiple optimized clustering clusters.

[0138] In the follow-up system of patients with chronic airway diseases, it is usually necessary to cluster patients based on the indicator data of various detection indicators of patients, so as to achieve targeted and personalized management of patients with similar disease manifestations in the same cluster. However, in the existing K-means clustering algorithm, each detection indicator is regarded as equal in the clustering process, that is, the existing clustering algorithm is indifferent clustering, and each detection indicator of the patient has different discrimination for different patient types. There are some detection indicators with high discrimination for patient types, which are more targeted for chronic airway diseases of different patient types, while other detection indicators have low discrimination for patient types, which are universal for chronic airway diseases of different patient types. Therefore, in order to improve the effect of patient clustering, it is necessary to reduce the reference of detection indicators with low discrimination in the clustering process, so as to avoid the dilution effect of detection indicators with low discrimination on detection indicators with high discrimination in the clustering process. Therefore, according to the patient discrimination of each detection indicator, the difference in indicator data of the same detection indicator between patients can be adjusted to obtain the adjusted distance measurement between patients, and then the patients can be accurately clustered based on the adjusted distance measurement between patients.

[0139] Preferably, in one embodiment of the present invention, the method for obtaining the adjusted distance metric between patients specifically includes:

[0140] Based on the calculation formula of the adjusted distance metric, the adjusted distance metric between patients is obtained. The calculation formula of the adjusted distance metric is:

[0141] ;

[0142] in, Indicates Patients and The adjusted distance measure between patients, ; Indicates The patient's The indicator data of each detection indicator; Indicates The patient's The indicator data of each detection indicator; Indicates The patient discrimination of each test indicator; Indicates the number of detection indicators.

[0143] in, In the process of calculating the Euclidean distance of each test indicator between patients, the patient discrimination of each test indicator is added Adjust the patient discrimination of the test indicators The smaller the value, the more universal the test indicator is in different types of chronic airway diseases. In the analysis of distance measurement, the reference value of the test indicator is low. On the contrary, the patient discrimination of the test indicator is The larger it is, the more reference the detection indicator has in the analysis of clustering distance measurement.

[0144] After obtaining the adjusted distance metric between any two patients, all patients can be clustered based on the adjusted distance metric between the patients to obtain multiple optimized clustering clusters, thereby improving the final clustering effect.

[0145] Preferably, in one embodiment of the present invention, the method for obtaining multiple optimized clusters specifically includes:

[0146] Using the K-means clustering algorithm, all patients are clustered based on the adjusted distance metric between patients to obtain different optimized clustering clusters, wherein the number of optimized clustering clusters can be determined using the existing elbow method. In other embodiments of the present invention, other clustering algorithms based on distance metrics may also be used for clustering, which is not limited here.

[0147] Through the above clustering method, patients with similar chronic airway disease manifestations can be accurately divided into the same cluster, which is conducive to the long-term intelligent follow-up system to carry out targeted and personalized management of patients.

[0148] It should be noted that the sequence of the above embodiments of the present invention is only for description and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0149] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.

Claims

1. A long-term intelligent follow-up system for patients with chronic airway diseases, characterized in that: The system comprises: The data collection module is used to obtain the scores of each question in different test items of patients of different types of patients in the current follow-up and multiple historical follow-ups within a preset time period, and to obtain the index data of multiple test indicators of each patient in the current follow-up; The individual analysis module is used to take any patient as a target patient, and obtain a first individual parameter of the target patient according to the scores of all questions in each test item of the target patient at the current follow-up, and the difference in scores of the target patient and other patients except the target patient in the same test item at the current follow-up; obtain a second individual parameter of the target patient according to the scores of all questions in the same test item of the target patient at each historical follow-up, the time sequence of the historical follow-ups, and the difference in scores of questions in the same test item between each historical follow-up and the current follow-up; A cluster analysis module, for taking any detection index as a target detection index, clustering all patients according to the difference of the index data of the target detection index between different patients, and obtaining a plurality of clustering clusters about the target detection index; obtaining the patient discrimination of the target detection index according to the distribution of the index data of the target detection index of each patient of each patient type in each clustering cluster, the first individual parameter and the second individual parameter of each patient of each patient type, and the difference in the number of patients of the same patient type between different clustering clusters; The clustering adjustment module is used to adjust the difference in indicator data of the same detection indicator between patients according to the patient discrimination of each detection indicator to obtain an adjusted distance metric between patients; based on the adjusted distance metric between patients, all patients are clustered to obtain multiple optimized clustering clusters.

2. A long-term intelligent follow-up system for patients with chronic airway diseases according to claim 1, characterized in that: The obtaining of the first physical parameter of the target patient comprises: The sum of the scores of all questions in each test item of the target patient at the current follow-up is used as the total score of each test item of the target patient at the current follow-up; the average of the total scores of all test items of the target patient at the current follow-up is used as the disease severity of the target patient at the current follow-up; Obtaining the degree of individual difference of the target patient at the current follow-up according to the difference in scores of the same questions in the same test items between the target patient and other patients at the current follow-up, and the difference in the total scores of the same test items between the target patient and other patients at the current follow-up; The severity of the disease and the individual difference are integrated and normalized to obtain the first individual parameter of the target patient.

3. A long-term intelligent follow-up system for patients with chronic airway diseases according to claim 2, characterized in that: The degree of individual differences in the target patients during the current follow-up includes: Any other patient except the target patient is used as the other patient to be tested, and any test item is used as the target test item; The cumulative value of the absolute value of the difference between the scores of all the same questions in the target test items between the target patient and other patients to be tested at the current follow-up is used as the patient characteristic difference degree of the target test items between the target patient and other patients to be tested at the current follow-up; Performing negative correlation normalization processing on the absolute value of the difference in the total score of the target test item between the target patient and other patients to be tested at the current follow-up, to obtain the reference weight of the target test item between the target patient and other patients to be tested at the current follow-up; Using the reference weight of each test item between the target patient and other patients to be tested at the current follow-up, weighted summing the patient characteristic difference degree of each test item between the target patient and other patients to be tested at the current follow-up is performed to obtain the initial difference degree between the target patient and other patients to be tested; The cumulative value of the initial difference degrees between the target patient and all other patients is taken as the individual difference degree of the target patient at the current follow-up.

4. A long-term intelligent follow-up system for patients with chronic airway diseases according to claim 3, characterized in that: The obtaining of the second individual parameter of the target patient comprises: Based on the calculation method of the total score of each test item of the target patient at the current follow-up, the total score of each test item of the target patient at each historical follow-up is obtained; Using non-zero natural numbers, numbering each historical follow-up of the target patient in time sequence to obtain the serial number of each historical follow-up of the target patient; using the serial number of each historical follow-up of the target patient and the total score of the target test item of the target patient at each historical follow-up as the symptom manifestation sequence of the target patient with respect to the target test item at each historical follow-up; Based on the calculation method of the patient characteristic difference degree of the target test item between the target patient and other patients to be tested at the current follow-up, according to the difference in the score of the target patient in the same question in the target test item between each historical follow-up and the current follow-up, the patient characteristic difference parameter of the target test item between each historical follow-up and the current follow-up is obtained, and the patient characteristic difference parameter is normalized to obtain the weight parameter of the target patient with respect to the target test item at each historical follow-up; Input the symptom manifestation sequence and the weight parameter of the target patient in all historical follow-ups with respect to the target test items into the weighted PCA algorithm to obtain the principal component direction of the target patient with respect to the target test items; take the average of the principal component directions of the target patient with respect to all test items as the overall principal component direction of the target patient; The slope of the overall principal component direction of the target patient is normalized to obtain a second individual parameter of the target patient.

5. The long-term intelligent follow-up system for patients with chronic airway diseases according to claim 1, characterized in that: The step of obtaining a plurality of clusters related to target detection indicators includes: The absolute value of the difference between the target detection index data of any two patients is used as the index distance measurement between any two patients with respect to the target detection index; Using the K-means clustering algorithm, all patients are clustered based on the indicator distance metric regarding the target detection indicator between any two patients to obtain multiple clusters regarding the target detection indicator.

6. The long-term intelligent follow-up system for patients with chronic airway diseases according to claim 1, characterized in that: The patient discrimination of the target detection index includes: Taking any clustering cluster about the target detection index as the target clustering cluster, taking the average value of the index data of the target detection index of all patients in the target clustering cluster as the cluster center value of the target clustering cluster; taking the average value of the index data of the target detection index of all patients of each patient type in the target clustering cluster as the category center value of each patient type in the target clustering cluster; Obtaining a first discrimination degree of the target detection indicator in the target cluster according to the distribution of the indicator data of the target detection indicator of each patient of each patient type in the target cluster, the first individual parameter and the second individual parameter of each patient of each patient type, and the difference between the category center value of each patient type in the target cluster and the cluster center value of the target cluster; According to the difference in the number of patients of the same patient type between the target cluster and other clusters except the target cluster, a second discrimination degree of the target detection index in the target cluster is obtained; Combining the first discrimination and the second discrimination to obtain a comprehensive discrimination of the target detection index in the target cluster; The average value of the comprehensive discrimination of the target detection index in all clusters is normalized to obtain the patient discrimination of the target detection index.

7. A long-term intelligent follow-up system for patients with chronic airway diseases according to claim 6, characterized in that: The obtaining of the first discrimination degree of the target detection index in the target cluster includes: The first individual parameter and the second individual parameter of each patient in the target cluster are integrated and normalized to obtain a weight coefficient of each patient in the target cluster; Based on the calculation formula of the first discrimination, the first discrimination of the target detection index in the target cluster is obtained. The calculation formula of the first discrimination is: ； in, Indicates the first discrimination of the target detection index in the target cluster; Indicates the first The category center value of each patient type; Indicates the cluster center value of the target cluster; Indicates the first Patient type The target detection index data of each patient; Indicates the first Patient type Weight coefficient for each patient; Indicates the first the number of patients with each patient type; Indicates the number of patient types in the target cluster; represents the normalization function; Represents the adjustment parameter, the value range is .

8. The long-term intelligent follow-up system for patients with chronic airway diseases according to claim 6, characterized in that: The obtaining of the second discrimination degree of the target detection index in the target cluster includes: The number of patients of each patient type in each cluster is taken as the numerator, the number of patients in each cluster is taken as the denominator, and the ratio is taken as the proportion of the number of each patient type in each cluster; Based on the calculation formula of the second discrimination, the second discrimination of the target detection index in the target cluster is obtained. The calculation formula of the second discrimination is: ； in, Indicates the second discrimination of the target detection index in the target cluster; Indicates the first The proportion of each patient type; Indicates the clusters other than the target cluster The other clusters The proportion of each patient type; Indicates the number of clusters other than the target cluster; Indicates the number of patient types in the target cluster; Represents the normalization function.

9. The long-term intelligent follow-up system for patients with chronic airway diseases according to claim 1, characterized in that: The obtaining of the adjusted distance metric between patients comprises: Based on the calculation formula of the adjusted distance metric, the adjusted distance metric between patients is obtained, and the calculation formula of the adjusted distance metric is: ； in, Indicates Patients and The adjusted distance measure between patients, ; Indicates The patient's The indicator data of each detection indicator; Indicates The patient's The indicator data of each detection indicator; Indicates The patient discrimination of each test indicator; Indicates the number of detection indicators.

10. The long-term intelligent follow-up system for patients with chronic airway diseases according to claim 1, characterized in that: The step of clustering all patients based on the adjusted distance metric between patients to obtain a plurality of optimized clustering clusters comprises: Using the K-means clustering algorithm, all patients are clustered based on the adjusted distance metric between patients to obtain multiple optimized clustering clusters.

Images