Intelligent analysis system for kidney disease personalized nursing scheme optimization

Abstract

The invention relates to the technical field of medical care, in particular to an intelligent analysis system for kidney disease personalized care scheme optimization. The system comprises a kidney multivariate data fusion module, a disease key factor analysis module, a personalized nursing scheme generation module and a nursing scheme dynamic optimization module, and can acquire a kidney disease multivariate data set in real time and perform multivariate data fusion and kidney association mining analysis; the disease association degree among the sub-metadata in each type of kidney patients is obtained; performing disease key factor screening based on the disease association degree among the sub-metadata in each type of kidney patients to obtain disease feature key factors corresponding to each type of kidney patients; successful nursing scheme cases corresponding to the same kind of patients are obtained, personalized nursing scheme generation processing and nursing effect feedback evaluation are carried out, dynamic optimization is carried out at the same time, and personalized nursing feedback optimization schemes corresponding to the kidney patients are generated. The invention can provide a scientific and effective nursing scheme.

Description

Technical Field

[0001] The present invention relates to the technical field of medical care, and particularly to an intelligent analysis system for optimizing personalized nursing plans for kidney diseases. Background Art

[0002] In recent years, with the rapid development of artificial intelligence, big data, and machine learning technologies, intelligent analysis methods have gradually been applied to the medical field, especially in aspects such as optimizing nursing plans. Remarkable results have been achieved. By analyzing a large amount of patient data, intelligent analysis methods can deeply explore individual differences, identify potential risk factors, predict the trend of disease progression, and provide personalized nursing suggestions according to the specific situation of patients. This data-driven method has high accuracy and real-time performance, and can significantly improve the pertinence of nursing plans and nursing effects. Existing nursing plans for kidney diseases mostly rely on doctors' experience judgments, combined with conventional examination data, such as blood indicators (such as serum creatinine, urinary protein, etc.), imaging examination results, etc., for basic nursing adjustments. However, this method often ignores individual differences among patients, such as factors like age, gender, living habits, genetic background, etc., and also lacks dynamic, comprehensive monitoring and real-time nursing adjustments. Summary of the Invention

[0003] Based on this, it is necessary for the present invention to provide an intelligent analysis system for optimizing personalized nursing plans for kidney diseases to solve at least one of the above technical problems.

[0004] To achieve the above object, an intelligent analysis system for optimizing personalized nursing plans for kidney diseases includes the following modules:

[0005] A kidney multi-source data fusion module, which is used to collect a multi-source dataset of kidney diseases in real time and perform multi-source data fusion on the multi-source dataset of kidney diseases to obtain a comprehensive dataset of kidney disease conditions;

[0006] A disease key factor analysis module, which is used to perform kidney association mining analysis between each sub-metadata corresponding to each type of kidney patient in the comprehensive dataset of kidney disease conditions to obtain the disease association degree between each sub-metadata within each type of kidney patient; based on the disease association degree between each sub-metadata within each type of kidney patient, perform disease key factor screening on each sub-metadata corresponding to each type of kidney patient to obtain the disease characteristic key factors corresponding to each type of kidney patient;

[0007] A personalized nursing plan generation module, which is used to obtain successful nursing plan cases corresponding to the same type of kidney disease patients, and based on the successful nursing plan cases corresponding to the same type of kidney disease patients and combined with a convolutional neural network, perform personalized nursing plan generation processing on the disease characteristic key factors corresponding to each type of kidney patient to generate a disease personalized nursing plan corresponding to each type of kidney patient;

[0008] The nursing plan dynamic optimization module is used to evaluate the nursing effect feedback of the corresponding renal function indexes of each type of kidney patients during the nursing process based on the disease personalized nursing plan corresponding to each type of kidney patients, so as to obtain the feedback effect of the change of the renal function indexes corresponding to each type of kidney patients; and dynamically optimize the corresponding disease personalized nursing plan based on the feedback effect of the change of the renal function indexes corresponding to each type of kidney patients, so as to generate the personalized nursing feedback optimization plan corresponding to each type of kidney patients.

[0009] Furthermore, the kidney multi-data fusion module includes the following functions:

[0010] By using clinical examination equipment to regularly collect the renal function indexes corresponding to kidney disease patients, including serum creatinine, blood urea nitrogen and glomerular filtration rate, the urine routine indexes including urinary protein, red blood cells and white blood cells, and the related biochemical indexes including blood sugar, blood lipid and electrolytes, so as to obtain clinical examination index data;

[0011] By using gene detection technology to obtain the gene data corresponding to kidney disease patients related to kidney diseases, so as to obtain the gene data of kidney patients;

[0012] By using wearable devices including smart bracelets and smart sphygmomanometers to continuously monitor the heart rate, blood pressure and sleep quality of kidney disease patients, so as to obtain the physiological condition data of kidney patients;

[0013] By means of questionnaire surveys, collect the lifestyle data corresponding to kidney disease patients, including diet preferences, exercise frequency and smoking and drinking status, and the psychological state data including anxiety and depression levels, so as to obtain the living condition data of kidney patients;

[0014] Merge the clinical examination index data, the gene data of kidney patients, the physiological condition data of kidney patients and the living condition data of kidney patients into a multi-data set of kidney diseases, and perform denoising, error correction and missing value processing on the multi-data set of kidney diseases, so as to remove the corresponding noise interference in the multi-data set of kidney diseases, correct the corresponding error values, and use linear interpolation to fill in the corresponding missing values, and at the same time perform normalization processing on the data corresponding to different types and different magnitudes, so as to unify the multi-data set of kidney diseases into the same numerical range, and obtain a standardized multi-data set of kidneys;

[0015] Perform multi-data fusion on the standardized multi-data set of kidneys to obtain a comprehensive data set of kidney disease conditions.

[0016] Furthermore, the disease key factor analysis module includes the following functions:

[0017] Classify and subdivide each sub - metadata corresponding to each kidney patient in the comprehensive dataset of kidney disease conditions according to the clinical manifestations and pathological conditions of the kidney disease patients, so as to obtain each sub - metadata corresponding to each type of kidney patient;

[0018] Perform kidney - related mining analysis among each sub - metadata corresponding to each type of kidney patient to obtain the disease correlation degree among the sub - metadata within each type of kidney patient;

[0019] Based on the disease correlation degree among the sub - metadata within each type of kidney patient, construct an association network among each sub - metadata corresponding to each type of kidney patient, taking each sub - metadata as a network node, and compare and judge the corresponding disease correlation degree according to a preset association threshold. If the corresponding disease correlation degree is greater than the preset association threshold, establish a connection edge between the corresponding two network nodes, otherwise do not establish, and generate a sub - metadata association network corresponding to each type of kidney patient;

[0020] Based on the sub - metadata association network corresponding to each type of kidney patient, screen the disease key factors for each sub - metadata corresponding to each type of kidney patient to obtain the disease characteristic key factors corresponding to each type of kidney patient.

[0021] Further, the kidney - related mining analysis among each sub - metadata corresponding to each type of kidney patient includes:

[0022] Extract and quantify the kidney characteristics of each sub - metadata corresponding to each type of kidney patient. For numerical sub - metadata, statistically extract its corresponding mean, standard deviation, maximum and minimum values, and amplitude. For text - type sub - metadata, use natural language processing technology to extract the corresponding keywords and theme features and convert them into numerical values that can be used for mathematical calculations. At the same time, uniformly statistically calculate the corresponding mean, standard deviation, maximum and minimum values, and amplitude for the same numerical type, so as to obtain the characteristic set of each sub - metadata corresponding to each type of kidney patient;

[0023] Form a feature vector from the characteristic sets of each sub - metadata corresponding to each type of kidney patient to generate the feature vectors of each sub - metadata corresponding to each type of kidney patient;

[0024] Quantify the disease - related correlation degree among the feature vectors of each sub - metadata corresponding to each type of kidney patient to obtain the disease correlation degree among the sub - metadata within each type of kidney patient.

[0025] Further, the screening of disease key factors for each sub - metadata corresponding to each type of kidney patient based on the sub - metadata association network corresponding to each type of kidney patient includes:

[0026] Obtain the degree centrality coefficients corresponding to each network node within each type of kidney patient through the sub-metadata association network corresponding to each type of kidney patient;

[0027] Based on the degree centrality coefficients corresponding to each network node within each type of kidney patient, screen the association hub nodes for each network node within the sub-metadata association network corresponding to each type of kidney patient. If the corresponding degree centrality coefficient is greater than or equal to the preset threshold, then screen out the sub-metadata features of the network node corresponding to the sub-metadata association network as the hub factors corresponding to a relatively high degree of association with other sub-metadata features, so as to obtain the disease sub-metadata hub factors corresponding to each type of kidney patient;

[0028] Calculate the valueability of the disease sub-metadata hub factors corresponding to each type of kidney patient screened out, so as to obtain the valueability corresponding to each sub-metadata factor within each type of kidney patient;

[0029] Based on the valueability corresponding to each sub-metadata factor within each type of kidney patient, screen the disease key factors for the corresponding disease sub-metadata hub factors. If the corresponding valueability is greater than 0.5, then screen out the corresponding disease sub-metadata hub factors as the key feature factors, otherwise eliminate them, so as to obtain the disease feature key factors corresponding to each type of kidney patient.

[0030] Further, the calculating the valueability of the disease sub-metadata hub factors corresponding to each type of kidney patient screened out includes:

[0031] Conduct eigenvalue statistical analysis on the disease sub-metadata hub factors corresponding to each type of kidney patient, so as to calculate the mean, standard deviation, maximum value and minimum value statistics corresponding to each sub-metadata factor, and form the corresponding feature statistical matrix, so as to generate the disease sub-metadata feature statistical matrix corresponding to each type of kidney patient;

[0032] Based on the disease sub-metadata feature statistical matrix corresponding to each type of kidney patient, calculate the factor variability score for the corresponding disease sub-metadata hub factors, and obtain the disease sub-metadata hub factor variability score corresponding to each type of kidney patient;

[0033] Based on the disease sub-metadata hub factor variability score corresponding to each type of kidney patient and in combination with the disease sub-metadata feature statistical matrix, conduct valueability weighted calculation on the corresponding disease sub-metadata hub factors, so as to obtain the valueability corresponding to each sub-metadata factor within each type of kidney patient.

[0034] Further, the personalized nursing plan generation module includes the following functions:

[0035] Obtain the successful nursing plan cases corresponding to the same type of kidney disease patients;

[0036] Analyze the disease characteristics of the successful nursing plans for patients with the same type of kidney disease to obtain the disease characteristics of the successful nursing plans for patients of the same type.

[0037] By combining a convolutional neural network, construct a corresponding personalized nursing plan generation model, and input the disease characteristics of the successful nursing plans for patients of the same type and the key factors of the disease characteristics for each type of kidney patient into the personalized nursing plan generation model for personalized nursing plan generation processing, so as to generate a disease-specific personalized nursing plan for each type of kidney patient.

[0038] Furthermore, the personalized nursing plan generation processing specifically analyzes the feature matching score between the disease characteristics of the successful nursing plan and the key factors of the disease characteristics for each type of kidney patient within the personalized nursing plan generation model, and makes a judgment according to a preset matching threshold. If the feature matching score is greater than or equal to the preset matching threshold, then apply the successful nursing plan for this type of patient; if the feature matching score is less than the preset matching threshold, then use case comparison analysis to identify and adjust the corresponding diet content, exercise intensity and frequency, and the dosage and time of drug use in the successful nursing plan, so as to generate a disease-specific personalized nursing plan for each type of kidney patient.

[0039] Furthermore, the nursing plan dynamic optimization module includes the following functions:

[0040] Apply the disease-specific personalized nursing plan for each type of kidney patient to the corresponding kidney disease patients to generate the nursing process for this type of kidney patient.

[0041] Analyze the renal function indicators during the nursing process of this type of kidney patient to obtain the corresponding renal function indicators during the nursing process of this type of kidney patient.

[0042] Obtain the expected target of the renal function indicators for this type of kidney patient, and based on the expected target of the renal function indicators and using time series analysis, conduct a nursing effect feedback evaluation between the corresponding renal function indicators during the nursing process of this type of kidney patient to obtain the feedback effect of the change in the renal function indicators for this type of kidney patient.

[0043] Dynamically optimize the corresponding disease-specific personalized nursing plan based on the feedback effect of the change in the renal function indicators for this type of kidney patient to generate a personalized nursing feedback optimization plan for this type of kidney patient.

[0044] Furthermore, the personalized nursing feedback optimization plan specifically analyzes the reasons if the improvement of the renal function indicators is not obvious according to the corresponding feedback effect of the change in the renal function indicators, and adjusts the corresponding diet, exercise, and drug-taking management suggestions in the disease-specific personalized nursing plan based on the reasons.

[0045] Advantages of the present invention:

[0046] The intelligent analysis system for optimizing the personalized care plan for kidney diseases proposed by the present invention is generally composed of a kidney multi-source data fusion module, a disease key factor analysis module, a personalized care plan generation module, and a care plan dynamic optimization module. Compared with the prior art, the beneficial effects of this application are as follows: by collecting multi-source data of kidney-related patients in real time, combining clinical test data, patient gene data, patient physiological conditions, patient living habits, etc., a comprehensive and accurate comprehensive dataset of kidney health status can be constructed. This dataset can reflect the multi-dimensional health information of patients, avoiding the deviation and incompleteness brought by a single data source. The data fusion process can not only integrate different types of data, but also remove noise data through algorithms during the analysis process, thereby improving the overall data quality and accuracy. In addition, data fusion helps to identify potential influencing factors and interactions, providing more valuable references for subsequent analysis. Through this comprehensive dataset, data support can be provided for further kidney health status analysis, personalized formulation of care plans, and patient effect evaluation. Secondly, based on the comprehensive dataset, through multi-dimensional kidney-related mining analysis of different categories of patients, the internal connections between various types of patients can be revealed. There is a certain degree of correlation between each sub-metadata. By mining these relationships, it can not only help to better understand the potential mechanisms of kidney health, but also further provide a basis for the formulation of personalized care plans. Through the analysis of disease correlation, key factors that have a greater impact on patient health can be identified, such as specific living habits, laboratory indicators, or certain behavior patterns, etc. These key factors will play an important role in the subsequent formulation of care plans, making the nursing work more precise and efficient. In addition, kidney patients face different risk factors under different pathological states. Through this multi-dimensional analysis, it helps to optimize the nursing intervention strategy according to the individual needs of each patient. Then, by obtaining successful nursing cases of patients with the same type of kidney disease, valuable experience and data support are provided for the formulation of personalized care plans. By combining deep learning technologies such as convolutional neural networks (CNNs), the most suitable care plan can be mined according to the key factors of the patient's disease characteristics. The convolutional neural network can automatically identify effective patterns and rules in the nursing process through learning a large number of historical cases, so as to customize personalized nursing plans for each type of kidney patient. This process can not only improve the accuracy and scientific nature of nursing, but also reduce the error of manual intervention by automatically generating care plans, improving the efficiency and operability of care plans. Through the intelligent generation based on patient data and historical successful nursing plans, the nursing team can respond more quickly to the needs of patients, thereby improving the quality of patient care.Finally, by regularly monitoring the renal function indicators during the care of kidney patients, the care effect can be evaluated in real time and feedback can be provided based on the changes of the patients. This feedback mechanism is the key to realizing the dynamic optimization of personalized care plans. During the care process, the changes in renal function indicators can help determine whether the care plan has achieved the expected effect and timely identify whether there are areas where the care plan needs to be adjusted. Based on this feedback data, the care plan can be optimized in real time through algorithms, which can continuously improve the adaptability and individualized characteristics of the care plan. During the long-term care process, this dynamic optimization mechanism can continuously track the health status of patients and adjust the care strategy according to the changing needs of each patient to ensure that patients always receive the most appropriate care services. Ultimately, this dynamic optimization mechanism can not only improve the care effect, but also provide more dynamic, comprehensive monitoring and real-time care adjustment. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Other features, objects, and advantages of the present invention will become more apparent from the following detailed description of non-limiting embodiments read in conjunction with the accompanying drawings:

[0048] Figure 1 It is a schematic diagram of the modules of the intelligent analysis system for optimizing the personalized care plan for kidney diseases of the present invention;

[0049] Figure 2 is Figure 1 a schematic diagram of the functional process of the kidney multi-data fusion module in;

[0050] Figure 3 is Figure 1 a schematic diagram of the functional process of the disease key factor analysis module in. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0051] The following clearly and completely describes the technical system of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all of them. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts fall within the scope of the present invention.

[0052] In addition, the accompanying drawings are only schematic diagrams of the present invention and are not necessarily drawn to scale. The same reference numerals in the drawings represent the same or similar parts, and thus their repeated description will be omitted. Some of the block diagrams shown in the drawings are functional entities and do not necessarily correspond to physically or logically independent entities. The functional entities can be implemented in software form, or in one or more hardware modules or integrated circuits, or in different networks and / or processor systems and / or microcontroller systems.

[0053] It should be understood that although terms such as "first", "second", etc. may be used herein to describe various units, these units should not be limited by these terms. These terms are only used to distinguish one unit from another. For example, without departing from the scope of the exemplary embodiments, the first unit may be referred to as the second unit, and similarly, the second unit may be referred to as the first unit. The term "and / or" used herein includes any and all combinations of one or more of the listed associated items.

[0054] To achieve the above object, please refer to Figures 1 to 3 , the present invention provides an intelligent analysis system for optimizing personalized care plans for kidney diseases, and the system includes the following modules:

[0055] A kidney multi-data fusion module, configured to obtain a comprehensive data set of kidney disease conditions by collecting a multi-data set of kidney diseases in real time and performing multi-data fusion on the multi-data set of kidney diseases;

[0056] A disease key factor analysis module, configured to perform kidney association mining analysis between each sub-metadata corresponding to each type of kidney patient in the comprehensive data set of kidney disease conditions to obtain the disease association degree between each sub-metadata within each type of kidney patient; based on the disease association degree between each sub-metadata within each type of kidney patient, perform disease key factor screening on each sub-metadata corresponding to each type of kidney patient to obtain the disease characteristic key factors corresponding to each type of kidney patient;

[0057] A personalized care plan generation module, configured to obtain successful care plan cases corresponding to the same type of kidney disease patients, and perform personalized care plan generation processing on the disease characteristic key factors corresponding to each type of kidney patient based on the successful care plan cases corresponding to the same type of kidney disease patients and in combination with a convolutional neural network to generate a disease personalized care plan corresponding to each type of kidney patient;

[0058] A care plan dynamic optimization module, configured to perform a nursing effect feedback evaluation on the corresponding renal function indexes of this type of kidney patient during the nursing process based on the disease personalized care plan corresponding to each type of kidney patient to obtain the feedback effect of the change in the renal function indexes corresponding to this type of kidney patient; perform dynamic optimization on the corresponding disease personalized care plan based on the feedback effect of the change in the renal function indexes corresponding to this type of kidney patient to generate a personalized care feedback optimization plan corresponding to this type of kidney patient.

[0059] In an embodiment of the present invention, please refer to Figure 1 shown, which is a schematic diagram of the modules of the intelligent analysis system for optimizing the personalized care plan for kidney diseases of the present invention. In this example, the intelligent analysis system for optimizing the personalized care plan for kidney diseases includes the following modules:

[0060] S1: A kidney multi - data fusion module, which is used to collect a multi - dataset of kidney diseases in real - time and perform multi - data fusion on the multi - dataset of kidney diseases to obtain a comprehensive dataset of kidney disease conditions;

[0061] In the embodiments of the present invention, in the hospital information system, a multi - dataset of kidney diseases is collected in real - time through an integrated data collection module. Renal function indicators such as serum creatinine, blood urea nitrogen, and glomerular filtration rate of patients are obtained from the laboratory information system, as well as routine urine test indicators such as urinary protein, red blood cells, and white blood cells. At the same time, relevant biochemical indicators such as blood glucose, blood lipids, and electrolytes are obtained. Gene data related to kidney diseases is extracted from the gene detection laboratory system. With the help of the wearable device management platform, physiological condition data such as the heart rate, blood pressure, and sleep quality of patients uploaded by smart bracelets and smart blood pressure monitors are collected in real - time. Through the online questionnaire system, lifestyle habit data such as patients' dietary preferences, exercise frequency, smoking and drinking status are regularly collected, and psychological state data such as anxiety and depression levels are collected using professional psychological assessment scales. The pandas library of Python is used to integrate these data from different sources and in different formats into a data structure. Then, the principal component analysis (PCA) algorithm, implemented with the help of the scikit - learn library, is used to perform multi - data fusion on the integrated data. For example, a multi - dataset of kidney diseases containing multiple dimensions such as renal function indicators, gene data, and physiological condition data is input into the PCA algorithm. This algorithm converts multiple related variables into a few uncorrelated comprehensive variables (principal components) through linear transformation. The first few principal components with a cumulative contribution rate of 85% - 95% are selected, and these principal components are combined to form a new dataset, that is, a comprehensive dataset of kidney disease conditions, which is stored in the hospital's big data platform.

[0062] S2: A disease key factor analysis module, which is used to perform kidney - related mining analysis between each sub - metadata corresponding to each type of kidney patient in the comprehensive dataset of kidney disease conditions to obtain the disease correlation degree between each sub - metadata within each type of kidney patient; based on the disease correlation degree between each sub - metadata within each type of kidney patient, disease key factors are screened for each sub - metadata corresponding to each type of kidney patient to obtain the disease - characteristic key factors corresponding to each type of kidney patient;

[0063] In an embodiment of the present invention, on a data analysis workstation, the mlxtend library of Python is used to perform kidney association mining analysis between each sub-metadata corresponding to each type of kidney patient in the comprehensive dataset of kidney disease conditions. First, according to the classification criteria of kidney diseases and the clinical manifestations and pathological conditions of patients, the patients in the comprehensive dataset are divided into different categories. For example, patients with IgA nephropathy and hypertension are classified into one category. For each category of patients, their respective sub-metadata are converted into a format suitable for processing by an association rule mining algorithm. For example, continuous data is discretized. Then, using the Apriori algorithm, with a minimum support of 0.2 and a minimum confidence of 0.6, frequent item sets are found and association rules are generated. For example, for a certain category of patients, the algorithm finds a high degree of association between an increase in serum creatinine value and a decrease in glomerular filtration rate. By calculating indicators such as the confidence and lift of the association rules, the disease association degrees between each sub-metadata within each type of kidney patient are obtained. Based on these association degrees, using a node centrality analysis method, such as calculating the degree centrality of the network nodes corresponding to each sub-metadata in Python using the NetworkX library, and setting the degree centrality threshold to 0.7, the nodes corresponding to the sub-metadata with a degree centrality greater than or equal to 0.7 are selected, and the corresponding sub-metadata are the key disease feature factors corresponding to each type of kidney patient, which are stored in the database.

[0064] S3: A personalized care plan generation module, configured to obtain successful care plan cases corresponding to the same type of kidney disease patients, and perform personalized care plan generation processing on the key disease feature factors corresponding to each type of kidney disease patients based on the successful care plan cases corresponding to the same type of kidney disease patients and in combination with a convolutional neural network, so as to generate a disease personalized care plan corresponding to each type of kidney disease patients;

[0065] In the embodiment of the present invention, in the case library of the nursing plan in the hospital, the SQL query statement "SELECT * FROM successful_care_plans WHERE disease_type ='a specific kidney disease type'" is used to retrieve the successful care plan cases corresponding to the patients with the same type of kidney disease from the table storing the successful care plan cases. For example, for patients with IgA nephropathy, 30 successful care plan cases are obtained. These cases include details such as diet, exercise, drug treatment, and psychological care. In the Python environment, a personalized care plan generation model is constructed using the deep learning framework TensorFlow combined with a convolutional neural network. The successful care plan case data and the key factors of the disease characteristics corresponding to each type of kidney patient are organized into a suitable input format. For example, the characteristics are represented in vector form, and these vectors are input into the model. The model analyzes the matching degree between the characteristics of the successful care plan and the key factors of the disease characteristics through convolutional and fully connected operations. The preset matching threshold is 0.6. If the matching degree is greater than or equal to 0.6, the successful care plan is directly applied; if it is less than 0.6, the successful care plan is adjusted, such as adjusting the diet content, exercise intensity and frequency, drug dosage and time according to the specific situation of the patient, to generate a disease personalized care plan corresponding to each type of kidney patient, and store it in the nursing plan management system in the form of a document.

[0066] S4: The nursing plan dynamic optimization module is used to evaluate the nursing effect feedback of the corresponding renal function indicators of each type of kidney patient during the nursing process based on the disease personalized care plan corresponding to each type of kidney patient, so as to obtain the feedback effect of the change of the renal function indicators corresponding to each type of kidney patient; dynamically optimize the corresponding disease personalized care plan based on the feedback effect of the change of the renal function indicators corresponding to each type of kidney patient, so as to generate a personalized nursing feedback optimization plan corresponding to each type of kidney patient.

[0067] In the embodiments of the present invention, in the hospital information management system, at a predetermined time interval (such as once a week), renal function index data of such kidney patients during the nursing process is obtained from the laboratory information management system, including serum creatinine, blood urea nitrogen, glomerular filtration rate, etc. On the data analysis workstation, using the pandas and numpy libraries of Python and combining with time series analysis methods, the nursing effect feedback of the renal function index data is evaluated. By using pandas to read the renal function index data stored in the information management system, a time series data structure indexed by time is constructed. The functions of numpy are used to calculate the difference between the actual renal function index and the pre-set expected target, such as calculating the deviation of the serum creatinine value from the expected target range. Through time series analysis, the changing trend of the renal function index over time is observed. These analysis results are sorted into an evaluation report to obtain the feedback effect of the corresponding renal function index change of such kidney patients, which is stored in the hospital data analysis database. Based on this feedback effect, a program is written in Python on the data analysis workstation to dynamically optimize the corresponding disease personalized nursing plan. If the evaluation report shows that the improvement of the renal function index is not obvious, such as the serum creatinine value has been higher than the upper limit of the expected target, the program analyzes the reasons by comparing the patient's nursing execution records (diet, exercise, medication situation) with the requirements of the personalized nursing plan. For example, it is found that the patient's actual salt intake exceeds the plan regulations and the exercise execution is irregular. According to these reasons, the corresponding diet, exercise, and medication management suggestions in the personalized nursing plan are adjusted, such as strengthening diet education, strictly controlling salt intake, adjusting the exercise plan and increasing the supervision frequency, and communicating with the doctor to adjust the drug dose if necessary. The adjusted plan is sorted into a document to form the corresponding personalized nursing feedback optimization plan for such kidney patients, which is stored in the nursing plan management system, and the patient's nursing file is updated to ensure that subsequent nursing is carried out according to the optimized plan.

[0068] Further, the kidney multi-data fusion module includes the following functions:

[0069] By using clinical testing equipment to regularly collect renal function indexes of kidney disease patients corresponding to serum creatinine, blood urea nitrogen, and glomerular filtration rate, routine urine indexes including urine protein, red blood cells, and white blood cells, and related biochemical indexes including blood glucose, blood lipids, and electrolytes to obtain clinical test index data;

[0070] By using gene detection technology to obtain gene data related to kidney diseases corresponding to kidney disease patients to obtain kidney patient gene data;

[0071] By using wearable devices including smart bracelets and smart sphygmomanometers to real-time monitor the heart rate, blood pressure, and sleep quality of kidney disease patients corresponding to obtain kidney patient physiological condition data;

[0072] Collect data on the living habits of kidney disease patients, including dietary preferences, exercise frequency, and smoking and drinking status, and data on psychological states, including anxiety and depression levels, through a questionnaire survey to obtain data on the living conditions of kidney patients;

[0073] Merge the clinical test index data, kidney patient gene data, kidney patient physiological condition data, and kidney patient living condition data into a multi - variable dataset for kidney diseases, and perform denoising, error correction, and missing value processing on the multi - variable dataset for kidney diseases to remove the corresponding noise interference in the multi - variable dataset for kidney diseases, correct the corresponding error values, and fill the corresponding missing values using linear interpolation. At the same time, perform normalization processing on data of different types and different magnitudes to unify the multi - variable dataset for kidney diseases into the same numerical range, obtaining a multi - variable standard dataset for kidneys;

[0074] Perform multi - variable data fusion on the multi - variable standard dataset for kidneys to obtain a comprehensive dataset on kidney disease conditions.

[0075] As an embodiment of the present invention, refer to Figure 2 shown in Figure 1 which is a schematic functional flow diagram of the multi - variable data fusion module for kidneys. In this embodiment, the multi - variable data fusion module for kidneys includes the following functions:

[0076] S11: Regularly collect, by using clinical test equipment, renal function indexes of kidney disease patients, including serum creatinine, blood urea nitrogen, and glomerular filtration rate, urine routine indexes, including urinary protein, red blood cells, and white blood cells, and related biochemical indexes, including blood glucose, blood lipids, and electrolytes, to obtain clinical test index data;

[0077] In the embodiment of the present invention, relevant index data are regularly collected for kidney disease patients by using professional clinical test equipment. In the hospital's clinical laboratory, a fully automatic biochemical analyzer is used to detect renal function indexes such as serum creatinine, blood urea nitrogen, and glomerular filtration rate. For example, 2 - 3 ml of fasting venous blood of the patient is collected and injected into a blood collection tube containing anticoagulant. After thorough mixing, the sample is placed on the sample rack of the fully automatic biochemical analyzer. The instrument detects serum creatinine by specific biochemical reactions, such as enzymatic method, and blood urea nitrogen by colorimetric method, and calculates the glomerular filtration rate based on a formula. The specific formula is when Scr ≤ 0.7mg / dl (female) or ≤ 0.9mg / dl (male), GFR = 141×(Scr / κ) -0.329 ×(0.993) 年龄 and when Scr > 0.7mg / dl (female) or > 0.9mg / dl (male), GFR = 141×(Scr / κ) -1.209 ×(0.993) 年龄, where Scr is serum creatinine (mg / dl), κ value is 0.7 for females and 0.9 for males, to obtain accurate renal function index values. For urine routine indicators, patients collect 10-15 ml of clean mid-morning urine in a special urine cup, and use a urine dry chemical analyzer to detect urine protein, red blood cells and white blood cells. The instrument reacts with the chemical components in the urine through the test strip and reads the value according to the color change. These data are collected regularly every day, and the patient's name, medical record number and collection time are recorded to form clinical test index data.

[0078] S12: Obtaining gene data related to kidney disease from patients with kidney disease through gene testing technology to obtain gene data of kidney patients;

[0079] In an embodiment of the present invention, genetic data related to kidney disease in patients with kidney disease are obtained by using advanced genetic testing technology. In a genetic testing laboratory, oral swabs are used to collect oral mucosal epithelial cells of patients. The swabs are gently scraped on the inner wall of the oral cavity 10-15 times to ensure that enough cells are collected. The collected oral swabs are placed in a sampling tube containing a cell preservation solution. Genomic DNA is extracted from the cells using a DNA extraction kit, and polymerase chain reaction (PCR) technology is used to amplify specific gene fragments related to kidney disease, such as the APOL1 gene, the PKD1 gene, etc. The amplified gene fragments are sequenced by a high-throughput sequencer to obtain gene sequence data. The sequencing data are compared with a normal gene database using bioinformatics analysis software, and the gene mutation is analyzed to obtain the gene data of kidney patients, which are stored in a special gene data management system.

[0080] S13: Use wearable devices including smart bracelets and smart blood pressure monitors to monitor the heart rate, blood pressure and sleep quality of patients with kidney disease in real time to obtain the physiological status data of kidney patients;

[0081] In an embodiment of the present invention, the physiological condition of a patient with kidney disease is monitored in real time with the help of wearable devices. The patient wears a smart bracelet and a smart blood pressure monitor. For example, a certain brand of smart bracelet uses photoplethysmogram (PPG) technology, green LED lights and photodiodes to detect changes in blood volume in the blood vessels at the wrist, thereby monitoring the heart rate in real time and recording data in units of minutes. The smart blood pressure monitor uses the oscillometric method. The patient ties the cuff to the upper arm and presses the measurement button. The blood pressure monitor measures the systolic blood pressure, diastolic blood pressure and pulse rate by detecting changes in pressure in the cuff. At the same time, the smart bracelet monitors the patient's heart rate, movements and other data during sleep, and uses built-in algorithms to analyze sleep quality, such as deep sleep time, light sleep time, etc. These devices transmit data to the patient's smartphone via Bluetooth, and then upload it to the hospital's cloud server through a special medical health APP to form the physiological condition data of kidney patients.

[0082] S14: Collect data on the living habits of kidney disease patients, including diet preferences, exercise frequency, and smoking and drinking status, and data on mental states, including anxiety and depression levels, through a questionnaire survey to obtain data on the living conditions of kidney patients;

[0083] In the embodiments of the present invention, data on the living conditions of kidney disease patients is collected through a carefully designed questionnaire survey. In the waiting area or ward of the hospital, medical staff distribute paper questionnaires to patients. The questionnaire content covers diet preferences (such as whether they prefer high-salt and high-protein foods), exercise frequency (number of times of exercise per week, duration of each exercise), smoking and drinking status (number of cigarettes smoked per day, number of times and amount of alcohol consumed per week), and mental state (using the Hospital Anxiety and Depression Scale HADS to evaluate anxiety and depression levels). After the patients fill out the questionnaires, the medical staff collect them on the spot. For patients with limited mobility or illiteracy, the medical staff record the answers through face-to-face interviews. All questionnaire data is entered into a spreadsheet, with the patient's name, medical record number corresponding to the questionnaire answers one by one, and finally data on the living conditions of kidney patients is obtained, providing a basis for subsequent analysis.

[0084] S15: Combine the clinical test index data, kidney patient gene data, kidney patient physiological condition data, and kidney patient living condition data into a multi-source dataset for kidney diseases, and perform denoising, error correction, and missing value processing on the multi-source dataset for kidney diseases to remove the corresponding noise interference in the multi-source dataset for kidney diseases, correct the corresponding error values, and fill in the corresponding missing values using linear interpolation. At the same time, perform normalization processing on data of different types and different magnitudes to unify the multi-source dataset for kidney diseases to the same numerical range, and obtain a standardized multi-source dataset for kidneys;

[0085] In the embodiment of the present invention, on a data analysis workstation, by using the Python programming language in combination with the pandas and numpy libraries, clinical test index data, gene data of kidney patients, physiological condition data of kidney patients, and living condition data of kidney patients are merged into a multivariate dataset of kidney diseases. First, data files from different sources are imported into the Python environment to ensure that the patient identifiers (such as medical record numbers) in each dataset are consistent, and by using the merge function of pandas, the data is merged according to the patient identifier. For denoising processing, a median filtering algorithm is adopted to process numerical data columns (such as serum creatinine values). For example, for a sequence containing serum creatinine values [80, 90, 1000, 110, 120], 1000 is obviously a noise point and is corrected to 100 after median filtering. For error correction, by setting a reasonable value range (such as the normal range of serum creatinine is 44 - 133 μmol / L), error values outside the range are checked and corrected. For missing values, linear interpolation is used. For example, if the glomerular filtration rate data of a certain patient is missing, according to the glomerular filtration rate values at the previous and subsequent time points, the missing value is estimated and filled using a linear relationship. Finally, the functions of numpy are used to normalize data of different types and magnitudes, and all data is unified to the range [0, 1], and finally a multivariate standard dataset of kidneys is obtained.

[0086] S16: Perform multivariate data fusion on the multivariate standard dataset of kidneys to obtain a comprehensive dataset of kidney disease conditions.

[0087] In the embodiment of the present invention, on a data analysis workstation, a data fusion algorithm is used to perform multivariate data fusion on the multivariate standard dataset of kidneys, and a data fusion method based on the feature level is adopted. The principal component analysis (PCA) algorithm is used to process the multivariate standard dataset of kidneys. The PCA algorithm converts multiple correlated variables into a few uncorrelated comprehensive variables (principal components) through linear transformation. For example, the multivariate standard dataset of kidneys containing multiple dimensions such as renal function indexes, urine routine indexes, and physiological condition data is input into the PCA algorithm. The algorithm calculates the covariance matrix of the dataset, and through eigenvalue decomposition, eigenvectors and eigenvalues are obtained. The first few principal components with a cumulative contribution rate reaching 85% - 95% are selected, and these principal components are combined to form a new dataset, that is, the comprehensive dataset of kidney disease conditions. This dataset synthesizes various data information and provides more comprehensive and effective data support for the intelligent analysis of optimizing the personalized care plan for kidney diseases in the future.

[0088] Furthermore, the disease key factor analysis module includes the following functions:

[0089] Classify and subdivide the respective sub-metadata corresponding to each kidney patient in the comprehensive dataset of kidney disease conditions according to the clinical manifestations and pathological conditions of the kidney disease patients, so as to obtain the respective sub-metadata corresponding to each class of kidney patients;

[0090] Perform kidney association mining analysis among the respective sub-metadata corresponding to each class of kidney patients to obtain the disease association degrees among the sub-metadata within each class of kidney patients;

[0091] Based on the disease association degrees among the sub-metadata within each class of kidney patients, construct an association network among the respective sub-metadata corresponding to each class of kidney patients, taking the respective sub-metadata as network nodes, and compare and judge the corresponding disease association degrees according to a preset association threshold. If the corresponding disease association degree is greater than the preset association threshold, then establish a connection edge between the corresponding two network nodes, otherwise do not establish, and generate the sub-metadata association network corresponding to each class of kidney patients;

[0092] Based on the sub-metadata association network corresponding to each class of kidney patients, screen the disease key factors for the respective sub-metadata corresponding to each class of kidney patients to obtain the disease characteristic key factors corresponding to each class of kidney patients.

[0093] As an embodiment of the present invention, refer to Figure 3 shown, for Figure 1 the functional flowchart of the disease key factor analysis module in

[0094] S21: Classify and subdivide the respective sub-metadata corresponding to each kidney patient in the comprehensive dataset of kidney disease conditions according to the clinical manifestations and pathological conditions of the kidney disease patients, so as to obtain the respective sub-metadata corresponding to each class of kidney patients;

[0095] In the embodiment of the present invention, on the data analysis workstation, according to the clinical manifestations (such as edema, proteinuria, hematuria, etc.) and pathological conditions (such as glomerulosclerosis, tubulointerstitial fibrosis, etc.) of the kidney disease patients, classify and subdivide the respective sub-metadata of each kidney patient in the comprehensive dataset of kidney disease conditions, write a script using Python, and combine with medical diagnostic criteria to group the dataset according to the clinical manifestations and pathological conditions of the patients. For example, patients with a large amount of proteinuria and microscopic lesions shown in the pathology are grouped into one category, and patients with edema and membranous nephropathy shown in the pathology are grouped into another category. For the patients in each group, extract their corresponding respective sub-metadata, including renal function indicators, gene data, physiological condition data, and living condition data, etc., and store the respective sub-metadata corresponding to each class of kidney patients after classification and subdivision in a new data table for convenient subsequent analysis.

[0096] S22: Conduct kidney association mining and analysis among the respective sub-metadata corresponding to each type of kidney patient to obtain the disease association degrees among the sub-metadata within each type of kidney patient;

[0097] In an embodiment of the present invention, for the respective sub-metadata corresponding to each type of kidney patient, the Apriori algorithm (such as the Apriori algorithm) is used for kidney association mining and analysis. In a Python environment, the Apriori algorithm is implemented using relevant data mining libraries (such as mlxtend). First, the sub-metadata of each type of patient is converted into a format suitable for algorithm processing. For example, continuous data is discretized. Then, the minimum support and minimum confidence thresholds are set, and the Apriori algorithm is run. The algorithm will find frequent item sets and generate association rules. For example, for a certain type of patient, the algorithm may find a relatively high association degree between an elevated serum creatinine value and a decreased glomerular filtration rate. By calculating indicators such as the confidence and lift of the association rules, the disease association degrees among the sub-metadata within each type of kidney patient are finally obtained, and the results are stored in a database.

[0098] S23: Based on the disease association degrees among the sub-metadata within each type of kidney patient, construct an association network among the respective sub-metadata corresponding to each type of kidney patient, taking the respective sub-metadata as network nodes, and compare and judge the corresponding disease association degrees according to a preset association threshold. If the corresponding disease association degree is greater than the preset association threshold, a connection edge is established between the corresponding two network nodes; otherwise, no connection edge is established, generating a sub-metadata association network corresponding to each type of kidney patient;

[0099] In an embodiment of the present invention, based on the disease association degrees among the sub-metadata within each type of kidney patient, the NetworkX library of Python is used to construct an association network, taking the respective sub-metadata of each type of patient as network nodes. Each node represents a specific sub-metadata, such as serum creatinine, urinary protein, etc. A preset association threshold, such as 0.6, is set. Traverse the disease association degrees among the sub-metadata within each type of patient and compare them with the preset association threshold. If the disease association degree between two sub-metadata is greater than 0.6, a connection edge is established between the corresponding two network nodes. For example, if the association degree between serum creatinine and blood urea nitrogen is 0.7, a connection edge is established between the nodes representing serum creatinine and blood urea nitrogen. Finally, a sub-metadata association network corresponding to each type of kidney patient is generated, and the network structure is stored in a file for subsequent visualization and analysis.

[0100] S24: Screen the disease key factors for the respective sub-metadata corresponding to each type of kidney patient based on the sub-metadata association network corresponding to each type of kidney patient to obtain the disease characteristic key factors corresponding to each type of kidney patient.

[0101] In the embodiments of the present invention, based on the sub - metadata association network corresponding to each type of kidney patient, the node centrality analysis method (such as degree centrality, betweenness centrality, etc.) is used to screen the disease key factors for each sub - metadata corresponding to each type of kidney patient. In the Python environment, the NetworkX library is used to calculate the degree centrality of each node. The degree centrality represents the number of connections of the node. A centrality threshold is set. For example, nodes with a degree centrality greater than or equal to 0.7 are considered key nodes. The sub - metadata corresponding to these key nodes are the disease characteristic key factors for each type of kidney patient. For example, if a node representing a specific gene mutation in the gene data has a high degree centrality, then this gene mutation is the disease characteristic key factor for this type of patient. The screened disease characteristic key factors are stored in a new data table, providing a basis for formulating personalized care plans in the future.

[0102] Furthermore, the kidney - related mining analysis of each sub - metadata corresponding to each type of kidney patient includes:

[0103] Extract and quantify the kidney characteristics of each sub - metadata corresponding to each type of kidney patient. For numerical sub - metadata, the corresponding mean, standard deviation, maximum and minimum values, and range are statistically extracted. For text - type sub - metadata, natural language processing techniques are used to extract the corresponding keywords and theme features and convert them into numerical values that can be used for mathematical calculations. At the same time, the corresponding mean, standard deviation, maximum and minimum values, and range are uniformly statistically calculated for the same numerical - type data to obtain the feature set of each sub - metadata corresponding to each type of kidney patient;

[0104] In the embodiments of the present invention, on the data analysis workstation, Python programming language and its rich libraries are used to extract and quantify kidney characteristics for each sub-metadata corresponding to each type of kidney patient. For numerical sub-metadata, such as the serum creatinine value in renal function indicators, the pandas library is used to read the dataset storing the serum creatinine data of this type of patient. By calling the mean() function of pandas to calculate the mean value, the std() function to calculate the standard deviation, and the max() and min() functions to obtain the maximum and minimum values respectively, and then calculating the difference between the maximum and minimum values to get the amplitude. For example, for the serum creatinine data [80, 90, 100, 110, 120] of a certain type of kidney patient, the mean value calculated by pandas is 100, the standard deviation is about 14.14, the maximum value is 120, the minimum value is 80, and the amplitude is 40. For text-type sub-metadata, such as the medical record text describing the patient's condition, natural language processing technology is adopted, and the NLTK (Natural Language Toolkit) library is used for text preprocessing, including operations such as word segmentation and stop word removal. Through the TextRank algorithm, with the help of NLTK or a dedicated TextRank implementation library, the keywords in the text are extracted. For example, for a medical record text "The patient has severe edema, accompanied by proteinuria, and pathological examination shows mesangial hyperplasia", the keywords "edema", "proteinuria", and "mesangial hyperplasia" are extracted by the TextRank algorithm. Using a topic model, such as Latent Dirichlet Allocation (LDA), implemented through the gensim library, the topic features of the text are analyzed. The extracted keywords and topic features are converted into numerical forms through one-hot encoding or a word vector model (such as Word2Vec). Taking one-hot encoding as an example, assuming the keywords are "edema", "proteinuria", and "hypertension", then "edema" can be encoded as [1, 0, 0], "proteinuria" is encoded as [0, 1, 0], etc. Then, similar to numerical data, statistical calculations of the mean value, standard deviation, maximum and minimum values, and amplitude are performed on the converted numerical values. Finally, the feature sets of each sub-metadata corresponding to each type of kidney patient are obtained and stored in a new data structure.

[0105] Preferably, the feature sets of each sub-metadata corresponding to each type of kidney patient are combined into a feature vector to generate the feature vectors of each sub-metadata corresponding to each type of kidney patient;

[0106] In the embodiments of the present invention, in the Python environment, each subclass of kidney patients corresponds to their respective sub - metadata feature sets to form feature vectors, so as to generate the respective sub - metadata feature vectors corresponding to each subclass of kidney patients. Regarding each previously obtained sub - metadata feature set as an element, the numpy library is used to combine these elements into a one - dimensional array in sequence to form a feature vector. For example, for a certain subclass of kidney patients, the feature set of serum creatinine is [mean 100, standard deviation 14.14, maximum value 120, minimum value 80, amplitude 40], and the feature set of urinary protein after processing is [mean 0.5, standard deviation 0.1, maximum value 0.8, minimum value 0.2, amplitude 0.6], etc. These sub - metadata feature sets are combined into a large one - dimensional array in a specific order (such as the order of renal function indicators, urine routine indicators, gene data features, etc.), such as [100, 14.14, 120, 80, 40, 0.5, 0.1, 0.8, 0.2, 0.6...]. This array is a sub - metadata feature vector corresponding to this subclass of kidney patients. Such combination operations are performed on all sub - metadata of this class of patients to generate multiple sub - metadata feature vectors, and these feature vectors are stored in a list or two - dimensional array for subsequent correlation degree quantification analysis.

[0107] Preferably, the disease - related correlation degree is quantified between the respective sub - metadata feature vectors corresponding to each subclass of kidney patients to obtain the disease correlation degree between the sub - metadata within each subclass of kidney patients.

[0108] In an embodiment of the present invention, by using the scikit-learn library of Python, the disease-related correlation degree is quantified among the respective sub-metadata feature vectors corresponding to each type of kidney patient. Data is read from a previously stored list or two-dimensional array of sub-metadata feature vectors, and a suitable correlation degree calculation method is selected, such as the Pearson correlation coefficient method. Through the pairwise_correlation function in the scikit-learn library, the sub-metadata feature vectors of this type of kidney patient are input, and the function will calculate the Pearson correlation coefficient between every two feature vectors. This coefficient reflects the degree of linear correlation between two sub-metadata feature vectors, and its value range is between -1 and 1. For example, when calculating the Pearson correlation coefficient between the serum creatinine feature vector and the urinary protein feature vector, if the calculation result is 0.7, it indicates that there is a strong positive correlation between the two, that is, there is a certain association between the change in serum creatinine value and the change in urinary protein content. Such calculations are performed pairwise for all sub-metadata feature vectors of this type of patient to obtain a series of correlation coefficient values. These values are the disease correlation degrees among the respective sub-metadata within each type of kidney patient. These correlation degree values are organized into a correlation degree matrix, where the rows and columns of the matrix correspond to different sub-metadata respectively, and the matrix elements are the corresponding disease correlation degrees, which are stored in the database to provide data support for subsequent construction of the association network and screening of disease key factors.

[0109] Further, the screening of disease key factors for the respective sub-metadata corresponding to each type of kidney patient based on the sub-metadata association network corresponding to each type of kidney patient includes:

[0110] Obtaining the degree centrality coefficient corresponding to each network node within each type of kidney patient through the sub-metadata association network corresponding to each type of kidney patient;

[0111] In an embodiment of the present invention, in the Python environment of a data analysis workstation, the NetworkX library is used to operate on the sub - metadata association network corresponding to each type of kidney patient, so as to obtain the degree - centrality coefficient corresponding to each network node within each type of kidney patient. Read the network structure data from the sub - metadata association network file corresponding to each type of kidney patient that was previously generated and stored in a file. For example, for the sub - metadata association network of a certain type of kidney patient, the network contains nodes representing sub - metadata such as serum creatinine, urinary protein, and a certain gene mutation, as well as the connection edges between them. By using the degree_centrality function in the NetworkX library, the read network structure data is input as a parameter into this function. The function will traverse each node in the network and calculate the degree - centrality coefficient of each node. The calculation method of the degree - centrality coefficient is the degree of the node (i.e., the number of edges directly connected to the node) divided by the total number of nodes in the network minus 1. For example, if there are 10 nodes in a network and the node representing serum creatinine is connected to 4 other nodes, then the degree - centrality coefficient of the serum creatinine node is 4÷(10 - 1)≈0.44. Such calculations are performed on all nodes in the sub - metadata association network of this type of kidney patient to obtain the degree - centrality coefficient corresponding to each network node, and the results are stored in a dictionary. The keys of the dictionary are the node names (such as "serum creatinine", "urinary protein", etc.), and the values are the corresponding degree - centrality coefficients.

[0112] Preferably, based on the degree - centrality coefficients corresponding to each network node within each type of kidney patient, screening for associated hub nodes is performed on each network node within the sub - metadata association network corresponding to each type of kidney patient. If the corresponding degree - centrality coefficient is greater than or equal to a preset threshold, then the sub - metadata feature of the network node corresponding to the sub - metadata association network is screened out as a hub factor corresponding to a relatively high degree of association with other sub - metadata features, so as to obtain the disease sub - metadata hub factor corresponding to each type of kidney patient;

[0113] In the embodiment of the present invention, based on the degree centrality coefficients corresponding to each network node within each type of kidney patient obtained previously, the associated hub nodes are screened in the Python environment. A threshold is preset, for example, 0.7. The dictionary storing the degree centrality coefficients generated previously is traversed. For each node and its corresponding degree centrality coefficient, the degree centrality coefficient is compared with the preset threshold. For example, for the sub - metadata association network of a certain type of kidney patient, when traversing to the node representing "glomerular filtration rate", its degree centrality coefficient is 0.75. Since 0.75 ≥ 0.7, the sub - metadata features corresponding to this node are screened out. Assume that the sub - metadata features corresponding to the "glomerular filtration rate" node are stored in a data structure after being processed in the previous steps, and the relevant feature information is extracted from it. In this way, all nodes in the sub - metadata association network of this type of kidney patient are judged and screened. The sub - metadata features corresponding to the nodes with degree centrality coefficients greater than or equal to the preset threshold are organized into a new list. These sub - metadata features are the hub factors corresponding to the relatively high correlation with other sub - metadata features. This list is stored in the database as the disease sub - metadata hub factors corresponding to this type of kidney patient.

[0114] Preferably, the value - availability of the disease sub - metadata hub factors corresponding to each type of kidney patient screened out is calculated to obtain the value - availability corresponding to each sub - metadata factor within each type of kidney patient;

[0115] In the embodiments of the present invention, for each disease sub-metadata hub factor corresponding to each type of kidney patient selected, the valueability is calculated using a custom valueability calculation function in a Python environment. The list of disease sub-metadata hub factors corresponding to each type of kidney patient previously stored is read from the database. Assuming that the valueability calculation function considers multiple factors, such as the occurrence frequency of the sub-metadata factor in different patient groups, its degree of association with the disease severity, etc., for each disease sub-metadata hub factor, by querying the database storing all patient data, the occurrence times of the factor in various types of patients are counted, and the occurrence frequency is calculated. For example, for the hub factor of "proteinuria", among 100 kidney patients of the same type, 70 patients have this factor, then the occurrence frequency is 70÷100 = 0.7. At the same time, by analyzing the relationship between this factor and the disease severity index (such as the deterioration rate of renal function, etc.), the correlation coefficient is calculated using methods such as linear regression. Assuming that the obtained correlation coefficient is 0.6, factors such as the occurrence frequency and the correlation coefficient are weighted and calculated according to a certain weight (such as the occurrence frequency weight of 0.4 and the correlation coefficient weight of 0.6) to obtain the valueability of the sub-metadata factor. For example, the valueability of the "proteinuria" sub-metadata factor = 0.4×0.7 + 0.6×0.6 = 0.64. Such calculations are performed for each sub-metadata factor in the disease sub-metadata hub factors corresponding to each type of kidney patient to obtain the valueability corresponding to each sub-metadata factor, and the results are stored in a new dictionary. The key of the dictionary is the sub-metadata factor name, and the value is the corresponding valueability.

[0116] Preferably, based on the valueability corresponding to each sub-metadata factor within each type of kidney patient, the disease key factors are screened from the corresponding disease sub-metadata hub factors. If the corresponding valueability is greater than 0.5, then the corresponding disease sub-metadata hub factor is screened out as a key feature factor; otherwise, it is eliminated to obtain the disease feature key factors corresponding to each type of kidney patient.

[0117] In an embodiment of the present invention, in a Python environment, based on the value degrees corresponding to each sub - metadata factor within each type of kidney patient obtained previously, disease key factor screening is performed on the corresponding disease sub - metadata hub factors. Data is read from the dictionary storing value degrees generated previously. The preset value degree screening threshold is 0.5. Each sub - metadata factor in the dictionary and its corresponding value degree are traversed, and the value degree is compared with the preset threshold. For example, for the "certain specific gene mutation" sub - metadata factor in the disease sub - metadata hub factor of a certain type of kidney patient, its value degree is 0.55. Since 0.55 > 0.5, the corresponding disease sub - metadata hub factor is screened out as a key feature factor. For the "family history of hypertension" sub - metadata factor, its value degree is 0.45. Since 0.45 < 0.5, it is excluded. In this way, all disease sub - metadata hub factors corresponding to each type of kidney patient are screened, and the screened key feature factors are organized into a new list. These key feature factors are the disease - characteristic key factors corresponding to each type of kidney patient and are stored in the database to provide a core basis for formulating personalized care plans in the future.

[0118] Further, the calculation of the value degrees for the disease sub - metadata hub factors corresponding to each type of kidney patient after screening includes:

[0119] Perform eigenvalue statistical analysis on the disease sub - metadata hub factors corresponding to each type of kidney patient to calculate the mean, standard deviation, maximum value, and minimum value statistics corresponding to each sub - metadata factor, and form a corresponding feature statistical matrix to generate the disease sub - metadata feature statistical matrix corresponding to each type of kidney patient;

[0120] In an embodiment of the present invention, in a Python environment, the pandas library is used to perform eigenvalue statistical analysis on the disease sub-metadata hub factors corresponding to each type of kidney patient. The list of disease sub-metadata hub factors corresponding to each type of kidney patient previously stored in the database is read. Suppose the disease sub-metadata hub factors of a certain type of kidney patient include "serum creatinine", "proteinuria", "glomerular filtration rate", etc. For each sub-metadata factor, data is read from the database table storing the data of all patients for this factor. For example, the serum creatinine values of 100 patients of this type are read. The mean() function of pandas is used to calculate the mean, the std() function to calculate the standard deviation, the max() function to obtain the maximum value, and the min() function to obtain the minimum value. Suppose the mean of these 100 serum creatinine values after calculation is 90 μmol / L, the standard deviation is 10 μmol / L, the maximum value is 120 μmol / L, and the minimum value is 60 μmol / L. These statistics are organized into a one-dimensional array in a specific order (mean, standard deviation, maximum value, minimum value), such as [90, 10, 120, 60]. Such operations are performed on all the disease sub-metadata hub factors of this type of patient, and the obtained one-dimensional arrays are combined into a two-dimensional array by rows. This two-dimensional array is the corresponding feature statistical matrix. For example, for the "proteinuria" factor, the statistical array [0.8, 0.2, 1.5, 0.3] is calculated. The statistical arrays of serum creatinine and proteinuria are combined into the feature statistical matrix [[90, 10, 120, 60], [0.8, 0.2, 1.5, 0.3]]. By analogy, the disease sub-metadata feature statistical matrix corresponding to each type of kidney patient is generated and stored in the database.

[0121] Preferably, based on the disease sub-metadata feature statistical matrix corresponding to each type of kidney patient, factor variability score calculation is performed on the corresponding disease sub-metadata hub factors to obtain the disease sub-metadata hub factor variability score corresponding to each type of kidney patient;

[0122] In the embodiment of the present invention, based on the disease sub-element feature statistical matrix corresponding to each type of kidney patient obtained previously, factor variability score calculation is performed using a custom function in the Python environment. The disease sub-element feature statistical matrix data corresponding to each type of kidney patient is read from the database. Assuming that the factor variability score calculation function considers two factors, namely the ratio of the standard deviation to the mean and the ratio of the data range (maximum value - minimum value) to the mean. For each row in the matrix (i.e., the statistic corresponding to each disease sub-element data hub factor), first calculate the ratio of the standard deviation to the mean. For example, for the serum creatinine statistic [90, 10, 120, 60], this ratio is 10÷90≈0.11. Then calculate the ratio of the data range to the mean, that is, (120 - 60)÷90≈0.67. These two ratios are weighted and calculated according to certain weights (such as the weight of the ratio of the standard deviation to the mean is 0.4, and the weight of the ratio of the data range to the mean is 0.6) to obtain the variability score of this factor. For example, the serum creatinine factor variability score = 0.4×0.11 + 0.6×0.67 = 0.446. Such calculations are performed for each row in the disease sub-element feature statistical matrix of this type of patient, obtaining a series of variability score values. These values are organized into a list, and this list is the disease sub-element hub factor variability score corresponding to each type of kidney patient, which is stored in the database.

[0123] Preferably, based on the disease sub-element hub factor variability score corresponding to each type of kidney patient and in combination with the disease sub-element feature statistical matrix, a valueability weighted calculation is performed on the corresponding disease sub-element data hub factor to obtain the valueability corresponding to each sub-element data factor within each type of kidney patient.

[0124] In an embodiment of the present invention, in a Python environment, based on the variability scores of disease sub-element hub factors corresponding to each type of kidney patient obtained previously and in combination with the disease sub-element feature statistical matrix, a custom value-degree weighted calculation function is used to perform value-degree weighted calculation on the corresponding disease sub-element data hub factors. The variability score list of disease sub-element hub factors corresponding to each type of kidney patient and the disease sub-element feature statistical matrix data are read from the database. Assume that the value-degree weighted calculation function also considers the occurrence frequency of sub-element data factors in different patient groups (statistically obtained from the database storing all patient data). For each disease sub-element data hub factor, first obtain its statistics such as mean and standard deviation from the disease sub-element feature statistical matrix, then obtain its variability score from the variability score list, and at the same time query the occurrence frequency of this factor in patients of the same type from the database. For example, for the "proteinuria" factor, its mean in the disease sub-element feature statistical matrix is 0.8, the standard deviation is 0.2, the variability score is 0.35, and the occurrence frequency in 100 patients of the same type is 0.6. These factors are weighted and calculated according to certain weights (such as mean weight 0.3, standard deviation weight 0.2, variability score weight 0.3, occurrence frequency weight 0.2) to obtain the value-degree of this sub-element data factor. For example, the value-degree of the "proteinuria" sub-element data factor = 0.3×0.8 + 0.2×0.2 + 0.3×0.35 + 0.2×0.6 = 0.515. Such calculations are performed on all disease sub-element data hub factors corresponding to each type of kidney patient to obtain the value-degree corresponding to each sub-element data factor, and the results are stored in a new dictionary. The key of the dictionary is the name of the sub-element data factor, and the value is the corresponding value-degree, providing a basis for subsequent screening of disease key factors.

[0125] Further, the personalized care plan generation module includes the following functions:

[0126] Obtain successful care plan cases corresponding to kidney disease patients of the same type;

[0127] In the embodiment of the present invention, in the nursing plan management database of the hospital, according to the classification criteria of kidney diseases and the disease characteristics of patients, successful nursing plan cases corresponding to patients of the same type are screened out. Using an SQL query statement, such as "SELECT * FROM care_plan WHERE disease_type ='a specific type of kidney disease' AND success_status ='success'", all nursing plans for a specific type of kidney disease that are marked as successful are retrieved from the nursing plan table in the database. Assuming that a certain type of kidney disease is IgA nephropathy, by executing the above query statement, 50 successful nursing plan cases for IgA nephropathy patients are obtained. These cases contain detailed records of nursing measures, such as diet arrangements, exercise guidance, drug nursing plans, and psychological nursing methods, etc. The obtained successful nursing plan case data is exported as a CSV format file for convenient subsequent analysis and processing.

[0128] Preferably, perform a disease characteristic analysis of the successful nursing plan for patients of the same type of kidney disease to obtain the disease characteristics of the successful nursing plan corresponding to patients of the same type;

[0129] In the embodiments of the present invention, by using text analysis libraries of Python, such as NLTK and pandas, disease feature analysis of successful nursing plan cases corresponding to patients with the same kidney disease is carried out to read the previously exported CSV file into the DataFrame of pandas, and preprocessing is performed on the text description of the nursing plan, including operations such as word segmentation and stop word removal. For example, for a nursing plan record "Patients should follow a low-salt, low-fat diet, do aerobic exercise three times a week, 30 minutes each time, and take sartan antihypertensive drugs on time", after word segmentation using NLTK, we get ["Patients", "should", "follow", "low-salt", "low-fat", "diet", "weekly", "do", "three times", "aerobic exercise", "each time", "30 minutes", "on time", "take", "sartan", "antihypertensive drugs"], and after removing stop words, the keywords related to disease care are retained ["low-salt", "low-fat", "diet", "weekly", "three times", "aerobic exercise", "each time", "30 minutes", "sartan", "antihypertensive drugs"]. By keyword matching and topic model analysis, such as using the Latent Dirichlet Allocation (LDA) model implemented with the gensim library, the key features related to the disease in the successful nursing plan are determined. For example, through LDA model analysis, it is found that in the successful nursing plan for patients with IgA nephropathy, themes such as "blood pressure control", "low-protein diet", and "moderate exercise" appear with a higher frequency. These themes and related keywords are sorted into the disease features of the successful nursing plan corresponding to patients of the same type and stored in a new DataFrame.

[0130] Preferably, a corresponding personalized nursing plan generation model is constructed by combining a convolutional neural network, and the disease features of the successful nursing plan corresponding to patients of the same type and the key factors of the disease features corresponding to each type of kidney patient are input into the personalized nursing plan generation model for personalized nursing plan generation processing to generate a disease personalized nursing plan corresponding to each type of kidney patient.

[0131] In the embodiment of the present invention, in the Python environment, a personalized care plan generation model is constructed by using the deep learning framework TensorFlow in combination with a convolutional neural network. First, the structure of the convolutional neural network is defined, including convolutional layers, pooling layers, and fully connected layers, etc. For example, a network structure including two convolutional layers and two fully connected layers is constructed. The convolutional layer is used to extract the features of the input data, the pooling layer is used to reduce the data dimension, and the fully connected layer is used to output the final result. The disease features of the successful care plans corresponding to the same type of patients obtained previously and the key disease feature factors corresponding to each type of kidney patients are sorted into a suitable input format, such as representing the features in vector form, and these vectors are input into the personalized care plan generation model. The model analyzes the feature matching scores between the disease features of the successful care plans and the key disease feature factors corresponding to each type of kidney patients through convolutional and fully connected operations. A matching threshold is preset, for example, 0.6. For the data of each type of kidney patients, the model calculates the feature matching score. Suppose for a certain type of IgA nephropathy patients, the feature matching score calculated by the model is 0.7. Since 0.7≥0.6, the successful care plan corresponding to this type of patients is directly applied. If for another type of patients, the feature matching score is 0.5, which is less than the preset threshold, then the case comparison analysis method is used to select a case from the successful care plan case library that is relatively similar to the features of this type of patients, compare the differences between the current patients and the case patients, such as diet preferences, exercise ability, disease severity, etc., and adjust the corresponding diet content (such as adjusting the types and intakes of foods), exercise intensity and frequency (such as increasing or decreasing the exercise time and times), and the dosage and time of drug use (such as adjusting the dosage of antihypertensive drugs according to the renal function of the patients) in the successful care plan. Finally, a disease personalized care plan corresponding to each type of kidney patients is generated and stored in the form of a document, recording in detail each care measure and its basis.

[0132] Further, the generation process of the personalized care plan specifically analyzes the feature matching scores between the disease features of the successful care plans and the key disease feature factors corresponding to each type of kidney patients in the personalized care plan generation model, and makes a judgment according to the preset matching threshold. If the feature matching score is greater than or equal to the preset matching threshold, the successful care plan corresponding to this type of patients is applied; if the feature matching score is less than the preset matching threshold, the case comparison analysis is used to identify and adjust the corresponding diet content, exercise intensity and frequency, and the dosage and time of drug use in the successful care plan to generate a disease personalized care plan corresponding to each type of kidney patients.

[0133] Further, the care plan dynamic optimization module includes the following functions:

[0134] Apply the disease personalized care plan corresponding to each type of kidney patient to the corresponding kidney disease patients to generate the care process corresponding to this type of kidney patient;

[0135] In the embodiment of the present invention, in the nursing execution system of the hospital, the disease personalized care plan corresponding to each type of kidney patient is associated with the corresponding kidney disease patients and implemented. For each kidney disease patient belonging to a specific category, the nursing staff retrieves the corresponding personalized care plan document from the system. For example, for a certain type of patient with IgA nephropathy, the personalized care plan includes a low-salt diet every day (salt intake not exceeding 3 grams), slow walking exercise three times a week for 30 minutes each time, and taking a specific dose of sartan antihypertensive drug at 8 am every day. According to the plan, the nursing staff supervises the patient's diet every day to ensure that they strictly follow the low-salt diet requirements, prepares low-salt food ingredients and arranges meals for the patient. During exercise, the nursing staff guides the patient to perform slow walking exercise on-site, records the exercise time and the patient's exercise status, reminds the patient to take medicine on time, and confirms whether the patient takes the medicine on time and in the correct dosage. During the entire nursing process, the nursing staff details the patient's nursing situation, such as the types and intakes of daily diet, the actual implementation of exercise, the time and dosage of medicine taking, etc., to form the nursing process record corresponding to this type of kidney patient, which is stored in the patient nursing file of the nursing execution system.

[0136] Preferably, analyze the renal function indicators during the care process corresponding to this type of kidney patient to obtain the renal function indicators corresponding to this type of kidney patient during the care process;

[0137] In the embodiment of the present invention, in the hospital's laboratory department, at a predetermined time interval (such as once a week), analyze the renal function indicators during the care process corresponding to this type of kidney patient, and use a fully automatic biochemical analyzer to detect renal function indicators such as serum creatinine and blood urea nitrogen. For example, collect 3 ml of fasting venous blood from the patient, inject it into a blood collection tube containing anticoagulant, mix well, and then place the sample on the sample rack of the fully automatic biochemical analyzer. The instrument performs specific biochemical reactions, such as using the enzyme method to detect serum creatinine and the colorimetric method to detect blood urea nitrogen. At the same time, by collecting the patient's 24-hour urine, detect indicators such as 24-hour urinary protein quantification. For the glomerular filtration rate, use the simplified MDRD formula or the CKD-EPI formula, combined with the patient's serum creatinine value, age, gender and other information for calculation. For example, given that a patient's serum creatinine value is 100 μmol / L, age 50 years old, male, the glomerular filtration rate is calculated to be approximately 75 ml / min / 1.73m by the simplified MDRD formula. 2The renal function index data obtained from each detection, including serum creatinine value, blood urea nitrogen value, 24-hour urinary protein quantification, glomerular filtration rate, etc., are recorded in the patient's test report in chronological order and simultaneously entered into the hospital's information management system for subsequent analysis, so as to obtain the corresponding renal function index data sequence of this type of kidney patient during the nursing process.

[0138] Preferably, obtain the expected target of the renal function index corresponding to this type of kidney patient, and based on the expected target of the renal function index and using time series analysis, conduct a nursing effect feedback evaluation among the corresponding renal function indexes of this type of kidney patient during the nursing process, so as to obtain the corresponding renal function index change feedback effect of this type of kidney patient;

[0139] In the embodiment of the present invention, the expected target of the renal function index corresponding to this type of kidney patient is obtained from the hospital's clinical nursing guideline database or the nursing plan document formulated by the doctor for this type of patient. For example, for a certain type of IgA nephropathy patient, the expected target of the renal function index is that the serum creatinine is maintained at 80-120 μmol / L, the blood urea nitrogen is maintained at 3.2-7.1 mmol / L, the 24-hour urinary protein quantification is less than 1 gram, and the glomerular filtration rate is stable at 60-90 ml / min / 1.73m 2 On the data analysis workstation, using the data analysis libraries of Python, such as pandas and numpy, combined with the time series analysis method, conduct a nursing effect feedback evaluation on the corresponding renal function index data sequence of this type of kidney patient during the nursing process. Read the renal function index data stored in the hospital information management system using pandas and convert it into a time series data structure. For example, with a weekly time interval, construct a time series of the change of serum creatinine value over time, and use the functions of numpy to calculate the difference between the actual renal function index and the expected target, such as calculating the deviation of the serum creatinine value from the expected target range. Through time series analysis, observe the change trend of the renal function index over time, such as whether the serum creatinine value is gradually decreasing, increasing or remaining stable. Organize these analysis results into an evaluation report, including the comparison of each renal function index with the expected target, trend analysis, etc., and finally obtain the corresponding renal function index change feedback effect of this type of kidney patient and store it in the hospital's data analysis database.

[0140] Preferably, dynamically optimize the corresponding disease personalized nursing plan based on the corresponding renal function index change feedback effect of this type of kidney patient to generate the corresponding personalized nursing feedback optimization plan for this type of kidney patient.

[0141] In the embodiments of the present invention, on a data analysis workstation, a program is written in Python to dynamically optimize the corresponding disease personalized care plan based on the feedback effect of the change in renal function indicators of this type of kidney patients. An evaluation report on the feedback effect of the change in renal function indicators obtained previously is read from the data analysis database of the hospital. If the evaluation report shows that the improvement of renal function indicators is not obvious, for example, the serum creatinine value has been higher than the upper limit of the expected target during the nursing process, the program starts the cause analysis process. By comparing the patient's nursing process records (diet, exercise, medication) with the requirements of the personalized care plan, possible reasons are analyzed. For example, it is found that the patient's actual salt intake exceeds the plan requirement, reaching 5 grams per day, and the exercise is not carried out regularly, only once a week. Based on these reasons, the corresponding diet, exercise, and medication management suggestions in the disease personalized care plan are adjusted. For example, strengthen the diet education for the patient, strictly control the salt intake within 3 grams, adjust the exercise plan to at least four times a week, 30 minutes of aerobic exercise each time, and increase the supervision frequency of the patient's exercise. For medication, if the improvement of the patient's renal function indicators is not obvious and other factors are excluded, consider communicating with the doctor and adjusting the dosage of antihypertensive drugs according to the patient's latest renal function. The adjusted care plan is sorted into a document to form a personalized care feedback optimization plan corresponding to this type of kidney patients, stored in the hospital's care plan management database, and the patient's nursing file is updated to ensure that the subsequent nursing work is carried out according to the optimized plan.

[0142] Further, the personalized care feedback optimization plan is specifically as follows: if it is reflected from the feedback effect of the corresponding change in renal function indicators that the improvement of renal function indicators is not obvious, then analyze the reasons and adjust the corresponding diet, exercise, and medication management suggestions in the disease personalized care plan based on the reasons.

[0143] Therefore, from any point of view, the embodiments should be regarded as exemplary and non-limiting. The scope of the present invention is defined by the appended claims rather than the above description. Therefore, it is intended to encompass all changes falling within the meaning and scope of the equivalent elements of the application documents within the present invention.

[0144] The above are only the specific embodiments of the present invention, enabling those skilled in the art to understand or implement the present invention. Various modifications to these embodiments will be obvious to those skilled in the art. The general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to these embodiments shown herein, but rather to the broadest scope consistent with the principles and novel features invented herein.

Claims

1. An intelligent analysis system for optimizing personalized care plans for kidney diseases, characterized in that, It includes the following modules: A kidney multi - data fusion module, which is used to collect a multi - data set of kidney diseases in real - time and perform multi - data fusion on the multi - data set of kidney diseases to obtain a comprehensive data set of kidney disease conditions; A disease key factor analysis module, which is used to conduct kidney - related mining analysis between each sub - metadata corresponding to each type of kidney patient in the comprehensive data set of kidney disease conditions to obtain the disease correlation degree between each sub - metadata within each type of kidney patient; screen disease key factors for each sub - metadata corresponding to each type of kidney patient based on the disease correlation degree between each sub - metadata within each type of kidney patient to obtain the disease - characteristic key factors corresponding to each type of kidney patient; A personalized nursing plan generation module, which is used to obtain successful nursing plan cases corresponding to the same type of kidney disease patients, and perform personalized nursing plan generation processing on the disease - characteristic key factors corresponding to each type of kidney patient based on the successful nursing plan cases corresponding to the same type of kidney disease patients and in combination with a convolutional neural network to generate a disease - personalized nursing plan corresponding to each type of kidney patient; A nursing plan dynamic optimization module, which is used to conduct a nursing effect feedback assessment on the renal function indicators corresponding to a certain type of kidney patient during the nursing process based on the disease - personalized nursing plan corresponding to each type of kidney patient to obtain the feedback effect of the change in the renal function indicators corresponding to this type of kidney patient; dynamically optimize the corresponding disease - personalized nursing plan based on the feedback effect of the change in the renal function indicators corresponding to this type of kidney patient to generate a personalized nursing feedback optimization plan corresponding to this type of kidney patient.

2. The intelligent analysis system for optimizing the personalized care plan for kidney diseases according to claim 1, wherein The kidney multi - data fusion module includes the following functions: Regularly collect renal function indicators including serum creatinine, blood urea nitrogen, and glomerular filtration rate, urine routine indicators including urinary protein, red blood cells, and white blood cells, and related biochemical indicators including blood glucose, blood lipids, and electrolytes corresponding to kidney disease patients by using clinical testing equipment to obtain clinical test index data; Obtain gene data related to kidney diseases corresponding to kidney disease patients through gene detection technology to obtain kidney patient gene data; Real - time monitor the heart rate, blood pressure, and sleep quality corresponding to kidney disease patients through wearable devices including smart bracelets and smart blood pressure monitors to obtain kidney patient physiological condition data; Collect lifestyle data including diet preferences, exercise frequency, and smoking and drinking status and psychological state data including anxiety and depression levels corresponding to kidney disease patients through questionnaire surveys to obtain kidney patient living condition data; Merge the clinical test index data, kidney patient gene data, kidney patient physiological condition data, and kidney patient living condition data into a multi - data set of kidney diseases, and perform denoising, error correction, and missing value processing on the multi - data set of kidney diseases to remove the corresponding noise interference in the multi - data set of kidney diseases, correct the corresponding error values, fill the corresponding missing values by using linear interpolation, and at the same time perform normalization processing on the data corresponding to different types and different magnitudes to unify the multi - data set of kidney diseases into the same numerical range to obtain a multi - standard data set of kidneys; Perform multivariate data fusion on the kidney multivariate standard dataset to obtain a comprehensive dataset of kidney disease status.

3. The intelligent analysis system for optimizing the personalized care plan for kidney diseases according to claim 1, wherein The disease key factor analysis module includes the following functions: Classify and subdivide each sub-metadata corresponding to each kidney patient in the comprehensive dataset of kidney disease status according to the clinical manifestations and pathological conditions of the kidney disease patients, so as to obtain each sub-metadata corresponding to each type of kidney patient; Perform kidney association mining analysis among the respective sub-metadata corresponding to each type of kidney patient to obtain the disease association degree among the sub-metadata within each type of kidney patient; Based on the disease association degree among the sub-metadata within each type of kidney patient, construct an association network among the respective sub-metadata corresponding to each type of kidney patient, taking each of the respective sub-metadata as network nodes, and compare and judge the corresponding disease association degree according to a preset association threshold. If the corresponding disease association degree is greater than the preset association threshold, then establish a connection edge between the corresponding two network nodes, otherwise do not establish, and generate a sub-metadata association network corresponding to each type of kidney patient; Based on the sub-metadata association network corresponding to each type of kidney patient, screen the disease key factors for each sub-metadata corresponding to each type of kidney patient to obtain the disease characteristic key factors corresponding to each type of kidney patient.

4. The intelligent analysis system for optimizing the personalized care plan for kidney diseases according to claim 3, wherein The kidney association mining analysis performed among the respective sub-metadata corresponding to each type of kidney patient includes: Extract and quantify the kidney characteristics of each sub-metadata corresponding to each type of kidney patient. For numerical sub-metadata, statistically extract its corresponding mean, standard deviation, maximum and minimum values, and range. For text sub-metadata, use natural language processing technology to extract the corresponding keywords and theme features and convert them into numerical values that can be used for mathematical calculations. At the same time, uniformly calculate the corresponding mean, standard deviation, maximum and minimum values, and range for the same numerical type, so as to obtain the sub-metadata feature set corresponding to each type of kidney patient; Form a feature vector from the sub-metadata feature sets corresponding to each sub-metadata of each type of kidney patient to generate the sub-metadata feature vectors corresponding to each type of kidney patient; Quantify the disease-related association degree among the sub-metadata feature vectors corresponding to each type of kidney patient to obtain the disease association degree among the sub-metadata within each type of kidney patient.

5. The intelligent analysis system for optimizing the personalized care plan for kidney diseases according to claim 3, characterized in that, The screening of disease key factors for each sub-metadata corresponding to each type of kidney patient based on the sub-metadata association network corresponding to each type of kidney patient includes: Obtain the degree centrality coefficient corresponding to each network node within each type of kidney patient through the sub-metadata association network corresponding to each type of kidney patient; Based on the degree centrality coefficient corresponding to each network node within each type of kidney patient, screen the association hub nodes in the sub-metadata association network corresponding to each type of kidney patient. If the corresponding degree centrality coefficient is greater than or equal to the preset threshold, then screen out the sub-metadata features corresponding to the network nodes of the sub-metadata association network as the hub factors with relatively high association degrees with other sub-metadata features, so as to obtain the disease sub-metadata hub factors corresponding to each type of kidney patient; Calculate the valueability of the disease sub-metadata hub factors corresponding to each type of kidney patient selected, so as to obtain the valueability corresponding to each sub-metadata factor within each type of kidney patient; Based on the valueability corresponding to each sub-metadata factor within each type of kidney patient, screen the disease key factors for the corresponding disease sub-metadata hub factors. If the corresponding valueability is greater than 0.5, then screen out the corresponding disease sub-metadata hub factors as key feature factors, otherwise eliminate them, so as to obtain the disease feature key factors corresponding to each type of kidney patient.

6. The intelligent analysis system for optimizing the personalized care plan for kidney diseases according to claim 5, characterized in that, The calculation of the valueability of the disease sub-metadata hub factors corresponding to each type of kidney patient selected includes: Conduct eigenvalue statistical analysis on the disease sub-metadata hub factors corresponding to each type of kidney patient, so as to calculate the mean, standard deviation, maximum value and minimum value statistics corresponding to each sub-metadata factor, and form the corresponding feature statistical matrix, so as to generate the disease sub-feature statistical matrix corresponding to each type of kidney patient; Based on the disease sub-feature statistical matrix corresponding to each type of kidney patient, calculate the factor variability score for the corresponding disease sub-metadata hub factors, and obtain the disease sub-hub factor variability score corresponding to each type of kidney patient; Based on the disease sub-hub factor variability score corresponding to each type of kidney patient and in combination with the disease sub-feature statistical matrix, conduct valueability weighted calculation on the corresponding disease sub-metadata hub factors, so as to obtain the valueability corresponding to each sub-metadata factor within each type of kidney patient.

7. The intelligent analysis system for optimizing the personalized care plan for kidney diseases according to claim 1, characterized in that, The personalized care plan generation module includes the following functions: Obtain the successful care plan cases corresponding to the same type of kidney disease patients; Conduct successful care plan disease feature analysis on the successful care plan cases corresponding to the same type of kidney disease patients, so as to obtain the successful care plan disease features corresponding to the same type of patients; Construct a corresponding personalized care plan generation model by combining a convolutional neural network, and input the successful care plan disease features corresponding to the same type of patients and the disease feature key factors corresponding to each type of kidney patient into the personalized care plan generation model for personalized care plan generation processing, so as to generate the disease personalized care plan corresponding to each type of kidney patient.

8. The intelligent analysis system for optimizing the personalized care plan for kidney diseases according to claim 7, wherein, The personalized care plan generation processing is specifically to analyze the feature matching score between the successful care plan disease features and the disease feature key factors corresponding to each type of kidney patient in the personalized care plan generation model, and make a judgment according to the preset matching threshold. If the feature matching score is greater than or equal to the preset matching threshold, then apply the successful care plan corresponding to this type of patient; If the feature matching score is less than the preset matching threshold, then use case comparison analysis to identify and adjust the corresponding diet content, exercise intensity and frequency, and the dosage and time corresponding to drug use in the successful care plan, so as to generate the disease personalized care plan corresponding to each type of kidney patient.

9. The intelligent analysis system for optimizing the personalized care plan for kidney diseases according to claim 1, characterized in that, The care plan dynamic optimization module includes the following functions: Apply the disease personalized care plan corresponding to each type of kidney patient to the corresponding kidney disease patients, so as to generate the care process corresponding to this type of kidney patient; During the nursing process for this type of kidney patients, renal function indicators are analyzed to obtain the corresponding renal function indicators of this type of kidney patients during the nursing process; Obtain the expected target of the renal function indicators for this type of kidney patients, and based on the expected target of the renal function indicators and using time series analysis, conduct a feedback evaluation of the nursing effect among the corresponding renal function indicators of this type of kidney patients during the nursing process to obtain the feedback effect of the change in the renal function indicators of this type of kidney patients; Dynamically optimize the corresponding disease personalized nursing plan based on the feedback effect of the change in the renal function indicators of this type of kidney patients to generate the corresponding personalized nursing feedback optimization plan for this type of kidney patients.

10. The intelligent analysis system for optimizing the personalized care plan for kidney diseases according to claim 9, characterized in that, The specific content of the personalized nursing feedback optimization plan is that if the improvement of the renal function indicators is not obvious according to the feedback effect of the change in the corresponding renal function indicators, analyze the reasons and adjust the corresponding diet, exercise and drug administration management suggestions in the disease personalized nursing plan based on the reasons.

Images