Method, system and equipment for evaluating early screening model of chronic kidney disease and medium

Abstract

The invention relates to the technical field of disease diagnosis, in particular to a chronic kidney disease early screening model evaluation method, which comprises the following steps: acquiring sample case data, preprocessing the sample case data, and constructing a data set; carrying out statistical analysis on the data set to obtain a characteristic factor variable related to a target factor, namely whether the target factor is a CKD onset risk or not, giving data quality and clinical expert opinions, and screening prediction factors; fitting data corresponding to the predictive factors by utilizing an algorithm model, performing comparative analysis according to a plurality of fitting results of different predictive factors, selecting the predictive factor corresponding to the optimal fitting result as an influence factor of the model, and determining an optimal training model; and finally, evaluating the effect of the model in a model verification set according to drawing a subject working characteristic curve and calculating the area under the curve to obtain a fitting result of the CKD early screening model. According to the fitting result of the CKD early screening model, whether the parameter or the data set of the CKD early screening model needs to be adjusted is determined, and the screening accuracy is high.

Description

Technical Field

[0001] The present invention relates to the technical field of disease diagnosis, and in particular to an early screening model evaluation method, system, equipment and medium for chronic kidney disease. Background Art

[0002] Among many chronic diseases, the incidence and prevalence of chronic kidney disease (CKD) rank first in the world. Therefore, it is very important to detect kidney damage early and prevent the occurrence of CKD.

[0003] The current research method for early CKD screening is based on selecting effective predictive factors to establish relevant prediction models, and then analyzing and evaluating the prediction models in combination with corresponding model data. However, all CKD early screening models have the following main problems that need to be improved:

[0004] 1. Outcome definition, namely the definition of chronic kidney disease (CKD). CKD is defined as abnormal kidney structure or function for more than three months, with a decreased glomerular filtration rate (eGFR) less than 60 ml·min-1·(1.73 m2)-1. Because the diagnosis of CKD is long, cumbersome, and time-consuming, current studies rely on single GFR estimates. This cannot rule out the possibility of some patients with acute reversible renal injury, leading to an overestimation of CKD incidence and interference with prediction models.

[0005] 2. The number of model predictors is large. The number of predictors included in existing prediction models ranges from 4 to 17. The more predictors in a model, the better. The more predictors included, the more likely overfitting will occur during modeling.

[0006] 3. Missing data. The existing prediction models process patient indicator data related to predictive factors by excluding patients with missing data or not performing any analysis or explanation on the indicator items with missing data. A few model cases use data interpolation methods to fill in the missing data items. Although a few models use multiple interpolation methods to process missing data because they can retain the randomness of missing data and have higher accuracy than single interpolation methods, the above data operation methods still cannot improve the accuracy of the data set to the same level as the real clinical data set, and fitting analysis of the real clinical validation set data will cause interference.

[0007] The existing CKD early screening model has many predictive factors and is prone to interference during the analysis process. The CKD early screening model has low screening accuracy for CKD. Summary of the Invention

[0008] The present invention aims to solve at least one of the technical problems existing in the prior art.

[0009] To this end, the present invention provides an early screening model evaluation method for chronic kidney disease, which improves the screening accuracy of CKD.

[0010] According to an embodiment of the present invention, the early screening model evaluation method for chronic kidney disease comprises the following steps:

[0011] S1, data collection and collation, obtaining sample case data, preprocessing the sample case data, and constructing a data set;

[0012] S2, statistical analysis of data: Statistical analysis of the data set to obtain characteristic factor variables related to the target factor, i.e., whether the risk of CKD is present. Based on data quality and clinical expert opinions, predictive factors are screened;

[0013] S3, data fitting analysis: use the algorithm model to fit the data corresponding to the predictor, conduct comparative analysis based on multiple fitting results of different predictors, select the predictor with the best fitting result as the influencing factor of the model, and determine the optimal training model. Finally, in the model validation set, the receiver operating characteristic curve is plotted and the area under the curve is calculated to evaluate the model effect, and the fitting results of the CKD early screening model are obtained;

[0014] S4, comprehensive evaluation, determines whether the parameters or data sets of the CKD early screening model need to be adjusted based on the fitting results of the CKD early screening model.

[0015] The beneficial effect of the present invention is that, based on the principles of evidence-based medicine and the mechanism of disease, the present invention statistically analyzes patient case data, and statistically analyzes the data set to select statistically significant patient data indicators as predictors, and uses a machine learning (classification) algorithm model to fit the data set corresponding to the predictor for learning and reasoning testing, thereby completing the establishment of a CDK disease risk prediction model; subsequently, the results of the inference of the test set data according to the early screening model are analyzed to evaluate the effect of the early screening model, and the (basic) model is analyzed for interpretability, and the analysis results are compared with medical principles and expert clinical experience; the reasoning evaluation results of the model and the conclusions of the interpretability analysis are in line with the expectations of the experts. The impact of diagnostic history (history of diabetes, history of hypertension, history of hyperuricemia, history of rheumatic immune system disease, history of urinary system stones) on the risk of CDK can be further studied in the future. Ultimately, based on the current analysis conclusions of the CDK disease risk factors and with reference to the theory of the clinical prediction model, the CDK disease risk prediction and evaluation model can be further optimized to achieve a more accurate early screening function for CDK.

[0016] According to one embodiment of the present invention, step S1 includes: acquiring the sample case data from the LIS system and the HIS system, wherein the sample case data includes the patient's case information and medical consultation information.

[0017] According to one embodiment of the present invention, fitting the data corresponding to the prediction factors using an algorithm model specifically includes:

[0018] Build a data model, update the data record set based on the predictive factors related to the data model, and verify whether the confirmed annotations correspond to the corresponding predictive factor data;

[0019] The algorithm model is used to fit the verified predictor data, and the particle swarm optimization algorithm is used to tune the parameters of each algorithm model;

[0020] Using a stacked generalization approach, the prediction results of multiple basic algorithm models are used as new features, and the predicted dependent variable (whether the patient is diagnosed with CKD) is added to form a new training dataset. A CKD early screening model is then constructed through machine learning and neural network regression model fitting.

[0021] The corresponding data in the test set were substituted into the CKD early screening model for inference, and the inference results were evaluated and analyzed. The model was then interpreted based on the analysis and evaluation of the inference results of the test set data.

[0022] According to one embodiment of the present invention, the CKD early screening model fitting result includes an AUC value and an accuracy rate.

[0023] According to one embodiment of the present invention, determining whether to adjust the machine learning parameters or data set based on the machine learning fitting result specifically includes:

[0024] When the AUC value is less than 75%, it indicates that the CKD early screening model is not suitable for the clinical research and application of CKD early screening;

[0025] When the AUC value is greater than 75%, the fitted training data and validation data and the CKD early screening model are optimized;

[0026] When the AUC reaches the expert's expected value, the accuracy is verified. When the accuracy is less than 85%, the parameters, data set distribution and verification method of the CKD early screening model need to be adjusted. Otherwise, no adjustment is required.

[0027] According to one embodiment of the present invention, the mathematical model includes an artificial neural network, a random forest or a k-nearest neighbor.

[0028] According to one embodiment of the present invention, the predictive factors include: gender, age, BMI, history of hypertension and diabetes, blood creatinine, urea, cystatin, uric acid, urine specific gravity, urine microalbumin, hemoglobin, glucose, red blood cell count, white blood cell count, sodium, history of hyperuricemia, history of rheumatic immune system disease, and history of urinary system stones.

[0029] An early screening model evaluation system for chronic kidney disease according to an embodiment of the present invention includes:

[0030] The data collection and collation module obtains sample case data, preprocesses the sample case data, and constructs a data set;

[0031] The data statistical analysis module conducts statistical analysis on sample case data to obtain characteristic factor variables related to the target factor, i.e., whether the target factor is the risk of CKD. Based on data quality and clinical expert opinions, it screens predictive factors.

[0032] The data fitting analysis module uses the algorithm model to fit the data corresponding to the predictor. Based on the comparative analysis of multiple fitting results of different predictors, the predictor with the best corresponding fitting result is selected as the influencing factor of the model, and the optimal training model is determined. Finally, the model effect is evaluated by plotting the receiver operating characteristic curve and calculating the area under the curve in the model validation set to obtain the fitting results of the CKD early screening model;

[0033] The comprehensive evaluation module determines whether the parameters or data sets of the CKD early screening model need to be adjusted based on the fitting results of the CKD early screening model.

[0034] A computer device according to an embodiment of the present invention includes:

[0035] processor;

[0036] a memory for storing executable instructions;

[0037] Wherein, the processor is used to read the executable instructions from the memory and execute the executable instructions to implement the early screening model evaluation method for chronic kidney disease as described above.

[0038] According to a computer-readable storage medium of an embodiment of the present invention, the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the processor implements the early screening model evaluation method for chronic kidney disease as described above.

[0039] Other features and advantages of the present invention will be described in the following description, and in part will become apparent from the description, or understood by practicing the present invention. The purposes and other advantages of the present invention are realized and obtained by the structures particularly pointed out in the description, claims and drawings.

[0040] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, preferred embodiments are given below and described in detail with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] The present invention will be further described below with reference to the accompanying drawings and examples.

[0042] Figure 1 It is a schematic diagram of the method flow of embodiment 1 of the present invention.

[0043] Figure 2 This is the overall histogram of data in Example 1 of the present invention

[0044] Figure 3 This is the overall analysis of data distribution in the first embodiment of the present invention.

[0045] Figure 4 This is the data density analysis of all indicator factors in the data set of the first embodiment of the present invention.

[0046] Figure 5 This is a heat map of the relationship between various characteristic indicator data in Example 1 of the present invention.

[0047] Figure 6 1 is a schematic diagram of the ROC curve and the area below the line ACU waveform of the chronic kidney disease early screening model according to the first embodiment of the present invention.

[0048] Figure 7 This is a display of the characteristic history factor analysis results of Example 1 of the present invention.

[0049] Figure 8 This is a schematic diagram of the system structure of the second embodiment of the present invention.

[0050] Figure 9 It is a schematic diagram of the device structure of embodiment 3 of the present invention.

[0051] In the figure, 21 is a data collection and organization module; 22 is a data statistical analysis module; 23 is a data statistical analysis module; 24 is a comprehensive evaluation module; 10 is a computer device; 1002 is a processor; 1004 is a memory; and 1006 is a transmission device. DETAILED DESCRIPTION

[0052] The present invention will now be described in further detail with reference to the accompanying drawings, which are simplified schematic diagrams that illustrate the basic structure of the present invention in a schematic manner.

[0053] Example 1

[0054] The present invention provides an early screening model evaluation method for chronic kidney disease. Figure 1 As shown, the method includes the following steps:

[0055] S1, data collection and collation, obtaining sample case data, preprocessing the sample case data, and constructing a data set;

[0056] For example, the sample case data is obtained from the LIS system and the HIS system, and includes the patient's case information and medical information. The patient's case information includes: gender, age, BMI, hypertension, and diabetes; the medical information includes renal function indicators (serum creatinine, urea, cystatin, uric acid, urine specific gravity, urine microalbumin, hemoglobin), laboratory test data (albumin, glucose, red blood cell count, white blood cell count, sodium, potassium, hematocrit), and diagnostic medical history data (such as a history of hyperuricemia, rheumatic immune system diseases, and urinary system stones).

[0057] Furthermore, preprocessing of the sample case data includes: collecting and cleaning the patient's case information and medical information based on time logic (such as the patient's medical information within three months), forming a full data record of the patient within the effective time period, and avoiding interference with the model; in addition, according to the standard definitions of CKD diagnosis and CKD exclusion and various data analysis strategies, case data with good integrity are sorted out and a data set is formed to ensure the accuracy and completeness of the data and avoid interference with the prediction model.

[0058] S2, data statistical analysis, statistical analysis of the data set, obtain characteristic factor variables related to the target factor, that is, whether it is the risk of CKD, and screen out predictive factors based on data quality and clinical expert opinions; among them, predictive factors are screening factors related to CKD diagnosis, including: gender, age, BMI, history of hypertension and diabetes, blood creatinine, urea, cystatin, uric acid, urine specific gravity, urine microalbumin, hemoglobin, glucose, red blood cell count, white blood cell count, sodium, history of hyperuricemia, history of rheumatic immune system diseases, and history of urinary system stones.

[0059] Statistical analysis was performed using SPSS 22.0 statistical analysis software and the statsmodels library in Python. Continuous data were expressed as mean ± standard deviation, and intergroup comparisons were performed using t-tests and analysis of variance. Continuous data were expressed as percentages, and intergroup comparisons were performed using the chi-square test. Significance tests for the characteristic items of the categorical data and the quantitative target (ultrafiltration volume) were performed using t-tests and variance tests. Multivariate statistical analysis and multivariate interaction analysis were used to identify characteristic factor variables associated with the target factor (CKD risk).

[0060] Using univariate statistical analysis, we screened out multiple factors influencing the risk of CKD, selecting statistically significant indicators with a P < 0.05 as model variables. Furthermore, we conducted a multivariate statistical analysis of the key definition of hypotension, namely the blood pressure drop (or the lowest blood pressure value available during dialysis), selecting clinically significant indicators with a P < 0.2 as model influencing factors. The two types of statistical factor indicators and analysis are shown in Table 1:

[0061] Table 1

[0062]

[0063]

[0064] From the above statistical results of CKD indicator factors, we can see that the P values of albumin, potassium, and hematocrit are greater than 0.2, so they are not statistically significant for the CKD early screening model. Considering the actual data missing situation (some indicators are missing), we eliminated the urine specific gravity and glucose indicators with more serious data missing ratios. The final indicators confirmed in the model and the related statistical analysis results are shown in Table 2:

[0065] Table 2

[0066]

[0067]

[0068] The dataset was updated based on the predicted factors screened out. The dataset included 414 confirmed patients, 126 excluded patients, 1207 confirmed records, and 162 excluded records. Figure 2 shows the overall histogram of the data, Figure 3 shows an overall analysis of the data distribution, Figure 4 shows the data density analysis of all indicator factors in the data set, Figure 5 A heat map showing the relationship between various feature indicator data.

[0069] S3, data fitting analysis, use the algorithm model to fit the data corresponding to the predictor, perform comparative analysis based on multiple fitting results of different predictors, select the predictor with the best corresponding fitting result as the influencing factor of the CKD early screening model, and determine the optimal training model as the final CKD early screening model. Finally, in the model validation set, the effect of the model is evaluated by drawing the receiver operating characteristic (ROC) curve and calculating the area under the curve (AUC), and the fitting result of the CKD early screening model is obtained.

[0070] Step S3 specifically includes:

[0071] Build a data model and update the data set based on the predictors associated with the data model. Verify that the confirmed label (0 or 1) corresponds to the corresponding predictor data. Mathematical models include but are not limited to artificial neural networks, random forests, or k-nearest neighbors. Divide the data set into training and test sets.

[0072] The algorithm model is used to fit the verified predictor data (training set data), and the particle swarm optimization algorithm (PSO) is used to tune the parameters of each algorithm model; the algorithm model can be a machine learning model or a (fully connected) neural network model.

[0073] To ensure optimal performance and generalization, a stacking approach was used. The prediction results of multiple algorithm models were used as new features, along with the predicted dependent variable (whether the patient was diagnosed with CKD) to form a new training set. This was then fitted through machine learning and neural network regression models to construct a CKD early screening model.

[0074] Finally, the corresponding data in the test set is substituted into the CKD early screening model for inference, and the inference results are evaluated and analyzed. Based on the analysis and evaluation of the inference results of the test set data, the model is subjected to an explanatory analysis.

[0075] S4, comprehensive evaluation, determines whether the CKD early screening model parameters or data set need to be adjusted based on the CKD early screening model fitting results. The CKD early screening model fitting results include AUC value and accuracy rate.

[0076] Determining whether machine learning parameters or data sets need to be adjusted based on the machine learning fitting results specifically includes:

[0077] When the AUC value is less than 75%, it indicates that the CKD early screening model is not suitable for the clinical research and application of CKD early screening;

[0078] When the AUC value is greater than 75%, the fitted training data and validation data and the CKD early screening model are optimized;

[0079] When the AUC reaches the expert's expected value (such as greater than or equal to 0.8), the accuracy is verified. If the accuracy is less than 85%, the parameters, data set distribution and verification method of the CKD early screening model need to be adjusted. Otherwise, no adjustment is required.

[0080] In this embodiment, due to the statistical caliber or other uncontrollable factors in the processing of step S1 and step S2, the data corresponding to the data model is subjected to repeated dimensionality reduction and dimensionality increase operations using principal component analysis (PCA) before fitting and inference to achieve the purpose of filtering and noise reduction; in addition, in order to improve the nonlinear characterization fitting ability of the model, this embodiment also introduces a clustering strategy for surface splicing to extract nonlinear manifold features, and uses unsupervised learning and k-means based on the Euclidean distance metric to quantize the data vector, ultimately achieving the enhanced nonlinear data characterization capability of the data set.

[0081] Model evaluation and result analysis:

[0082] 1,369 early screening records were used as fitting analysis data, of which the test data set accounted for 30%, namely 958 training data records and 411 test data records. The CKD early screening model was fitted and analyzed on the training data records with an accuracy of 99%; the CKD early screening model was fitted and analyzed on the test data records with an accuracy of 92.49%; the receiver operating characteristic (ROC) curve and the area under the curve (AUC value) were calculated. Figure 6 shown.

[0083] Model analysis:

[0084] This embodiment uses the SHAP (SHapley Additive exPlanations) algorithm as the interpretation technology and method of the CKD early screening model. The importance of each characteristic factor in the model is used to measure the degree of influence on the prediction of confirmed CKD. Generally speaking, not all characteristic factors have a decisive role in the prediction of the results. This embodiment uses the test data set as an example to perform interpretability analysis on multiple current basic models, and selects the basic model with the best effect evaluation as the model interpretation case. Then, the SHAP (Shapley Additive exPlanations) value and related chart visualization analysis methods and tools are combined to evaluate and display the feature importance and feature effect. The characteristic factor analysis results in the case interpretation model are shown as follows Figure 7 :

[0085] like Figure 7As shown in the figure, cystatin C, estimated glomerular filtration rate, and urea indicators have a greater impact on the model, and are the core indicator factors of the basic model. At the same time, attention should be paid to the impact of age and gender on the model. From the above figure, we can see that the age of CKD patients tends to be younger, and the probability of males suffering from CKD is significantly increased compared with females. The larger the value of cystatin C, urea, history of hypertension, white blood cell count, and history of hyperuricemia indicator factors (positive impact), the greater the probability that the model predicts that the result is confirmed CKD. Therefore, when the above indicators are higher than the values of most groups, attention should be paid to whether the patient has the risk of CKD in the subsequent follow-up period. It is worth noting that the indicators of urinary stone history, such as Figure 7 As shown in the data, when patients have a history of urinary stones, the risk of being diagnosed with CKD may be reduced (negative effect); the smaller the estimated glomerular filtration rate index value, the patient's risk of being diagnosed with CKD will significantly increase (estimated glomerular filtration rate is the core indicator of the prediction model), which is consistent with the basic medical theory of CKD.

[0086] Result analysis:

[0087] The above evaluation and analysis results show that the current CKD early screening model predicts the risk of patients diagnosed with CKD. Fitting and inference analysis are performed based on current real patient early screening data records. The prediction results are relatively ideal (AUC greater than 0.85, meeting expert expectations). However, overfitting occurs during the analysis using machine learning because there are too few data cases and the distribution is uneven. The reasoning and generalization ability of the model can be increased by collecting more relevant cases.

[0088] Problems and solutions during implementation: There are too few excluded records, and the proportion is low, resulting in serious data skew; but compared with the previous model (excluding albumin, potassium, hematocrit urine specific gravity, and glucose), the patient data records have increased significantly, enriching the sample size of the data. However, due to the high proportion of confirmed records, the data set has over-sampling of confirmed records, that is, the entire data set has over-sampling relative to the confirmed records. However, the AUC of the entire model is good (greater than 0.85), and it is necessary to increase the number of non-confirmed record data samples in the future, or to counter-generate and exclude CKD record data (mainly laboratory test data).

[0089] In summary, the present invention is based on the principles of evidence-based medicine and the mechanism of disease. By statistically analyzing patient case data and selecting statistically significant patient data indicators as predictors from the data set statistical analysis, the machine learning (classification) algorithm model is used to fit the data set corresponding to the predictor for learning and reasoning testing, and the CDK disease risk prediction model is established. Subsequently, the results of the early screening model are analyzed to evaluate the effect of the early screening model, and the (basic) model is analyzed for interpretability, and the analysis results are compared with medical principles and expert clinical experience. The reasoning evaluation results and interpretability analysis conclusions of the model are in line with the expert expectations. The impact of diagnostic history (diabetes history, hypertension history, hyperuricemia history, rheumatic immune system disease history, urinary system stones history) on the risk of CDK can be further studied. Finally, based on the current analysis conclusions of CDK risk factors and with reference to the theory of clinical prediction models, the CDK disease risk prediction and evaluation model can be further optimized to achieve a more accurate early screening function for CDK. And based on the prediction model reasoning results, a CKD patient hierarchical management system can be established.

[0090] Example 2

[0091] Based on the same inventive concept as the early screening model evaluation method for chronic kidney disease in the aforementioned embodiment, the embodiment of the present application provides an early screening model evaluation system for chronic kidney disease, such as Figure 2 As shown, the system includes:

[0092] The data collection and collation module 21 obtains sample case data, pre-processes the sample case data, and constructs a data set;

[0093] The data statistical analysis module 22 statistically analyzes the sample case data to obtain characteristic factor variables related to the target factor, i.e., whether the target factor is the CKD risk, and screens predictive factors based on data quality and clinical expert opinions;

[0094] The data fitting analysis module 23 uses the algorithm model to fit the data corresponding to the predictor, compares and analyzes multiple fitting results of different predictors, selects the predictor with the best fitting result as the influencing factor of the model, and determines the optimal training model. Finally, the model effect is evaluated by plotting the receiver operating characteristic curve and calculating the area under the curve in the model validation set to obtain the fitting result of the CKD early screening model;

[0095] The comprehensive evaluation module 24 determines whether the parameters or data set of the CKD early screening model need to be adjusted based on the fitting results of the CKD early screening model.

[0096] The foregoing Figure 1The various variations and specific examples of the early screening model evaluation method for chronic kidney disease in Example 1 are also applicable to the early screening model evaluation method device for chronic kidney disease in this embodiment. Through the above detailed description of the early screening model evaluation method for chronic kidney disease, those skilled in the art can clearly know the implementation method of the early screening model evaluation method device for chronic kidney disease in this embodiment, so for the sake of brevity of the specification, it will not be described in detail here.

[0097] Example 3

[0098] An embodiment of the present application provides a computer device, which includes a processor and a memory, wherein the memory stores at least one instruction or at least one program, and the at least one instruction or the at least one program is loaded and executed by the processor to implement an early screening model evaluation method for chronic kidney disease as provided in the above method embodiment.

[0099] Figure 9 The hardware structure diagram of a device for implementing an early screening model evaluation method for chronic kidney disease provided in an embodiment of the present application is shown. The device may participate in or include the apparatus or system provided in an embodiment of the present application. Figure 9 As shown, the computer device 10 may include one or more processors 1002 (the processor may include but is not limited to a microprocessor MCU or a programmable logic device FPGA, etc.), a memory 1004 for storing data, and a transmission device 1006 for communication functions. In addition, it may also include: a display, an input / output interface (I / O interface), a universal serial bus (USB) port (which may be included as one of the ports of the I / O interface), a network interface, a power supply and / or a camera. It will be understood by those skilled in the art that Figure 9 The structure shown is only for illustration and does not limit the structure of the above electronic device. Figure 9 More or fewer components than shown, or with Figure 9 Different configurations shown.

[0100] It should be noted that the one or more processors and / or other data processing circuits described above may generally be referred to herein as "data processing circuitry". The data processing circuitry may be embodied in whole or in part as software, hardware, firmware, or any other combination thereof. In addition, the data processing circuitry may be a single independent processing module, or may be incorporated in whole or in part into any of the other components of the computer device 10 (or mobile device). As described in the embodiments of the present application, the data processing circuitry serves as a processor control (e.g., selection of a variable resistor terminal path connected to an interface).

[0101] The memory 1004 can be used to store software programs and modules of application software, such as the program instructions / data storage device corresponding to the early screening model evaluation method for chronic kidney disease in the embodiment of the present application. The processor executes various functional applications and data processing by running the software programs and modules stored in the memory 1004, that is, implementing one of the above methods. The memory 1004 may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory 1004 may further include a memory remotely located relative to the processor, and these remote memories may be connected to the computer device 10 via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0102] Transmission device 1006 is configured to receive or transmit data via a network. A specific example of the aforementioned network may include a wireless network provided by the communications provider of computer device 10. In one embodiment, transmission device 1006 includes a network interface controller (NIC), which can be connected to other network devices via a base station to enable communication with the Internet. In another embodiment, transmission device 1006 may be a radio frequency (RF) module configured to communicate with the Internet wirelessly.

[0103] The display may be, for example, a touch screen liquid crystal display (LCD) that enables a user to interact with a user interface of the computer device 10 (or mobile device).

[0104] Example 3

[0105] An embodiment of the present application also provides a computer-readable storage medium, which can be set in a server to store at least one instruction or at least one program related to an early screening model evaluation method for chronic kidney disease in a method embodiment. The at least one instruction or the at least one program is loaded and executed by the processor to implement an early screening model evaluation method for chronic kidney disease provided in the above method embodiment.

[0106] Optionally, in this embodiment, the storage medium may be located in at least one of a plurality of network servers in a computer network. Optionally, in this embodiment, the storage medium may include, but is not limited to, various media capable of storing program code, such as a USB flash drive, a read-only memory (ROM), a random access memory (RAM), a mobile hard disk, a magnetic disk, or an optical disk.

[0107] Example 4

[0108] An embodiment of the present invention further provides a computer program product or computer program, comprising computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform a method for evaluating an early screening model for chronic kidney disease provided in any of the aforementioned optional embodiments.

[0109] It should be noted that the order of the embodiments of the present application described above is for descriptive purposes only and does not represent the superiority or inferiority of the embodiments. The above description is of specific embodiments of the present application. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps described in the claims can be performed in an order different from that in the embodiments and still achieve the desired results. In addition, the processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0110] The various embodiments in this application are described in a progressive manner. Similar portions between the various embodiments can be referred to in conjunction with each other. Each embodiment focuses on the differences between the other embodiments. In particular, the device, equipment, and storage medium embodiments are generally similar to the method embodiments, so their descriptions are relatively simple. For relevant portions, refer to the descriptions of the method embodiments.

[0111] Those skilled in the art will understand that all or part of the steps to implement the above embodiments may be accomplished by hardware, or by a program to instruct the relevant hardware, and the program may be stored in a computer-readable storage medium, which may be a read-only memory, a disk, or an optical disk, etc.

[0112] With the above-described preferred embodiments of the present invention as a guide, and with reference to the above description, relevant personnel are fully capable of making various changes and modifications without departing from the technical scope of this invention. The technical scope of this invention is not limited to the contents of the specification and must be determined according to the scope of the claims.

Claims

1. A method for evaluating an early screening model for chronic kidney disease, characterized in that: The method comprises the following steps: S1, data collection and collation, obtaining sample case data, preprocessing the sample case data, and constructing a data set; S2, statistical analysis of data: Statistical analysis of the data set to obtain characteristic factor variables related to the target factor, i.e., whether the risk of CKD is present. Based on data quality and clinical expert opinions, predictive factors are screened; S3, data fitting analysis: use the algorithm model to fit the data corresponding to the predictor, conduct comparative analysis based on multiple fitting results of different predictors, select the predictor with the best fitting result as the influencing factor of the model, and determine the optimal training model. Finally, in the model validation set, the receiver operating characteristic curve is plotted and the area under the curve is calculated to evaluate the model effect, and the fitting results of the CKD early screening model are obtained; S4, comprehensive evaluation, determines whether the parameters or data sets of the CKD early screening model need to be adjusted based on the fitting results of the CKD early screening model.

2. The early screening model evaluation method for chronic kidney disease according to claim 1, characterized in that: The step S1 includes: acquiring the sample case data from the LIS system and the HIS system, wherein the sample case data includes the patient's case information and medical consultation information.

3. The early screening model evaluation method for chronic kidney disease according to claim 1, characterized in that: Use the algorithm model to fit the data corresponding to the predictor, including: Build a data model, update the data record set based on the predictive factors related to the data model, and verify whether the confirmed annotations correspond to the corresponding predictive factor data; The algorithm model is used to fit the verified predictor data, and the particle swarm optimization algorithm is used to tune the parameters of each algorithm model; Using a stacked generalization approach, the prediction results of multiple basic algorithm models are used as new features, and the predicted dependent variable (whether the patient is diagnosed with CKD) is added to form a new training dataset. A CKD early screening model is then constructed through machine learning and neural network regression model fitting. The corresponding data in the test set were substituted into the CKD early screening model for inference, and the inference results were evaluated and analyzed. The model was then interpreted based on the analysis and evaluation of the inference results of the test set data.

4. The early screening model evaluation method for chronic kidney disease according to claim 1, characterized in that: The CKD early screening model fitting results include AUC value and accuracy rate.

5. The early screening model evaluation method for chronic kidney disease according to claim 4, characterized in that: Determining whether machine learning parameters or data sets need to be adjusted based on the machine learning fitting results specifically includes: When the AUC value is less than 75%, it indicates that the CKD early screening model is not suitable for the clinical research and application of CKD early screening; When the AUC value is greater than 75%, the fitted training data and validation data and the CKD early screening model are optimized; When the AUC reaches the expert's expected value, the accuracy is verified. When the accuracy is less than 85%, the parameters, data set distribution and verification method of the CKD early screening model need to be adjusted. Otherwise, no adjustment is required.

6. The early screening model evaluation method for chronic kidney disease according to claim 1, characterized in that: Mathematical models include artificial neural networks, random forests or k-nearest neighbors.

7. The early screening model evaluation method for chronic kidney disease according to claim 1, characterized in that: Predictors included sex, age, BMI, history of hypertension and diabetes, serum creatinine, urea, cystatin, uric acid, urine specific gravity, urine microalbumin, hemoglobin, glucose, red blood cell count, white blood cell count, sodium, history of hyperuricemia, history of rheumatic immune system disease, and history of urinary stones.

8. An early screening model evaluation system for chronic kidney disease, characterized in that: The system comprises: A data collection and collation module (21) obtains sample case data, pre-processes the sample case data, and constructs a data set; The data statistical analysis module (22) statistically analyzes the sample case data to obtain characteristic factor variables related to the target factor, i.e., whether it is the risk of CKD, and screens predictive factors based on data quality and clinical expert opinions; The data fitting analysis module (23) uses the algorithm model to fit the data corresponding to the predictor, conducts comparative analysis based on multiple fitting results of different predictors, selects the predictor with the best corresponding fitting result as the influencing factor of the model, and determines the optimal training model. Finally, in the model validation set, the effect of the model is evaluated by drawing the receiver operating characteristic curve and calculating the area under the curve to obtain the fitting result of the CKD early screening model; The comprehensive evaluation module (24) determines whether the parameters or data sets of the CKD early screening model need to be adjusted according to the fitting results of the CKD early screening model.

9. A computer device, characterized in that: include: processor; a memory for storing executable instructions; Wherein, the processor is used to read the executable instructions from the memory and execute the executable instructions to implement the early screening model evaluation method for chronic kidney disease as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the processor implements the early screening model evaluation method for chronic kidney disease according to any one of claims 1 to 7.

Images