Diabetic nephropathy early warning method, system, equipment and medium

Abstract

The invention provides a diabetic nephropathy early warning method, system, equipment and medium, and relates to the technical field of medical data analysis and prediction.The method comprises the steps that multiple urine samples of a user are collected, target urine protein detection is conducted, and multiple urine protein concentrations are obtained; each urine protein concentration is combined with a normal urine protein concentration set to be input into an abnormal data recognition tree, the number of layers of multiple recognition nodes is obtained, and the detection confidence coefficient is obtained through calculation; configuring an abnormal proportion according to the detection confidence coefficient, inputting a plurality of urine protein concentration combination normal urine protein concentration sets into an abnormal data identification tree, deleting urine protein concentrations divided into abnormal data in the abnormal proportion before, and obtaining screened urine protein concentrations; according to the screened urine protein concentration, an auxiliary early warning coefficient is obtained through calculation, and early warning is carried out. The technical problem of low early warning accuracy of diabetic nephropathy in the prior art is solved.

Description

Technical Field

[0001] This invention relates to the field of medical data analysis and prediction, and in particular to a method, system, device and medium for early warning of diabetic nephropathy. Background Art

[0002] Detecting urinary protein concentration can provide early warning of diabetic nephropathy. However, current technologies suffer from low accuracy due to the susceptibility of single-sample tests to physiological fluctuations and detection errors. For example, a short-term high-protein diet or strenuous exercise can cause momentary abnormalities in urinary protein concentration, leading to false positives. Furthermore, factors such as instrument precision errors and non-standard operating procedures can further amplify data deviations. The combined effect of these accidental factors results in low accuracy in predicting diabetic nephropathy. Summary of the Invention

[0003] This invention addresses the technical problem of low accuracy in early warning of diabetic nephropathy in existing technologies by providing a method, system, device, and medium for early warning of diabetic nephropathy.

[0004] The technical solution of the present invention to solve the above-mentioned technical problems is as follows:

[0005] In a first aspect, the present invention provides a method for early warning of diabetic nephropathy, comprising:

[0006] Multiple urine samples were collected from the user to detect the target urinary protein and obtain multiple urinary protein concentrations;

[0007] Each urine protein concentration is combined with a normal urine protein concentration set and input into an abnormal data identification tree to obtain multiple identification node layers, and the detection confidence is calculated.

[0008] Based on the detection confidence level, configure the abnormal proportion, input the multiple urine protein concentrations into the normal urine protein concentration set into the abnormal data identification tree, delete the urine protein concentrations that were classified as abnormal data before the abnormal proportion, and obtain the screened urine protein concentrations.

[0009] Based on the screened urinary protein concentration, an auxiliary early warning coefficient is calculated and an early warning is issued.

[0010] In a second aspect, the present invention provides a diabetic nephropathy early warning system, comprising:

[0011] The data acquisition module is used to collect multiple urine samples from the user, perform target urine protein detection, and obtain multiple urine protein concentrations;

[0012] The anomaly identification module is used to combine each urine protein concentration with the normal urine protein concentration set and input it into the anomaly data identification tree to obtain multiple identification node layers and calculate the detection confidence.

[0013] The concentration screening module is used to configure the abnormal proportion according to the detection confidence level, input the multiple urine protein concentrations into the normal urine protein concentration set into the abnormal data identification tree, delete the urine protein concentrations that were classified as abnormal data before the abnormal proportion, and obtain the screened urine protein concentrations.

[0014] The early warning output module is used to calculate the auxiliary early warning coefficient based on the screened urine protein concentration and to issue an early warning.

[0015] Thirdly, the present invention provides an electronic device, comprising:

[0016] Memory, used to store computer software programs;

[0017] A processor is used to read and execute the computer software program, thereby implementing a method for early warning of diabetic nephropathy as described in the first aspect.

[0018] Fourthly, the present invention provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements a method for early warning of diabetic nephropathy as described in the first aspect.

[0019] The beneficial effects of the present invention are:

[0020] Compared to existing technologies, this application first collects multiple urine samples from users for target urinary protein detection, obtaining multiple urinary protein concentrations. Multiple Klotho protein concentrations are obtained through multi-sample detection, effectively reducing the impact of sample randomness and detection errors, providing a reliable data foundation for subsequent accurate early warning of diabetic nephropathy. Secondly, each urinary protein concentration is combined with a set of normal urinary protein concentrations and input into an anomaly identification tree to obtain multiple identification node layers. The detection confidence score is calculated, and the anomaly identification tree constructed using the isolated forest algorithm achieves adaptive identification of abnormal urinary protein data. The reliability of the detection data is assessed using the detection confidence score, providing reliable support for accurate early warning of diabetic nephropathy. Thirdly, based on the detection confidence score, an anomaly ratio is configured. Multiple urinary protein concentrations are combined with a set of normal urinary protein concentrations and input into the anomaly identification tree. The urinary protein concentrations that were previously classified as abnormal before deletion are then selected, obtaining the filtered urinary protein concentrations, providing a reliable data foundation for subsequent early warning of diabetic nephropathy. Finally, based on the screening of urinary protein concentration, an auxiliary early warning coefficient is calculated and an early warning is issued. The entire process, through rigorous calculations and scientific threshold settings, achieves efficient transformation from data processing to risk warning, thereby improving the accuracy of early warning for diabetic nephropathy.

[0021] Through the above technical solution, this application collects multiple urine samples from users, detects multiple urine protein concentrations, identifies abnormal data through an anomaly identification tree, obtains confidence levels for multiple urine protein concentrations, configures an anomaly ratio based on the confidence levels, and deletes some abnormal data accordingly to obtain a screening urine protein concentration that better reflects the user's true kidney status. Finally, it makes a judgment and issue an early warning based on this. In this way, it effectively eliminates abnormal data caused by sample randomness and detection errors, and improves the accuracy of early warning for diabetic nephropathy. Attached Figure Description

[0022] Figure 1 This is a flowchart illustrating a method for early warning of diabetic nephropathy provided by the present invention.

[0023] Figure 2 This is a schematic diagram of the structure of a diabetic nephropathy early warning system provided by the present invention;

[0024] Figure 3 This is a schematic diagram of the structure of the electronic device provided by the present invention.

[0025] In the attached diagram, the components represented by each number are as follows:

[0026] Data acquisition module 11, anomaly identification module 12, concentration screening module 13, early warning output module 14, electronic device 200, memory 210, processor 220, computer program 211. Detailed Implementation

[0027] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0028] In the description of this invention, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the stated features. In the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0029] In the description of this invention, the term "for example" is used to mean "used as an example, illustration, or description." Any embodiment described as "for example" in this invention is not necessarily to be construed as being more preferred or advantageous than other embodiments. The following description is provided to enable any person skilled in the art to make and use the invention. Details are set forth in the following description for purposes of explanation. It should be understood that those skilled in the art will recognize that the invention can be made without using these specific details. In other instances, well-known structures and processes will not be described in detail to avoid obscuring the description of the invention with unnecessary detail. Therefore, the invention is not intended to be limited to the embodiments shown, but is consistent with the broadest scope of the principles and features disclosed herein.

[0030] Example 1, as Figure 1 As shown, this embodiment of the invention provides a method for early warning of diabetic nephropathy, including:

[0031] S10: Collect multiple urine samples from the user, perform target urine protein detection, and obtain multiple urine protein concentrations;

[0032] In existing technologies, diabetic nephropathy is predicted by collecting a single urine sample and detecting urine protein concentration. However, the composition of human urine fluctuates daily due to factors such as diet, exercise, and lifestyle. For example, a single urine sample may show protein concentration concentration due to insufficient water intake, or transient proteinuria may be triggered by strenuous exercise, leading to false positive results. Furthermore, in the early stages of diabetic nephropathy, urine protein concentration may be intermittently abnormal, and a single sample may easily miss early, subtle abnormalities.

[0033] To address the aforementioned issues, this application collects multiple urine samples from users, performs target urinary protein detection, and obtains multiple urinary protein concentrations.

[0034] Specifically, step S10 in the method includes:

[0035] Collect multiple urine samples from the user;

[0036] Multiple urine samples were subjected to target urinary protein concentration detection to obtain multiple urinary protein concentrations, among which the urinary protein concentration was Klotho protein.

[0037] In this embodiment, multiple urine samples are first collected from the user. For example, the user's morning urine or random urine can be collected for 3-5 consecutive days, or multiple samples can be collected on the same day under different physiological conditions (such as fasting or post-meal) to obtain multiple urine samples. After collection, the samples need to be refrigerated (e.g., 4°C) or preservatives (such as toluene) should be added in a timely manner to avoid protein degradation in the urine samples. In this way, the interference of accidental factors can be reduced and the accuracy of the test results can be ensured.

[0038] Secondly, the target urinary protein concentration was measured in multiple urine samples to obtain multiple urinary protein concentrations. Among these, the urinary protein concentration was for Klotho protein. Klotho protein is a transmembrane glycoprotein secreted by the proximal tubular epithelial cells of the kidney. It can detach and enter the urine. Klotho protein exerts a renal protective function by inhibiting oxidative stress, inflammatory responses, and renal tubular epithelial cell fibrosis. In diabetes, factors such as hyperglycemia and insulin resistance inhibit the expression of Klotho in the kidneys, leading to a decrease in urinary Klotho concentration. Furthermore, the urinary Klotho concentration begins to decline before the appearance of urinary microalbumin in diabetic patients. For example, a clinical study showed that the urinary Klotho concentration in patients with stage G1 diabetic nephropathy (normal glomerular filtration rate) was about 30% lower than that in healthy individuals. Therefore, Klotho protein can be used as an earlier warning indicator of kidney disease. For example, Klotho protein detection can be achieved by the following methods: enzyme-linked immunosorbent assay (ELISA), which utilizes specific antibodies to bind to Klotho protein and quantifies it through enzymatic reaction with color development, achieving a detection sensitivity at the pg / mL level; or mass spectrometry (MS), which provides high-resolution detection and can simultaneously analyze multiple proteins.

[0039] For example, after preprocessing multiple urine samples from a user (such as centrifugation, dilution, etc.), the concentration of Klotho protein is detected by enzyme-linked immunosorbent assay (ELISA): Klotho-specific antibody is added, an antigen-antibody reaction is performed, colorimetric quantification is achieved through enzymatic reaction, and the concentration is converted using a standard curve to output the concentration of multiple urine proteins from the user.

[0040] In summary, compared to existing technologies, this application collects multiple urine samples from users and performs target urinary protein detection to obtain multiple urinary protein concentrations. Thus, obtaining multiple Klotho protein concentrations through multi-sample detection effectively reduces the impact of sample randomness and detection errors, providing a reliable data foundation for subsequent accurate early warning of diabetic nephropathy.

[0041] S20: Combine each urine protein concentration with the normal urine protein concentration set and input them into the abnormal data identification tree to obtain multiple identification node layers, and calculate the detection confidence.

[0042] Because the composition of human urine is affected by factors such as diet, exercise, and rest, and because the urine protein concentration detection process may have detection errors, there may be abnormal values ​​among the multiple urine protein concentrations obtained. Therefore, using urine protein concentration to predict diabetic nephropathy may result in a large error.

[0043] To address the aforementioned issues, this application combines each urinary protein concentration with a set of normal urinary protein concentrations and inputs them into an abnormal data identification tree to obtain multiple identification node layers and calculate the detection confidence level.

[0044] Specifically, step S20 in the method includes:

[0045] Based on the set of sample urinary protein concentrations, an anomaly identification tree is constructed. The anomaly identification tree includes multiple layers of classification nodes. Each layer of classification nodes performs binary classification on the input urinary protein concentration. Urinary protein concentrations classified as isolated data are considered anomaly data.

[0046] Each urine protein concentration is combined with a set of normal urine protein concentrations and input into the abnormal data identification tree to obtain the number of classification node layers for each urine protein concentration as abnormal data. Multiple identification node layers are obtained. If the urine protein concentration is not classified as abnormal data, the maximum number of classification node layers is taken as the identification node layer.

[0047] The detection confidence level is calculated based on the number of recognition node layers.

[0048] In this embodiment, an anomaly identification tree is first constructed based on a set of sample urinary protein concentrations. This anomaly identification tree includes multiple layers of classification nodes. Each layer performs binary classification on the input urinary protein concentrations, and the urinary protein concentrations classified as isolated data are considered anomaly data. Specifically, the anomaly identification tree is constructed based on the Isolation Forest algorithm and includes multiple layers of classification nodes. Each layer contains a random sample urinary protein concentration (e.g., 25 ng / mL) from historical testing data, which serves as the classification threshold for that layer. The input urinary protein concentrations are categorized into two classes: those greater than or equal to the classification threshold and those less than the threshold. These are stored in the left and right child nodes respectively, achieving binary classification of the input data. Based on the characteristic that anomaly data is more easily "isolated" in the data space, data in child nodes with very small sample sizes (e.g., <5%) or significantly fewer samples than the other side (e.g., the left node contains 10% of the data, and the right node contains 90%) are marked as isolated data. The urinary protein concentrations classified as isolated data by each layer are considered anomaly data.

[0049] Secondly, each urinary protein concentration is combined with a set of normal urinary protein concentrations (such as historical test data from healthy individuals) and input into the abnormal data identification tree. The number of classification node layers at which each urinary protein concentration is classified as abnormal data is obtained, resulting in multiple identification node layers. If a urinary protein concentration is not classified as abnormal data, the highest classification node layer is used as the identification node layer. Specifically, each urinary protein concentration is combined with a set of normal urinary protein concentrations. During the classification process at each level of the abnormal data identification tree, normal urinary protein concentrations are clustered into one side of the child node, while abnormal urinary protein concentrations are classified into the other side of the child node, forming isolated abnormal data points. This achieves multi-level binary classification, and urinary protein concentrations classified as isolated data are considered abnormal data.

[0050] The method of inputting a user's urine protein concentration into the tree model by combining it separately with a normal concentration set ensures that each node is referenced to normal data. Further, the number of classification nodes for each urine protein concentration classified as abnormal data is obtained, resulting in multiple identification node layers. For example, a urine protein concentration of 30 ng / mL is classified as abnormal data at layer 5. The depth at which a urine protein concentration is classified in the abnormal data identification tree (i.e., the identification node layer) is closely related to the degree of abnormality. If a urine protein concentration is classified as abnormal data at a shallow node (e.g., layer 1), it indicates a large deviation from the normal distribution and a high abnormality score. Conversely, if it is isolated at a deep node (e.g., layer 20), it indicates that it is close to the normal distribution and has a low abnormality score. For urine protein concentrations that are never classified as abnormal data, the maximum number of classification nodes in the abnormal data identification tree is used as their identification node layer. For example, a urine protein concentration of 5 ng / mL that is never classified as abnormal data in a 20-layer abnormal data identification tree is used as its identification node layer.

[0051] For example, a set of normal urine protein concentrations (e.g., 58 ng / mL) is combined with a set of normal urine protein concentrations from 100 samples and input into an anomaly identification tree. During the classification process at the fourth-level classification node, the 100 normal urine protein concentrations are assigned to the right child node, while the urine protein concentration is assigned to the left child node, forming an isolated anomaly data point. In this case, the identification node level for the urine protein concentration is 4. This step is repeated, and each urine protein concentration is combined with the normal urine protein concentration set and input separately. The output yields multiple identification node levels (e.g., 4, 5, 5, 6, 6, 6, 6, 7, 7, 8).

[0052] Finally, the detection confidence level is calculated based on the number of identification node layers. Specifically, the absolute difference between the maximum and minimum values ​​within multiple identification node layers is calculated, and then the ratio of the absolute difference to the maximum value is calculated to obtain the detection confidence level. The closer the number of identification node layers, the smaller the dispersion of the detected urinary protein concentrations, the more stable the data, and the higher the detection confidence level. Detection confidence level quantifies the reliability of the detection data. High confidence level indicates concentrated data distribution and high reliability of the detection results, while low confidence level indicates large data dispersion and potential detection errors.

[0053] Furthermore, the "construction of an anomaly identification tree based on the set of sample urinary protein concentrations" includes:

[0054] Based on historical data of the target urinary protein, a set of urinary protein concentrations from the samples was collected.

[0055] Randomly select the first sample urine protein concentration from the set of sample urine protein concentrations to construct a first-level classification node. The first-level classification node performs binary classification on the input urine protein concentration to obtain two categories: greater than or equal to or less than the first sample urine protein concentration. The results are then input into the previous-level classification node for further binary classification.

[0056] Continue to construct multi-level classification nodes to obtain an anomaly data identification tree, where the urine protein concentration classified as isolated data in each classification node is considered anomaly data.

[0057] In this embodiment, an anomaly identification tree is constructed based on the Isolation Forest algorithm. Isolation Forest is an unsupervised anomaly detection algorithm that leverages the characteristic that anomalous data is more easily "isolated" in the data space. It recursively segments the data by randomly selecting features and thresholds. In Isolation Forest, the depth of sample segmentation (number of node layers) is inversely proportional to the degree of anomaly. Shallow segmentation (e.g., layer 2): samples are quickly isolated, indicating a significant difference from the normal distribution and a high anomaly score. Deep segmentation (e.g., layer 8): samples require multiple segmentations to be isolated, indicating a close approximation of the normal distribution and a low anomaly score. The anomaly identification tree is a binary decision tree constructed based on the Isolation Forest algorithm. Each internal node contains a classification threshold (e.g., "protein concentration 25 ng / mL"), dividing the data into a left subtree (< threshold) and a right subtree (≥ threshold), storing them in the corresponding leaf nodes. If the sample size within a leaf node is extremely small (e.g., <10%), the sample at that node is considered isolated data. Further, the specific construction process of the anomaly identification tree is as follows:

[0058] First, a set of sample urinary protein concentrations is collected based on historical detection data of the target urinary protein. Specifically, based on historical urinary protein data from a large number of diabetic patients (e.g., 1000 cases), including samples with normal, early-stage, and late-stage lesions, ensuring data coverage of the entire disease course, a subset (e.g., 200 samples) is randomly selected from the historical data as the sample urinary protein concentration set. Based on this, an anomaly identification tree is constructed to reduce computational complexity and enhance model robustness.

[0059] Secondly, a first sample urine protein concentration is randomly selected from the set of sample urine protein concentrations to construct a first-level classification node. This first-level classification node performs binary classification on the input urine protein concentration, resulting in two classes: one greater than or equal to the first sample urine protein concentration, and the result is input into the next-level classification node for further binary classification. Randomly selecting the first sample urine protein concentration from the set of sample urine protein concentrations is to improve the model's robustness and generalization ability. For example, a urine protein concentration (e.g., 25 ng / mL) is randomly selected from the set of sample urine protein concentrations as the first sample urine protein concentration to construct the first-level classification node. The classification process uses the first sample urine protein concentration as a classification threshold. If the input urine protein concentration is greater than or equal to the first sample urine protein concentration, it is classified into the right child node; if the input urine protein concentration is less than the first sample urine protein concentration, it is classified into the left child node. Then, the result is input into the next-level classification node for further binary classification.

[0060] Finally, a multi-level classification node is constructed to obtain an anomaly identification tree. In this tree, the urinary protein concentration classified as isolated data by each classification node is considered anomaly data. Specifically, the random selection of sample urinary protein concentrations within the set of sample urinary protein concentrations in the above steps is repeated as the classification node. This process continues to construct multiple levels of classification nodes, for example, up to 20 levels, to obtain the anomaly identification tree. Based on the characteristic that anomaly data is more easily "isolated" in the data space, the anomaly identification tree marks data in leaf nodes with very small sample sizes (e.g., <5%) at each level as isolated data, and the urinary protein concentration classified as isolated data by each classification node is considered anomaly data.

[0061] Furthermore, the phrase "calculating the detection confidence based on the number of multiple identification node layers" includes:

[0062] Calculate the absolute difference between the maximum and minimum values ​​within the multiple identification node layers;

[0063] Calculate the ratio of the absolute difference to the maximum value, and then calculate the detection confidence level.

[0064] In this embodiment, the absolute difference between the maximum and minimum values ​​within the multiple identification node layers is first calculated. For example, the multiple identification node layers are (e.g., 4, 5, 5, 6, 6, 6, 6, 7, 7, 8), where the maximum and minimum identification node layers are 8 and 4 respectively. The absolute difference between them is calculated as |8 - 4| = 4. The absolute difference reflects the dispersion of the multiple identification node layers. The smaller the absolute difference, the more concentrated the data, meaning the detection data is more stable and reliable.

[0065] Next, the ratio of the absolute difference to the maximum value is calculated, and the detection confidence level is obtained, where the detection confidence level = 1 - absolute difference / maximum value. For example, if the absolute difference is 4 and the maximum value is 8, then the detection confidence level = 1 - 4 / 8 = 0.5. The smaller the absolute difference, the more concentrated the data, and the higher the calculated detection confidence level.

[0066] In summary, compared with existing technologies, this application combines each urinary protein concentration with a normal urinary protein concentration set and inputs it into an abnormal data identification tree to obtain multiple identification node layers and calculate the detection confidence. In this way, the abnormal data identification tree constructed by the isolated forest algorithm achieves adaptive identification of abnormal urinary protein data, and the reliability of the detection data is evaluated by the detection confidence, providing reliable support for accurate early warning of diabetic nephropathy.

[0067] S30: Configure the abnormal proportion according to the detection confidence level, input the multiple urine protein concentrations into the normal urine protein concentration set into the abnormal data identification tree, delete the urine protein concentrations that were classified as abnormal data before the abnormal proportion, and obtain the screened urine protein concentrations.

[0068] Because of sample randomness and detection errors, outliers may exist in the multiple urine protein data obtained from the test. Therefore, some abnormal data should be deleted when using urine protein to predict diabetic nephropathy, in order to improve the accuracy and reliability of the prediction.

[0069] To address the aforementioned issues, this application configures the abnormality ratio based on the detection confidence level, inputs multiple urine protein concentrations into a normal urine protein concentration set into an abnormal data identification tree, deletes the urine protein concentrations that were previously classified as abnormal data from the abnormality ratio, and obtains the screened urine protein concentrations.

[0070] Specifically, step S30 in the method includes:

[0071] The detection confidence level is input into the abnormality ratio classification library to obtain the abnormality ratio. The abnormality ratio classification library is constructed by mapping the sample detection confidence level set of sample users to the sample abnormality ratio set of urine protein concentration. The detection confidence level and the abnormality ratio are negatively correlated.

[0072] The multiple urine protein concentration combinations are input into the abnormal data identification tree to obtain all urine protein concentrations that are classified as isolated data, and sorted in ascending order of classification node layer number to obtain the abnormal data sequence.

[0073] Delete the urinary protein concentrations with the highest proportion of abnormality in the abnormal data sequence, and retain the other urinary protein concentrations to obtain the screened urinary protein concentrations.

[0074] In this embodiment, the detection confidence level is first input into the anomaly ratio classification library to obtain the anomaly ratio. The anomaly ratio classification library is constructed using the mapping relationship between the sample user's sample detection confidence level set and the sample anomaly ratio set of urine protein concentration. The detection confidence level and the anomaly ratio are negatively correlated; that is, the higher the confidence level, the stronger the data reliability, and the lower the proportion of abnormal data to be deleted. Conversely, the lower the confidence level, the higher the data dispersion, and the more abnormal data to be deleted to ensure the reliability of the filtered data. Furthermore, the abnormality proportion classification library can be constructed based on the mapping relationship between the detection confidence level and the abnormality proportion of historical data. For example, a large amount of urine protein detection data of diabetic patients can be collected, the detection confidence level of each group of data can be calculated, and combined with the clinical diagnosis results, the proportion of abnormal values ​​in each group of data can be marked (such as confirmed by kidney biopsy or long-term follow-up). For example, a detection confidence level of 0.8 corresponds to an abnormality proportion of 10%, a detection confidence level of 0.7 corresponds to an abnormality proportion of 15%, a detection confidence level of 0.6 corresponds to an abnormality proportion of 20%, a detection confidence level of 0.5 corresponds to an abnormality proportion of 25%, and so on. In this way, a mapping model between confidence level and the actual abnormality proportion can be established.

[0075] For example, the detection confidence level (e.g., 0.5) is input into the anomaly ratio classification library to obtain the anomaly ratio (e.g., 25%). The anomaly ratio represents the proportion of abnormal data that needs to be deleted among the multiple urinary protein concentrations obtained from the detection. The higher the anomaly ratio, the higher the proportion of abnormal data that needs to be deleted.

[0076] Secondly, the multiple urine protein concentrations combined with the normal urine protein concentration set are input into an anomaly identification tree to obtain all urine protein concentrations classified as isolated data. These are then sorted in ascending order of the number of classification node layers to obtain an anomaly data sequence. Specifically, multiple urine protein concentrations of the user and a normal urine protein concentration set (such as historical test data of healthy individuals) are input into an anomaly identification tree constructed based on the isolated forest algorithm. Using normal data as a reference, outlier isolated data in the user data is identified, and all urine protein concentrations classified as isolated data are obtained. These are then sorted in ascending order of the number of classification node layers to obtain an anomaly data sequence. For example, based on the isolated data identified by the anomaly identification tree and sorted in ascending order of the number of classification node layers, the following anomaly data sequence is obtained: node layer 2 (5 ng / mL), node layer 4 (12 ng / mL), node layer 6 (35 ng / mL), and node layer 8 (36 ng / mL). The smaller the layer number, the earlier the data was classified as isolated, indicating a higher degree of abnormality and requiring priority deletion.

[0077] Finally, the urine protein concentrations with the highest abnormal proportion within the abnormal data sequence are deleted, while the remaining urine protein concentrations are retained to obtain the screening urine protein concentrations. Specifically, based on the abnormal proportion (e.g., X%), the top X% of urine protein concentrations in the abnormal data sequence are deleted, and the remaining urine protein concentrations are retained to obtain the screening urine protein concentrations. For example, if there are 10 abnormal data points in the abnormal data sequence, with an abnormal proportion of 30%, then the top 3 urine protein concentrations in the abnormal data sequence (i.e., the 3 with the lowest layer number) are deleted, and the remaining undeleted urine protein concentrations constitute the screening urine protein concentrations. These data are considered to be more stable and reliable data that are closer to the true state and can be used for subsequent early warning of diabetic nephropathy.

[0078] In summary, compared to existing technologies, this application configures the proportion of potentially abnormal data (abnormal data due to sample randomness and detection errors) in the user's urine protein concentration detection based on the detection confidence level. The higher the confidence level, the smaller the proportion of abnormal data. Then, multiple urine protein concentrations are sorted by abnormality. The urine protein concentrations with the abnormal proportion before deletion are classified as abnormal data, and the remaining urine protein concentrations that are not deleted are used as screening urine protein concentrations, i.e., stable normal data, providing a reliable data foundation for subsequent early warning of diabetic nephropathy.

[0079] S40: Calculate the auxiliary early warning coefficient based on the screened urinary protein concentration, and issue an early warning.

[0080] After the aforementioned steps of data collection, abnormal data identification and screening, the obtained screening urine protein concentration has eliminated abnormal data caused by factors such as sample randomness and detection error. Therefore, diabetic nephropathy can be predicted based on the screening urine protein concentration.

[0081] To address the aforementioned issues, this application calculates an auxiliary early warning coefficient based on the screening of urinary protein concentration to provide early warning of diabetic nephropathy.

[0082] Obtain standard urine protein concentration;

[0083] The range by which the screened urinary protein concentration exceeds the standard urinary protein concentration is calculated and the average value is calculated to obtain an auxiliary warning coefficient for discrimination and warning.

[0084] In this embodiment, a standard urinary protein concentration is first obtained. Specifically, the standard urinary protein concentration is usually derived from statistical analysis of test data from a large number of healthy individuals or determined according to clinical treatment guidelines. It serves as an important reference benchmark for determining whether urinary protein concentration is abnormal. For example, for Klotho protein, the normal concentration reference range may be 15-30 ng / mL, which reflects the concentration range of this protein in human urine under healthy conditions.

[0085] Secondly, the range by which the screened urinary protein concentration exceeds the standard urinary protein concentration is calculated and the average is calculated to obtain an auxiliary warning coefficient for judgment and warning. Specifically, after the aforementioned steps of data collection, abnormal data identification and screening, the obtained screened urinary protein concentrations have eliminated abnormal data caused by factors such as sample randomness and detection errors, and can better reflect the user's true kidney status. By calculating the range by which each screened urinary protein concentration exceeds the standard urinary protein concentration, for example, if the upper limit of the standard urinary protein concentration is 30 ng / mL, and a certain screened urinary protein concentration is 35 ng / mL, then the deviation range is 5 ng / mL. By iterating through the screened urinary protein concentrations, multiple deviation ranges are calculated, for example, 3.5 ng / mL, 2.1 ng / mL, 3.9 ng / mL, and 2.5 ng / mL. Then, the average of multiple deviation ranges is calculated, for example, (5 + 3.5 + 2.1 + 3.9 + 2.5) / 5 = 3.4 ng / mL, to obtain the auxiliary warning coefficient. The auxiliary warning coefficient integrates the abnormality of multiple screened urinary protein concentrations, and quantifies the degree to which the patient's urinary protein concentration deviates from the normal range as a whole.

[0086] Finally, an auxiliary warning coefficient is used for judgment and warning. Specifically, when the auxiliary warning coefficient exceeds a preset threshold, it indicates that the patient's urine protein concentration is persistently abnormal, suggesting a risk of diabetic nephropathy, requiring timely further examination or intervention. When the auxiliary warning coefficient is within the threshold range, it indicates that the current kidney condition is relatively stable. Based on extensive clinical data, this application recommends a preset threshold of 5 ng / mL, which can be adjusted by those skilled in the art according to actual circumstances. For example, if the threshold is set to 5 ng / mL, an warning is triggered when the auxiliary warning coefficient reaches 6 ng / mL, reminding doctors and patients to pay attention to kidney health and gain valuable time for early diagnosis and treatment.

[0087] In summary, compared with existing technologies, this application calculates an auxiliary early warning coefficient based on the screening of urinary protein concentration to provide early warning of diabetic nephropathy. In this way, the entire process, through rigorous calculation and scientific threshold setting, achieves efficient transformation from data processing to risk warning, thereby improving the accuracy of early warning of diabetic nephropathy.

[0088] In summary, the embodiments of this application have at least the following technical effects:

[0089] Compared to existing technologies, this application first collects multiple urine samples from the user, performs target urinary protein detection, and obtains multiple urinary protein concentrations. In this way, by obtaining multiple Klotho protein concentrations through multi-sample detection, the impact of sample randomness and detection errors is effectively reduced, providing a reliable data foundation for subsequent accurate early warning of diabetic nephropathy.

[0090] Secondly, this application combines each urinary protein concentration with a normal urinary protein concentration set and inputs it into an abnormal data identification tree to obtain multiple identification node layers and calculate the detection confidence. In this way, the abnormal data identification tree constructed by the isolated forest algorithm realizes adaptive identification of abnormal urinary protein data, and the reliability of the detection data is evaluated by the detection confidence, providing reliable support for accurate early warning of diabetic nephropathy.

[0091] Furthermore, this application configures the proportion of potentially abnormal data (abnormal data due to sample randomness and detection error) in the user's urine protein concentration detection based on the detection confidence level. The higher the confidence level, the smaller the proportion of abnormal data. Then, multiple urine protein concentrations are sorted by abnormality. The urine protein concentrations with the abnormal proportion before deletion are classified as abnormal data, and the remaining urine protein concentrations that are not deleted are used as screening urine protein concentrations, i.e., stable normal data, providing a reliable data foundation for subsequent early warning of diabetic nephropathy.

[0092] Finally, this application calculates an auxiliary early warning coefficient based on the screening urinary protein concentration to provide early warning of diabetic nephropathy. In this way, the entire process, through rigorous calculation and scientific threshold setting, achieves efficient transformation from data processing to risk warning, thereby improving the accuracy of early warning of diabetic nephropathy.

[0093] Through the above technical solution, this application collects multiple urine samples from users, detects multiple urine protein concentrations, identifies abnormal data through an anomaly identification tree, obtains confidence levels for multiple urine protein concentrations, configures an anomaly ratio based on the confidence levels, and deletes some abnormal data accordingly to obtain a screening urine protein concentration that better reflects the user's true kidney status. Finally, it makes a judgment and issue an early warning based on this. In this way, it effectively eliminates abnormal data caused by sample randomness and detection errors, and improves the accuracy of early warning for diabetic nephropathy.

[0094] Example 2, as Figure 2 As shown, based on the same inventive concept as the diabetic nephropathy early warning method provided in Embodiment 1, this embodiment of the invention also provides a diabetic nephropathy early warning system, including:

[0095] Data acquisition module 11 is used to collect multiple urine samples from the user, perform target urine protein detection, and obtain multiple urine protein concentrations;

[0096] The anomaly identification module 12 is used to combine each urine protein concentration with the normal urine protein concentration set and input it into the anomaly data identification tree to obtain multiple identification node layers and calculate the detection confidence.

[0097] Concentration screening module 13 is used to configure the abnormal proportion according to the detection confidence level, input the multiple urine protein concentrations into the normal urine protein concentration set into the abnormal data identification tree, delete the urine protein concentrations that were classified as abnormal data before the abnormal proportion, and obtain the screened urine protein concentrations.

[0098] The early warning output module 14 is used to calculate the auxiliary early warning coefficient based on the screened urine protein concentration and to issue an early warning.

[0099] Specifically, the data acquisition module 11 is used for:

[0100] Collect multiple urine samples from the user;

[0101] Multiple urine samples were subjected to target urinary protein concentration detection to obtain multiple urinary protein concentrations, among which the urinary protein concentration was Klotho protein.

[0102] The anomaly detection module 12 is specifically used for:

[0103] Based on the set of sample urinary protein concentrations, an anomaly identification tree is constructed. The anomaly identification tree includes multiple layers of classification nodes. Each layer of classification nodes performs binary classification on the input urinary protein concentration. Urinary protein concentrations classified as isolated data are considered anomaly data.

[0104] Each urine protein concentration is combined with a set of normal urine protein concentrations and input into the abnormal data identification tree to obtain the number of classification node layers for each urine protein concentration as abnormal data. Multiple identification node layers are obtained. If the urine protein concentration is not classified as abnormal data, the maximum number of classification node layers is taken as the identification node layer.

[0105] The detection confidence level is calculated based on the number of recognition node layers.

[0106] Furthermore, the "construction of an anomaly identification tree based on the set of sample urinary protein concentrations" includes:

[0107] Based on historical data of the target urinary protein, a set of urinary protein concentrations from the samples was collected.

[0108] Randomly select the first sample urine protein concentration from the set of sample urine protein concentrations to construct a first-level classification node. The first-level classification node performs binary classification on the input urine protein concentration to obtain two categories: greater than or equal to or less than the first sample urine protein concentration. The results are then input into the previous-level classification node for further binary classification.

[0109] Continue to construct multi-level classification nodes to obtain an anomaly data identification tree, where the urine protein concentration classified as isolated data in each classification node is considered anomaly data.

[0110] Furthermore, the phrase "calculating the detection confidence based on the number of multiple identification node layers" includes:

[0111] Calculate the absolute difference between the maximum and minimum values ​​within the multiple identification node layers;

[0112] Calculate the ratio of the absolute difference to the maximum value, and then calculate the detection confidence level.

[0113] The concentration screening module 13 is specifically used for:

[0114] The detection confidence level is input into the abnormality ratio classification library to obtain the abnormality ratio. The abnormality ratio classification library is constructed by mapping the sample detection confidence level set of sample users to the sample abnormality ratio set of urine protein concentration. The detection confidence level and the abnormality ratio are negatively correlated.

[0115] The multiple urine protein concentration combinations are input into the abnormal data identification tree to obtain all urine protein concentrations that are classified as isolated data, and sorted in ascending order of classification node layer number to obtain the abnormal data sequence.

[0116] Delete the urinary protein concentrations with the highest proportion of abnormality in the abnormal data sequence, and retain the other urinary protein concentrations to obtain the screened urinary protein concentrations.

[0117] The warning output module 14 is specifically used for:

[0118] Obtain standard urine protein concentration;

[0119] The range by which the screened urinary protein concentration exceeds the standard urinary protein concentration is calculated and the average value is calculated to obtain an auxiliary warning coefficient for discrimination and warning.

[0120] In summary, the embodiments of this application have at least the following technical effects:

[0121] Compared to existing technologies, this application firstly collects multiple urine samples from users through a data acquisition module, performs target urinary protein detection, and obtains multiple urinary protein concentrations. Multiple Klotho protein concentrations are obtained through multi-sample detection, effectively reducing the impact of sample randomness and detection errors, providing a reliable data foundation for subsequent accurate early warning of diabetic nephropathy. Secondly, through an anomaly identification module, each urinary protein concentration is combined with a set of normal urinary protein concentrations and input into an anomaly data identification tree to obtain multiple identification node layers. The detection confidence level is calculated, and the anomaly data identification tree constructed using the isolated forest algorithm achieves adaptive identification of abnormal urinary protein data. The reliability of the detection data is assessed through the detection confidence level, providing reliable support for accurate early warning of diabetic nephropathy. Thirdly, through a concentration screening module, an anomaly ratio is configured according to the detection confidence level. Multiple urinary protein concentrations are combined with a set of normal urinary protein concentrations and input into the anomaly data identification tree. The urinary protein concentrations that were previously classified as abnormal are deleted, obtaining the screened urinary protein concentrations, providing a reliable data foundation for subsequent early warning of diabetic nephropathy. Finally, the early warning output module calculates an auxiliary early warning coefficient based on the screened urinary protein concentration and issues an early warning. The entire process, through rigorous calculations and scientifically set thresholds, achieves efficient transformation from data processing to risk warning, improving the accuracy of diabetic nephropathy early warning. This effectively eliminates abnormal data caused by sample randomness and detection errors, thus enhancing the accuracy of diabetic nephropathy early warning.

[0122] Example 3, as Figure 3 As shown, an embodiment of the present invention provides an electronic device 200, including a memory 210, a processor 220, and a computer program 211 stored in the memory 210 and executable on the processor 220. When the processor 220 executes the computer program 211, it implements a diabetic nephropathy early warning method in Embodiment 1.

[0123] Example 4: This example provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements a method for early warning of diabetic nephropathy as described in Example 1.

[0124] It should be noted that the descriptions of each embodiment in the above embodiments have different focuses. For parts that are not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.

[0125] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0126] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded computer, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0127] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0128] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0129] Although preferred embodiments of the invention have been described, those skilled in the art, once they have learned the basic inventive concept, can make other changes and modifications to these embodiments.

[0130] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of this invention and its equivalents, this invention also intends to include these modifications and variations.

Claims

1. A method for early warning of diabetic nephropathy, characterized in that, The method comprises: Multiple urine samples were collected from the user to detect the target urinary protein and obtain multiple urinary protein concentrations; Each urine protein concentration is combined with a normal urine protein concentration set and input into an abnormal data identification tree to obtain multiple identification node layers, and the detection confidence is calculated. Based on the detection confidence level, configure the abnormal proportion, input the multiple urine protein concentrations into the normal urine protein concentration set into the abnormal data identification tree, delete the urine protein concentrations that were classified as abnormal data before the abnormal proportion, and obtain the screened urine protein concentrations. Based on the screened urinary protein concentration, an auxiliary early warning coefficient is calculated and an early warning is issued.

2. The method for early warning of diabetic nephropathy according to claim 1, characterized in that, Multiple urine samples were collected from the user for target urinary protein detection, yielding multiple urinary protein concentrations, including: Collect multiple urine samples from the user; Multiple urine samples were subjected to target urinary protein concentration detection to obtain multiple urinary protein concentrations, among which the urinary protein concentration was Klotho protein.

3. The method for early warning of diabetic nephropathy according to claim 1, characterized in that, Each urine protein concentration is combined with a set of normal urine protein concentrations and input into an abnormal data identification tree to obtain multiple identification node layers. The detection confidence is then calculated, including: Based on the set of sample urinary protein concentrations, an anomaly identification tree is constructed. The anomaly identification tree includes multiple layers of classification nodes. Each layer of classification nodes performs binary classification on the input urinary protein concentration. Urinary protein concentrations classified as isolated data are considered anomaly data. Each urine protein concentration is combined with a set of normal urine protein concentrations and input into the abnormal data identification tree to obtain the number of classification node layers for each urine protein concentration as abnormal data. Multiple identification node layers are obtained. If the urine protein concentration is not classified as abnormal data, the maximum number of classification node layers is taken as the identification node layer. The detection confidence level is calculated based on the number of recognition node layers.

4. The method for early warning of diabetic nephropathy according to claim 3, characterized in that, Based on the set of sample urine protein concentrations, an anomaly identification tree is constructed, including: Based on historical data of the target urinary protein, a set of urinary protein concentrations from the samples was collected. Randomly select the first sample urine protein concentration from the set of sample urine protein concentrations to construct a first-level classification node. The first-level classification node performs binary classification on the input urine protein concentration to obtain two categories: greater than or equal to or less than the first sample urine protein concentration. The results are then input into the previous-level classification node for further binary classification. Continue to construct multi-level classification nodes to obtain an anomaly data identification tree, where the urine protein concentration classified as isolated data in each classification node is considered anomaly data.

5. The method for early warning of diabetic nephropathy according to claim 3, characterized in that, Based on the number of recognition node layers, the detection confidence is calculated, including: Calculate the absolute difference between the maximum and minimum values ​​within the multiple identification node layers; Calculate the ratio of the absolute difference to the maximum value, and then calculate the detection confidence level.

6. The method for early warning of diabetic nephropathy according to claim 1, characterized in that, Based on the detection confidence level, an abnormality ratio is configured. The multiple urine protein concentrations combined with the normal urine protein concentration set are input into an abnormal data identification tree. Urine protein concentrations that were previously classified as abnormal data are deleted from the abnormality ratio to obtain the screened urine protein concentrations, including: The detection confidence level is input into the abnormality ratio classification library to obtain the abnormality ratio. The abnormality ratio classification library is constructed by mapping the sample detection confidence level set of sample users to the sample abnormality ratio set of urine protein concentration. The detection confidence level and the abnormality ratio are negatively correlated. The multiple urine protein concentration combinations are input into the abnormal data identification tree to obtain all urine protein concentrations that are classified as isolated data, and sorted in ascending order of classification node layer number to obtain the abnormal data sequence. Delete the urinary protein concentrations with the highest proportion of abnormality in the abnormal data sequence, and retain the other urinary protein concentrations to obtain the screened urinary protein concentrations.

7. The method for early warning of diabetic nephropathy according to claim 1, characterized in that, Based on the screened urine protein concentration, an auxiliary early warning coefficient is calculated, and an early warning is issued, including: Obtain standard urine protein concentration; The range by which the screened urinary protein concentration exceeds the standard urinary protein concentration is calculated and the average value is calculated to obtain an auxiliary warning coefficient for discrimination and warning.

8. A diabetic nephropathy early warning system, characterized in that, For performing the method according to any one of claims 1-7, comprising: The data acquisition module is used to collect multiple urine samples from the user, perform target urine protein detection, and obtain multiple urine protein concentrations; The anomaly identification module is used to combine each urine protein concentration with the normal urine protein concentration set and input it into the anomaly data identification tree to obtain multiple identification node layers and calculate the detection confidence. The concentration screening module is used to configure the abnormal proportion according to the detection confidence level, input the multiple urine protein concentrations into the normal urine protein concentration set into the abnormal data identification tree, delete the urine protein concentrations that were classified as abnormal data before the abnormal proportion, and obtain the screened urine protein concentrations. The early warning output module is used to calculate the auxiliary early warning coefficient based on the screened urine protein concentration and to issue an early warning.

9. An electronic device, characterized in that, include: Memory, used to store computer software programs; A processor is configured to read and execute the computer software program, thereby implementing the method for early warning of diabetic nephropathy as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The storage medium stores a computer program, which, when executed by a processor, implements a method for early warning of diabetic nephropathy as described in any one of claims 1-7.

Images