Detection method of non-white proteinuria and application of non-white proteinuria in auxiliary diagnosis of metabolic syndrome
By combining qualitative and quantitative urine detection and using machine learning model analysis, the problem of inability to effectively distinguish albuminuria and non-albuminuria in the existing technology is solved, and the accuracy and efficiency of urine protein detection is improved, and for metabolism The auxiliary diagnosis of syndrome provides new approaches.
Patent Information
- Application Number
- CN202411529375.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-10-30
- Publication Date
- 2025-02-11
AI Technical Summary
Existing urine protein detection methods cannot effectively distinguish between albuminuria and non-albuminuria, resulting in difficulty in clinical diagnosis, and lack of systematic research and detection methods, affecting the auxiliary diagnosis of metabolic syndrome.
Using a combination of qualitative and quantitative detection methods, the urine samples are initially screened and accurately tested. Through the qualitative analysis of urine pH, urine protein and urethritis, the appropriate quantitative detection mode is selected to provide the total urine protein and microalbumin. Quantitative testing was performed and data analysis was performed using machine learning models to determine whether it was nonalbuminuria.
Improve the accuracy and efficiency of urine protein detection, can quickly and accurately distinguish albuminuria and non-albuminuria, provide a new way to assist in the diagnosis of metabolic syndrome, and significantly reduce the review rate of high-value specimens, saving Testing cost.
Smart Images
Figure BDA0005110111230000092 
Figure BDA0005110111230000101 
Figure BDA0005110111230000111
Abstract
Description
Technical Field
[0001] The present invention belongs to the fields of biomedical technology and clinical testing technology, and in particular relates to a method for detecting non-albuminuria and its application in auxiliary diagnosis of metabolic syndrome. Background Art
[0002] Urine protein detection can be roughly divided into dry chemical detection and protein quantitative detection. Urine protein qualitative detection uses dry chemical test strips, which can detect urine protein qualitatively and semi-quantitatively, which is convenient and fast. Protein quantitative detection is a quantitative test of relevant urine disease markers, among which total protein in urine (TPU) and microalbumin in urine (MALB) are the representatives. The protein quantitative detection process is relatively complicated and requires a long detection time. In theory, the results of dry chemical detection and protein quantitative detection are consistent, which we can call the consistency of urine protein detection results. In some cases, the results of dry chemical test and urine protein quantitative detection can support each other, and the urine protein results obtained by two different detection methods can refer to each other. However, at present, the two different urine protein detection methods cannot share a unified platform, the degree of automation is low, the degree of mutual support is low, and the performance of the urine protein detection system cannot be improved to meet the clinical detection needs. Moreover, the problem of inconsistency between the qualitative and quantitative results of urine protein is often encountered in clinical work. The main problem is that the clinic pays too much attention to albuminuria and ignores non-albuminuria. This also poses a challenge to the reasonable interpretation of the test results.
[0003] If the problem of consistency of urine protein results can be solved, it will bring the performance of urine protein detection to a new level. Then, the next question is to explore new clinical applications or significance after the improvement of urine protein performance, specifically to distinguish albuminuria from non-albuminuria. The detection of albuminuria has always been the focus of clinical practice, but there has been no good systematic detection method for non-albuminuria internationally and domestically, and it has not attracted enough attention. Non-albuminuria is a common clinical symptom, which refers to the abnormal increase of non-albumin components in proteinuria, that is, samples with elevated urine protein in urine can be considered as non-albuminuria as long as the main protein component is excluded as albumin. A small number of literature reports that non-albuminuria can be defined by the ratio of urine microalbumin to urine total protein. Generally, the ratio is selected to be less than 0.4, which is non-albuminuria. In actual work, we often encounter patients with different urine protein concentrations, whose urine microalbumin to urine total protein ratios are too different, and the use of ratios to judge albuminuria / non-albuminuria will be greatly affected. Therefore, it will be of great significance to establish a method for determining non-albuminuria based on urine protein concentration.
[0004] Currently known causes of non-albuminuria include hematuria, proteinuria caused by urinary tract infection, Bence Jones proteinuria, and physiological factors such as exercise and menstruation. In addition, there are few clear reports on other pathological factors that lead to non-albuminuria, and there is a lack of systematic research on non-albuminuria detection methods and pathogenic factors. The chronic kidney disease clinical practice guidelines issued by KDIGO (Kidney Disease Improving Global Outcomes) are blank in their recommendations on pathogenic factors that lead to non-albuminuria, and only suggest that urine dry chemical test strips may miss the detection of non-albuminuria. Therefore, it will be of great significance to establish a non-albuminuria detection system and study its disease spectrum / pathogenic factors.
[0005] Metabolic syndrome (MetS) forms a series of metabolic disorders, including insulin resistance, atherosclerotic dyslipidemia, central obesity and hypertension. It is causally related to non-alcoholic fatty liver disease (NAFLD) and type 2 diabetes, and together promotes the occurrence of diseases such as atherosclerotic cardiovascular disease (CVD), chronic kidney disease (CKD) and liver decompensation. The pathogenesis of MetS includes multiple genetic and acquired entities under insulin resistance and chronic low-grade inflammation. If left untreated, MetS is significantly associated with an increased risk of diabetes, CVD and all-cause mortality. There are currently no studies on non-albuminuria associated with metabolic syndrome. The current staging of hyperlipidemia, hypertension and diabetes patients, efficacy monitoring and prognosis judgment all rely on the detection of albuminuria to a certain extent. However, the role of non-albuminuria in MetS needs to be studied urgently. Summary of the invention
[0006] In view of the problems and shortcomings in the prior art, the purpose of the present invention is to provide a method for detecting non-albuminuria and its application in auxiliary diagnosis of metabolic syndrome.
[0007] To achieve the purpose of the invention, the technical solution adopted by the present invention is as follows:
[0008] The first aspect of the present invention provides a method for detecting non-albuminuria, comprising the following steps:
[0009] S1: Performing a qualitative analysis on the urine sample to obtain a qualitative test result of the urine sample, wherein the qualitative test result includes urine pH, urine protein PRO and urine occult blood BLD; if the urine occult blood BLD in the qualitative test result is positive, the urine sample is judged to be suspected non-albuminuria; if the urine occult blood BLD in the qualitative test result is negative, executing step S2;
[0010] S2: According to the urine qualitative test result obtained in step S1, the quantitative test mode of the urine sample is selected, and the urine sample is subjected to quantitative test of total urine protein TPU and microalbumin MALB according to the selected quantitative test mode to obtain the quantitative test results of total urine protein TPU and microalbumin MALB of the urine sample; then, whether the urine sample is non-albuminuria is jointly determined based on the quantitative test result of total urine protein TPU, the quantitative test result of microalbumin MALB and the qualitative test result of urine protein PRO.
[0011] According to the above detection method, preferably, in step S2, the quantitative detection mode includes an original multiple quantitative detection mode and a diluted quantitative detection mode; the original multiple quantitative detection mode is: the urine sample does not need to be diluted, and the urine total protein TPU and microalbumin MALB are directly detected on the urine sample; the diluted quantitative detection mode is: the urine sample is first diluted, and then the diluted urine sample is subjected to quantitative detection of total urine protein TPU and microalbumin MALB.
[0012] According to the above detection method, preferably, in step S2, according to the urine pH, urine protein PRO and urine occult blood BLD test results obtained in step S1, the specific operation of selecting the quantitative detection mode of the urine sample is:
[0013] If the urine occult blood BLD in the qualitative test result is negative and the pH is ≥8.0, the urine sample is subjected to quantitative detection of total urine protein TPU and microalbumin MALB using the original quantitative detection mode;
[0014] If the urine occult blood BLD is negative, pH < 8.0 and PRO is negative in the qualitative test results, the urine sample is subjected to quantitative detection of total urine protein TPU and microalbumin MALB using the original quantitative detection mode;
[0015] If the urine occult blood BLD is negative, pH < 8.0 and PRO 1+ in the qualitative test results, the urine sample is subjected to quantitative detection of total urine protein TPU and microalbumin MALB using the original quantitative detection mode;
[0016] If the urine occult blood BLD is negative, pH is less than 8.0, and PRO is weakly positive in the qualitative test results, the urine sample is subjected to quantitative detection of total urine protein TPU and microalbumin MALB using the original quantitative detection mode;
[0017] If the urine occult blood BLD is negative, pH is less than 8.0, and PRO is greater than or equal to 2+ in the qualitative test results, the urine sample is quantitatively tested using a dilution quantitative test mode.
[0018] According to the above detection method, preferably, if the urine occult blood BLD is negative, pH <8.0 and PRO 2+ in the qualitative test result, the urine sample is diluted 3 times and then subjected to TPU quantitative detection, and the urine sample is diluted 11 times and then subjected to MALB quantitative detection; if the urine occult blood BLD is negative, pH <8.0 and PRO ≥3+ in the qualitative test result, the urine sample is diluted 5 times and then subjected to TPU quantitative detection, and the urine sample is diluted 20 times and then subjected to MALB quantitative detection.
[0019] According to the above detection method, preferably, in step S2, the specific operation of jointly judging whether the urine sample is non-albuminuria based on the quantitative detection result of total urine protein TPU, the quantitative detection result of microalbumin MALB and the qualitative detection result of urine protein PRO is: inputting the quantitative detection result of total urine protein TPU, the quantitative detection result of microalbumin MALB and the qualitative detection result of urine protein PRO of the urine specimen into the trained machine learning model for detection and judgment, to obtain a judgment result of whether the urine sample is suspected non-albuminuria.
[0020] According to the above detection method, preferably, the machine learning model is a support vector machine model, a decision tree model, a random forest model, an AdaBoost model or an XGBoost model.
[0021] According to the above detection method, preferably, the training method of the machine learning model is:
[0022] (1) obtaining a sample set, wherein the sample set includes a plurality of samples, each sample including a quantitative detection result of total urine protein TPU, a quantitative detection result of microalbumin MALB, a qualitative detection result of urine protein PRO, and a true classification label of non-albuminuria corresponding to the sample; dividing the sample set into a training set and a test set in proportion;
[0023] (2) Inputting the urine total protein TPU quantitative detection result, microalbumin MALB quantitative detection result and urine protein PRO qualitative detection result of each training sample in the training set into the machine learning model for training, obtaining the non-albuminuria prediction result of each urine sample, comparing the non-albuminuria prediction result with the non-albuminuria true classification label corresponding to the sample, and constructing a loss function; then using the back propagation method to update the network parameters of the machine learning model, reduce the loss function value, until the loss function converges, and obtain the trained machine learning model;
[0024] (3) The trained machine learning model is tested using a test set to evaluate the performance of the trained machine learning model in diagnosing and distinguishing non-albuminuria, until the accuracy of the trained machine learning model in diagnosing and distinguishing non-albuminuria is greater than or equal to a preset threshold, thereby obtaining a trained machine learning model.
[0025] According to the above detection method, preferably, in step S1, the qualitative analysis is to perform qualitative analysis on the urine sample using a urine test paper method.
[0026] According to the above detection method, preferably, the urine sample is random urine or 24-hour urine.
[0027] The second aspect of the present invention provides a non-albuminuria detection system, which includes a laboratory information system, an IS system and a urine quantitative analysis system; the laboratory information system receives and stores the qualitative detection results of urine samples by a urine qualitative analyzer, and sends the qualitative detection results to the IS system, wherein the qualitative detection results include urine pH, urine protein PRO and urine occult blood BLD; the IS system receives the qualitative detection results of the urine samples sent by the laboratory information system, and analyzes the qualitative detection results of the urine samples. If the urine occult blood BLD in the qualitative detection result is positive, the IS system determines that the urine sample is suspected non-albuminuria and sends the judgment result to the laboratory information system; if the urine occult blood BLD in the qualitative detection result is negative, the IS system will determine that the urine sample is suspected non-albuminuria and send the judgment result to the laboratory information system; if the urine occult blood BLD in the qualitative detection result is negative, the IS system will determine that the urine sample is suspected non-albuminuria according to the qualitative detection result of the urine sample. The method comprises the following steps: selecting a quantitative detection mode for a urine sample according to the quantitative detection result, and sending the selected urine sample quantitative detection mode instruction to the urine quantitative analysis system; the urine quantitative analysis system performs urine total protein TPU and microalbumin MALB quantitative detection on the urine sample according to the received urine sample quantitative detection mode instruction, obtains the quantitative detection results of urine total protein TPU and microalbumin MALB of the urine sample, and sends the quantitative detection results of urine total protein TPU and microalbumin MALB to the IS system; the IS system jointly determines whether the urine sample is non-albuminuria based on the received urine total protein TPU quantitative detection result, microalbumin MALB quantitative detection result and urine protein PRO qualitative detection result of the urine sample, and sends the judgment result to the laboratory information system.
[0028] According to the above-mentioned non-albuminuria detection system, preferably, the specific operation of the IS system for jointly judging whether the urine sample is non-albuminuria based on the received urine total protein TPU quantitative detection result, microalbumin MALB quantitative detection result, and urine protein PRO qualitative detection result of the urine sample is: inputting the urine total protein TPU quantitative detection result, microalbumin MALB quantitative detection result, and urine protein PRO qualitative detection result of the urine specimen into the trained machine learning model installed in the IS system for detection and judgment, and obtaining a judgment result of whether the urine sample is suspected to be non-albuminuria.
[0029] According to the above-mentioned non-albuminuria detection system, preferably, the machine learning model is a support vector machine model, a decision tree model, a random forest model, an AdaBoost model or an XGBoost model.
[0030] According to the above non-albuminuria detection system, preferably, the training method of the machine learning model is:
[0031] (1) obtaining a sample set, wherein the sample set includes a plurality of samples, each sample including a quantitative detection result of total urine protein TPU, a quantitative detection result of microalbumin MALB, a qualitative detection result of urine protein PRO, and a true classification label of non-albuminuria corresponding to the sample; dividing the sample set into a training set and a test set in proportion;
[0032] (2) Inputting the urine total protein TPU quantitative detection result, microalbumin MALB quantitative detection result and urine protein PRO qualitative detection result of each training sample in the training set into the machine learning model for training, obtaining the non-albuminuria prediction result of each urine sample, comparing the non-albuminuria prediction result with the non-albuminuria true classification label corresponding to the sample, and constructing a loss function; then using the back propagation method to update the network parameters of the machine learning model, reduce the loss function value, until the loss function converges, and obtain the trained machine learning model;
[0033] (3) The trained machine learning model is tested using a test set to evaluate the performance of the trained machine learning model in diagnosing and distinguishing non-albuminuria, until the accuracy of the trained machine learning model in diagnosing and distinguishing non-albuminuria is greater than or equal to a preset threshold, thereby obtaining a trained machine learning model.
[0034] According to the above-mentioned non-albuminuria detection system, preferably, the urine quantitative analysis system includes a middleware platform and a urine quantitative biochemical analyzer, the middleware platform is used to receive the urine sample quantitative detection mode instruction sent by the IS system, and control the urine quantitative biochemical analyzer to perform urine total protein TPU and microalbumin MALB quantitative detection on the urine sample according to the urine sample quantitative detection mode instruction; the urine quantitative biochemical analyzer is used to perform urine total protein TPU and microalbumin MALB quantitative detection on the urine sample according to the instruction of the middleware platform, and after the detection is completed, the urine total protein TPU and microalbumin MALB quantitative detection results of the urine sample are sent to the IS system. More preferably, the middleware platform is cobas infinity; the urine quantitative biochemical analyzer is Cobas c701 analyzer.
[0035] According to the above non-albuminuria detection system, preferably, the urine sample is random urine or 24-hour urine.
[0036] The third aspect of the present invention provides the use of a detection reagent and / or device for detecting non-albuminuria in a urine sample in the preparation of a product for auxiliary diagnosis of metabolic syndrome.
[0037] According to the above application, preferably, the detection reagent or / and instrument is a detection reagent or / and instrument used when the urine sample is detected by the non-albuminuria detection method described in the first aspect. More preferably, the detection reagent or / and instrument is a detection reagent or / and instrument used for qualitative detection of urine samples, quantitative detection of total protein in urine TPU, and quantitative detection of microalbumin MALB. Most preferably, the detection reagent or / and instrument is a detection reagent or / and instrument used for qualitative detection of urine protein PRO, quantitative detection of total protein in urine TPU, and quantitative detection of microalbumin MALB in urine samples.
[0038] Compared with the prior art, the present invention has the following positive and beneficial effects:
[0039] (1) In the actual detection process of existing urine protein, the inaccurate and untimely ALBU and TPU test results of extremely low-concentration and high-concentration urine samples will have a significant impact on the non-albuminuria test results. For example, high-concentration albuminuria often leads to negative test results. Multiple dilutions and retests of high-concentration albuminuria greatly prolong the time to produce results and the test performance. In view of this technical problem, the non-albuminuria detection method of the present invention first performs a qualitative analysis on the urine sample to obtain the qualitative test results of the urine sample, and then makes a preliminary judgment based on the qualitative test results. If the urine occult blood BLD in the qualitative test results is positive, the urine sample is directly judged to be suspected non-albuminuria; if the urine occult blood BLD in the qualitative test results is positive, the urine sample is directly judged to be suspected non-albuminuria; If the blood BLD is negative, different quantitative detection modes are selected for urine samples, and the urine samples are quantitatively detected according to the selected quantitative detection modes to obtain quantitative detection results of total urine protein TPU and microalbumin MALB of the urine samples, which greatly improves the detection limit and detection accuracy of the quantitative detection results of total urine protein TPU and microalbumin MALB of the urine samples, and the detection linear range is greatly expanded. In actual operation, the re-examination rate of related high-value specimens is reduced from the original 12.94% to 0.84%, which greatly saves the detection cost; at the same time, it shortens the detection time and solves the technical problem of multiple dilution tests for high-concentration albuminuria that greatly prolongs the result time and reduces the detection performance.
[0040] (2) The non-albuminuria detection method of the present invention detects non-albuminuria according to a method combining urine dry chemical detection and urine quantitative detection. Specifically, the urine total protein TPU quantitative detection result, microalbumin MALB quantitative detection result and urine protein PRO qualitative detection result of the urine sample are jointly input into a machine learning model to determine whether the urine sample is non-albuminuria or albuminuria. The detection result is fast and accurate.
[0041] (3) Based on the non-albuminuria detection method developed by the present invention, the present invention studied the disease spectrum / pathogenic factors of non-albuminuria for the first time; and found that non-albuminuria is used as an auxiliary diagnostic indicator for MetS for the auxiliary diagnosis of MetS. Moreover, the AUC of the ROC curve for diagnosing and distinguishing MetS from normal healthy controls based on the non-albuminuria detection results of urine samples reached 0.94, which has a high diagnostic value and provides a new way for the auxiliary diagnosis of clinical MetS. BRIEF DESCRIPTION OF THE DRAWINGS
[0042] Figure 1 is a schematic diagram of a non-albuminuria detection system of the present invention;
[0043] Figure 2 ROC curves for the diagnosis and differentiation of non-albuminuria and albuminuria for five machine learning models;
[0044] Figure 3 This is a comparison chart of the changes in the in-laboratory turnover time of specimen testing in the actual application of the non-albuminuria detection method. The unit of the vertical axis is seconds;
[0045] Figure 4 The figure is a comparison chart of the detection ranges of different non-albuminuria detection systems, wherein A is the detection limit and linear range of TPU in the detection modes of Pattern 1, Pattern 2 and Pattern 3; B is the detection limit and linear range of ALBU in the detection modes of Pattern 1, Pattern 2 and Pattern 3;
[0046] Figure 5The CICOS diagram is a spectrum analysis of diseases related to non-albuminuria; A is a CICOS diagram of diseases that appeared more than 6 times in the clinical diagnosis of 108 cases; B is a CICOS diagram of diseases that appeared in the top 6 times in the clinical diagnosis of 108 cases; Hyt is hypertension; T2DM is type 2 diabetes; NAFLD is non-alcoholic fatty liver disease; Hyl is hyperlipidemia; ThN is thyroid nodule; NeS is nephrotic syndrome; SLE is systemic lupus erythematosus; Pnm is pneumonia; CKD is chronic kidney disease; LN is lupus nephritis; RCI-CCI is chronic cerebral infarction; CHD is coronary artery atherosclerotic heart disease; VD3D is vitamin D3 deficiency; Nepr is nephritis; ReA is renal anemia; Gar is gastritis; ReC is renal cyst; Phyp is prostatic hyperplasia; Ploy K stands for polycystic kidney disease; IgAN stands for IgA nephropathy; PAD stands for peripheral arterial atherosclerosis; Pul N stands for pulmonary nodules; Osp stands for osteoporosis; MM stands for multiple myeloma; Memnep stands for membranous nephropathy; Hyi stands for hyperuricemia; Ks stands for kidney stones; Gs stands for gallstones;
[0047] Figure 6 ROC curve for diagnosing and distinguishing metabolic syndrome patients from healthy controls by non-albuminuria, wherein the predicted probability represents the ROC curve for diagnosing and distinguishing metabolic syndrome patients from healthy controls by using the non-albuminuria test results of the present invention, and PRO, TPU, and MALB represent the ROC curves for diagnosing and distinguishing metabolic syndrome patients from healthy controls by using urine protein PRO, urine total protein TPU, and microalbumin MALB test results alone. DETAILED DESCRIPTION
[0048] In order to enable those skilled in the art to more clearly understand the technical solution of the present invention, the technical solution of the present invention will be described in detail below in conjunction with specific embodiments.
[0049] The following detailed descriptions are exemplary and are intended to provide further explanation of the present invention. Unless otherwise specified, all technical and scientific terms used herein have the same meanings as those commonly understood by those skilled in the art to which the present invention belongs.
[0050] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit exemplary embodiments according to the present invention. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that when the terms "comprise" and / or "include" are used in this specification, it indicates the presence of features, steps, operations, components and / or combinations thereof.
[0051] The experimental methods in the following examples without specifying specific conditions all adopt conventional techniques in the art or follow the conditions recommended by the manufacturers; the reagents or instruments used without specifying the manufacturers are all conventional products that can be obtained commercially.
[0052] Embodiment 1:
[0053] A method for detecting non-albuminuria comprises the following steps:
[0054] S1: Performing qualitative analysis on the urine sample by urine test paper method (also known as dry chemical test paper method) to obtain qualitative test results of the urine sample, wherein the qualitative test results include urine pH, urine protein PRO and urine occult blood BLD; if the urine occult blood BLD in the qualitative test result is positive, the urine sample is judged to be suspected non-albuminuria; if the urine occult blood BLD in the qualitative test result is negative, executing step S2;
[0055] S2: According to the urine qualitative test results obtained in step S1, select the quantitative test mode of the urine sample, and perform quantitative tests on the urine total protein TPU and microalbumin MALB on the urine sample according to the selected quantitative test mode to obtain the quantitative test results of the urine total protein TPU and microalbumin MALB of the urine sample; then jointly judge whether the urine sample is non-albuminuria based on the quantitative test results of the urine total protein TPU, the quantitative test results of the microalbumin MALB and the qualitative test results of the urine protein PRO. The quantitative test modes include the original quantitative test mode and the diluted quantitative test mode; the original quantitative test mode is that the urine sample does not need to be diluted, and the urine sample is directly subjected to quantitative tests of the urine total protein TPU and microalbumin MALB; the diluted quantitative test mode is that the urine sample is first diluted, and then the diluted urine sample is subjected to quantitative tests of the urine total protein TPU and microalbumin MALB.
[0056] According to the urine pH, urine protein PRO and urine occult blood BLD test results obtained in step S1, the specific operation of selecting the quantitative detection mode of the urine sample is:
[0057] If the urine occult blood BLD in the qualitative test result is negative and the pH is ≥8.0, the urine sample is subjected to quantitative detection of total urine protein TPU and microalbumin MALB using the original quantitative detection mode;
[0058] If the urine occult blood BLD is negative, pH < 8.0 and PRO is negative in the qualitative test results, the urine sample is subjected to quantitative detection of total urine protein TPU and microalbumin MALB using the original quantitative detection mode;
[0059] If the urine occult blood BLD is negative, pH < 8.0 and PRO 1+ in the qualitative test results, the urine sample is subjected to quantitative detection of total urine protein TPU and microalbumin MALB using the original quantitative detection mode;
[0060] If the urine occult blood BLD is negative, pH is less than 8.0, and PRO is weakly positive in the qualitative test results, the urine sample is subjected to quantitative detection of total urine protein TPU and microalbumin MALB using the original quantitative detection mode;
[0061] If the urine occult blood BLD is negative, pH < 8.0 and PRO ≥ 2+ in the qualitative test results, the urine sample is quantitatively tested using the dilution quantitative test mode. Further, if the urine occult blood BLD is negative, pH < 8.0 and PRO 2+ in the qualitative test results, the urine sample is preferably diluted 3 times for TPU quantitative test, and the urine sample is preferably diluted 11 times for MALB quantitative test; if the urine occult blood BLD is negative, pH < 8.0 and PRO ≥ 3+ in the qualitative test results, the urine sample is preferably diluted 5 times for TPU quantitative test, and the urine sample is preferably diluted 20 times for MALB quantitative test.
[0062] Furthermore, the specific operation of jointly judging whether the urine sample is non-albuminuria based on the quantitative detection results of total urine protein TPU, the quantitative detection results of microalbumin MALB and the qualitative detection results of urine protein PRO is as follows: the quantitative detection results of total urine protein TPU, the quantitative detection results of microalbumin MALB and the qualitative detection results of urine protein PRO of the urine specimen are input into the trained machine learning model for detection and judgment, and the judgment result of whether the urine sample is suspected non-albuminuria is obtained. Preferably, the machine learning model is a support vector machine model (SVM), a decision tree model, a random forest model, an AdaBoost model or an XGBoost model; more preferably, the machine learning model is a support vector machine model (SVM). The non-albuminuria detection method of the present invention is suitable for detecting random urine or 24-hour urine samples.
[0063] The derivation process of the conclusion that the non-albuminuria detection method of the present invention is suitable for detecting random urine or 24-hour urine samples is as follows:
[0064] uAPR=ACR / PCR, because ACR=MALB / urine creatinine, PCR=TPU / urine creatinine, therefore, uAPR=MALB / TPU; wherein ACR represents spot albumin, PCR represents spot total protein, MALB represents microalbumin, and TPU represents total urine protein.
[0065] Assuming that within 24 hours, the total protein and microalbumin results of the nth spot proteinuria are TPUn and MALBn, respectively, and the corresponding spot protein results are ACRn and PCRn, respectively.
[0066]
[0067] After transforming the above formula 1, we get:
[0068]
[0069] It can be seen from Formula 2 and Formula 3 that this rule can be used to determine albuminuria / non-albuminuria for both random urine (uAPR) and 24-hour urine (uAPR').
[0070] Embodiment 2:
[0071] A non-albuminuria detection system (such as Figure 1As shown), the detection system includes a laboratory information system (LIS), an IS system and a urine quantitative analysis system, and the urine quantitative analysis system includes a middleware platform and a urine quantitative biochemical analyzer. The laboratory information system (LIS) receives and stores the qualitative detection results of the urine sample by the urine qualitative analyzer, and sends the qualitative detection results to the IS system, and the qualitative detection results include urine pH, urine protein PRO and urine occult blood BLD; the IS system receives the qualitative detection results of the urine sample sent by the laboratory information system, and analyzes the qualitative detection results of the urine sample. If the urine occult blood BLD in the qualitative detection result is positive, the IS system determines that the urine sample is suspected of non-albuminuria, and sends the judgment result to the laboratory information system; if the urine occult blood BLD in the qualitative detection result is negative, the IS system will select the quantitative detection mode of the urine sample according to the qualitative detection result of the urine sample, and send the selected urine sample quantitative detection mode instruction to the middleware platform of the urine quantitative analysis system. The middleware platform (preferably the middleware platform is cobas infinity) is used to receive the urine sample quantitative detection mode instruction sent by the IS system, and control the urine quantitative biochemical analyzer to perform urine total protein TPU and microalbumin MALB quantitative detection on the urine sample according to the urine sample quantitative detection mode instruction; the urine quantitative biochemical analyzer (preferably the urine quantitative biochemical analyzer is Cobas c701 analyzer) is used to perform urine total protein TPU and microalbumin MALB quantitative detection on the urine sample according to the instruction of the middleware platform, and after the detection is completed, the urine total protein TPU and microalbumin MALB quantitative detection results of the urine sample are sent to the IS system. The IS system jointly determines whether the urine sample is non-albuminuria based on the urine total protein TPU quantitative detection result, microalbumin MALB quantitative detection result, and urine protein PRO qualitative detection result of the received urine sample, and sends the judgment result to the laboratory information system.
[0072] Wherein, the IS system is equipped with a trained machine learning model, preferably, the machine learning model is a support vector machine model, a decision tree model, a random forest model, an AdaBoost model or an XGBoost model; more preferably, the machine learning model is a support vector machine model (SVM). The specific operation of the IS system to jointly judge whether the urine sample is non-albuminuria based on the quantitative detection results of the urine total protein TPU, the quantitative detection results of the microalbumin MALB, and the qualitative detection results of the urine protein PRO of the received urine sample is: the urine total protein TPU quantitative detection results, the microalbumin MALB quantitative detection results, and the urine protein PRO qualitative detection results of the urine specimen are input into the trained machine learning model carried in the IS system for detection and judgment, and the judgment result of whether the urine sample is suspected to be non-albuminuria is obtained. The non-albuminuria detection system of the present invention is suitable for detecting random urine or 24-hour urine samples.
[0073] (I) Training of the machine learning model in the present invention:
[0074] 1. Experimental sample collection:
[0075] This study used a cross-sectional research method to collect urine samples from 1,187 patients randomly collected from the First Affiliated Hospital of Zhengzhou University for 3 days. Among them, 218 patients' urine samples were identified as non-albuminuria, 355 patients' urine samples were identified as albuminuria, and 614 patients' urine samples were identified as normal urine. The urine samples of 1,187 patients were tested using steps S1 and S2 in the non-albuminuria detection method described in Example 1 of the present invention to obtain the qualitative test results (urine pH, urine protein PRO and urine occult blood BLD test results) of the urine samples of 1,187 patients and the quantitative test results of total urine protein TPU and microalbumin MALB. The statistical results of the baseline data of 1,187 patients are shown in Table 1.
[0076] Table 1 shows the clinical baseline data of 1187 patients
[0077]
[0078]
[0079] Note: -, +-, 1+, 2+ and 3+ represent five levels of PRO results; a is the upper and lower values (mean) of the 95% confidence interval; b is the 5th and 95th percentiles; AER is the 24-hour urine microalbumin excretion rate; ACR is the ratio of urine microalbumin to urine creatinine; PER is the 24-hour urine total protein excretion rate; PCR is the ratio of urine total protein to urine creatinine.
[0080] 2. Division of training set and test set and training of machine learning model
[0081] (1) The true classification labels of non-albuminuria in urine samples of 1187 patients were marked as 1, and the true classification labels of non-albuminuria in albuminuria samples and normal urine samples were marked as 0. Then the 1187 patients were randomly divided into training set and test set in a ratio of 8:2.
[0082] (2) The quantitative detection results of total urine protein TPU, the quantitative detection results of microalbumin MALB, and the qualitative detection results of urine protein PRO of each training sample in the training set are input into the machine learning model for training to obtain the non-albuminuria prediction results of each urine sample, and the non-albuminuria prediction results are compared with the true classification labels of non-albuminuria corresponding to the samples to construct a loss function; then the network parameters of the machine learning model are updated using the back propagation method to reduce the loss function value until the loss function converges to obtain the trained machine learning model.
[0083] (3) Testing the trained machine learning model with a test set to evaluate the performance of the trained machine learning model in diagnosing and distinguishing non-albuminuria, until the accuracy of the trained machine learning model in diagnosing and distinguishing non-albuminuria is greater than or equal to a preset threshold, thereby obtaining a trained machine learning model. The machine learning model is a support vector machine model (SVM), a decision tree model, a random forest model, an AdaBoost model, or an XGBoost model.
[0084] The five machine learning models (SVM, decision tree model, random forest model, AdaBoost model, XGBoost model) trained by the present invention were used to diagnose and differentiate the samples in the test set, and the ROC curves of the five machine learning models for diagnosing and distinguishing non-albuminuria and albuminuria were drawn (such as Figure 2 The test results of accuracy, precision, sensitivity, specificity and F1 score for the classification of non-albuminuria and albuminuria are shown in Table 2.
[0085] Table 2 Test results of accuracy, precision, sensitivity, specificity and F1 score of five machine learning models for non-albuminuria and albuminuria classification
[0086] Accuracy Precision Sensitivity Specificity F1 Rating AUC Adaboost 0.933 0.831 0.785 0.964 0.804 0.946 Decision Tree 0.931 0.868 0.713 0.977 0.782 0.894 Random Forest 0.929 0.847 0.735 0.970 0.784 0.957 Support Vector Machine 0.927 0.915 0.644 0.986 0.752 0.967 XGBoost 0.943 0.881 0.787 0.976 0.829 0.945
[0087] Depend on Figure 2As can be seen from Table 2, the AUC values of the ROC curves for diagnosing and distinguishing non-albuminuria and albuminuria using the five trained machine learning models of the present invention are all greater than 0.89, and the accuracy, precision, sensitivity and specificity of the five trained machine learning models for the classification and judgment of non-albuminuria and albuminuria are all good, indicating that the five trained machine learning models can all be used for the diagnosis of non-albuminuria; among them, the AUC value of the ROC curve of the SVM machine learning model for diagnosing and distinguishing non-albuminuria and albuminuria reached a maximum of 0.9665. Therefore, the present invention further prefers the SVM machine learning model for diagnosing and distinguishing non-albuminuria and albuminuria according to the quantitative detection results of total urine protein TPU, the quantitative detection results of microalbumin MALB, and the qualitative detection results of urine protein PRO of urine samples.
[0088] (II) Discussion on the influence of different urine specimen quantitative detection methods on non-albuminuria detection:
[0089] In order to compare with the quantitative detection method of urine total protein TPU and microalbumin MALB recorded in step S2 of Example 1 of the present invention, and to study the influence of different quantitative detection methods on non-albuminuria detection, the "specific operation mode of selecting the quantitative detection mode of the urine specimen according to the detection results of urine pH, urine protein PRO and urine occult blood BLD obtained in step S1" recorded in step S2 of Example 1 of the present invention is recorded as Pattern 3, and two different urine quantitative detection methods (Pattern 1 and Pattern 2) are designed to quantitatively detect urine total protein TPU and microalbumin MALB in urine samples. Pattern 1 is: the urine sample after qualitative analysis is not diluted, and the urine sample after qualitative analysis is directly subjected to quantitative detection of urine total protein TPU and microalbumin MALB using a urine quantitative biochemical analyzer (that is, the urine sample is not diluted, and the urine sample is directly subjected to quantitative detection of urine total protein TPU and microalbumin MALB using a urine quantitative biochemical analyzer). Pattern 2 is: directly use a urine quantitative biochemical analyzer to perform a quantitative test on the urine total protein TPU in the urine sample after qualitative analysis; then dilute the urine sample after qualitative analysis by 5 times to obtain a diluted urine sample, and use a quantitative biochemical analyzer to perform a quantitative test on the diluted urine sample for microalbumin MALB.
[0090] Pattern 1 and Pattern 2 were used to replace step S2 in Example 1, respectively, to complete the non-albuminuria detection of urine samples, and the laboratory turnover time of urine samples to complete the quantitative detection operation was calculated, and the detection limit and linear range of total protein in urine TPU and microalbumin in urine MALB were obtained by quantitative calibration.
[0091] The results of laboratory turnaround time for urine specimens to complete non-albuminuria testing are as follows: Figure 3 shown.
[0092] Depend on Figure 3 It can be seen that when Pattern 3 is used for quantitative detection of urine samples, the time to complete non-albuminuria detection is the shortest, and the detection of urine samples can be completed most quickly to solve the problem of untimely detection.
[0093] The statistical results of the detection limit and linear range of urine total protein TPU and microalbumin MALB are as follows: Figure 4 shown.
[0094] Depend on Figure 4 It can be seen that when Pattern 3 is used for quantitative detection of urine samples, the detection linear range of TPU and ALBU is greatly expanded. Moreover, in actual operation, it is found that when Pattern 3 is used for quantitative detection of urine samples and completion of non-albuminuria identification, the re-examination rate of related high-value specimens is reduced from the original 12.94% to 0.84%, which greatly saves detection costs.
[0095] Example 3: Evaluation of non-albuminuria testing for diagnosis of metabolic syndrome
[0096] 1. Discovery of the spectrum of diseases associated with non-albuminuria:
[0097] This study used a cross-sectional research method to collect urine samples from 1,187 patients randomly collected from the First Affiliated Hospital of Zhengzhou University for 3 days. Among them, urine samples of 218 patients were identified as non-albuminuria, and 134 of the 218 patients had clear clinical disease diagnosis information. At the same time, cases of chronic kidney disease stage 5, renal dialysis arteriovenous fistula, ileal salpingostomy, colorectal cancer, multiple organ failure and liver cancer were excluded from the 134 patients with clinical disease diagnosis information, and finally 108 cases were included in the group.
[0098] All clinical diagnoses of the 108 cases were reviewed and analyzed according to the types and characteristics of the diseases, and a CICOS diagram was drawn to present them intuitively. Figure 5 shown.
[0099] Depend on Figure 5It can be seen that the proportion of cases with one of the pathogenic factors of metabolic syndrome, such as hypertension (Hyt), type 2 diabetes (T2DM), fatty liver disease (FLD) and hyperlipidemia (Hyl), was 57.9%; kidney-related diseases accounted for the highest proportion, at 59.8%. The overlap rate of the two diagnoses was 27.1%. Metabolic syndrome pathogenic factors and kidney factors are the two main causes of non-albuminuria-related diseases. In addition, multiple myeloma light chain proteinuria and kidney stone-related hematuria are independent of metabolic syndrome pathogenic factors and kidney factors.
[0100] The present invention re-collected 30 cases of metabolic syndrome diagnosed clinically by the metabolic syndrome diagnostic criteria and 31 normal healthy controls, and divided the samples into a metabolic syndrome group and a normal control group. Figure 5 It can be seen from the conclusion that non-albuminuria mainly exists in metabolic syndrome and kidney disease. Therefore, in actual research, when kidney disease is excluded, non-albuminuria patients can be diagnosed as metabolic syndrome. The non-albuminuria detection method recorded in Example 1 of the present invention is used to detect urine samples of the metabolic syndrome group and the normal control group, and the urine total protein TPU quantitative detection results, microalbumin MALB quantitative detection results and urine protein PRO qualitative detection results of the urine samples are obtained. The urine total protein TPU quantitative detection results, microalbumin MALB quantitative detection results and urine protein PRO qualitative detection results are brought into the trained machine learning model SVM to obtain the non-albuminuria detection results of urine samples of the metabolic syndrome group and the normal control group, and the ROC curve for diagnosing and distinguishing metabolic syndrome patients and healthy controls is prepared according to the non-albuminuria detection results of urine samples of the metabolic syndrome group and the normal control group. At the same time, for comparison, the present invention further prepares the ROC curve for diagnosing and distinguishing metabolic syndrome patients and healthy controls using the urine total protein TPU quantitative detection results, microalbumin MALB quantitative detection results, and urine protein PRO qualitative detection results of urine samples of the metabolic syndrome group and the normal control group. The ROC curve is as follows: Figure 6 The AUC value statistics of the ROC curve are shown in Table 3.
[0101] Table 3 Area under the curve for non-albuminuria diagnosis and differentiation of metabolic syndrome patients and healthy controls
[0102]
[0103] Test outcome variables: PRO, TPU and MALB represent the prediction of non-white blood cells using PRO, TPU and MALB test results respectively.
[0104] ROC curve obtained for proteinuria / metabolic syndrome. The statistic may be biased. a. Under nonparametric assumptions; b. Null hypothesis:
[0105] Area = 0.5.
[0106] Depend on Figure 6 As shown in Table 3, the non-albuminuria detection method of the present invention is used to detect non-albuminuria, and the ROC curve AUC value of the non-albuminuria detection result for the diagnosis of metabolic syndrome reaches 0.94, which is significantly higher than the AUC value of using urine protein PRO, urine total protein TPU, and microalbumin MALB detection results alone to diagnose and distinguish metabolic syndrome patients from healthy controls. Further statistical analysis of the area under the ROC curve of different test result variables, PRO / non-albuminuria detection result (prediction probability): p = <0.05, the difference is statistically significant, it can be considered that the diagnostic efficacy of the two is different; TPU / non-albuminuria detection result (prediction probability): p = <0.05, the difference is statistically significant, it can be considered that the diagnostic efficacy of the two is different; MALB / non-albuminuria detection result (prediction probability): p = <0.05, the difference is statistically significant, it can be considered that the diagnostic efficacy of the two is different. That is, the prediction model (prediction probability) of metabolic syndrome has significantly different diagnostic efficacy. The experimental results further demonstrate the value of the non-albuminuria diagnostic model of the present invention in distinguishing patients with metabolic syndrome from healthy controls in practical applications.
[0107] The above embodiments are intended to illustrate the essential content of the present invention, but are not intended to limit the protection scope of the present invention. Those skilled in the art should understand that the technical solution of the present invention can be modified or replaced by equivalents without departing from the essence and protection scope of the technical solution of the present invention.
Claims
1. A method for detecting non-albuminuria, characterized in that: The following steps are involved: S1: Performing a qualitative analysis on the urine sample to obtain a qualitative test result of the urine sample, wherein the qualitative test result includes urine pH, urine protein PRO and urine occult blood BLD; if the urine occult blood BLD in the qualitative test result is positive, the urine sample is judged to be suspected non-albuminuria; if the urine occult blood BLD in the qualitative test result is negative, executing step S2; S2: According to the urine qualitative test result obtained in step S1, the quantitative test mode of the urine sample is selected, and the urine sample is subjected to quantitative test of total urine protein TPU and microalbumin MALB according to the selected quantitative test mode to obtain the quantitative test results of total urine protein TPU and microalbumin MALB of the urine sample; then, whether the urine sample is non-albuminuria is jointly determined based on the quantitative test result of total urine protein TPU, the quantitative test result of microalbumin MALB and the qualitative test result of urine protein PRO.
2. The detection method according to claim 1, characterized in that: In step S2, the quantitative detection mode includes an original quantitative detection mode and a diluted quantitative detection mode; the original quantitative detection mode is: the urine sample does not need to be diluted, and the urine total protein TPU and microalbumin MALB are directly detected on the urine sample; the diluted quantitative detection mode is: the urine sample is first diluted, and then the urine total protein TPU and microalbumin MALB are quantitatively detected on the diluted urine sample; According to the urine pH, urine protein PRO and urine occult blood BLD test results obtained in step S1, the specific operation of selecting the quantitative detection mode of the urine sample is: If the urine occult blood BLD in the qualitative test result is negative and the pH is ≥8.0, the urine sample is subjected to quantitative detection of total urine protein TPU and microalbumin MALB using the original quantitative detection mode; If the urine occult blood BLD is negative, pH < 8.0 and PRO is negative in the qualitative test results, the urine sample is subjected to quantitative detection of total urine protein TPU and microalbumin MALB using the original quantitative detection mode; If the urine occult blood BLD is negative, pH < 8.0 and PRO 1+ in the qualitative test results, the urine sample is subjected to quantitative detection of total urine protein TPU and microalbumin MALB using the original quantitative detection mode; If the urine occult blood BLD is negative, pH is less than 8.0, and PRO is weakly positive in the qualitative test results, the urine sample is subjected to quantitative detection of total urine protein TPU and microalbumin MALB using the original quantitative detection mode; If the urine occult blood BLD is negative, pH is less than 8.0, and PRO is ≥2+ in the qualitative test results, the urine sample is quantitatively tested using a dilution quantitative test mode.
3. The detection method according to claim 2, characterized in that: If the urine occult blood BLD is negative, pH < 8.0 and PRO 2+ in the qualitative test results, the urine sample is diluted 3 times and then subjected to TPU quantitative test, and the urine sample is diluted 11 times and then subjected to MALB quantitative test; if the urine occult blood BLD is negative, pH < 8.0 and PRO ≥ 3+ in the qualitative test results, the urine sample is diluted 5 times and then subjected to TPU quantitative test, and the urine sample is diluted 20 times and then subjected to MALB quantitative test.
4. The detection method according to any one of claims 1 to 3, characterized in that: In step S2, the specific operation of jointly judging whether the urine sample is non-albuminuria based on the quantitative detection result of total urine protein TPU, the quantitative detection result of microalbumin MALB and the qualitative detection result of urine protein PRO is as follows: the quantitative detection result of total urine protein TPU, the quantitative detection result of microalbumin MALB and the qualitative detection result of urine protein PRO of the urine specimen are input into the trained machine learning model for detection and judgment, so as to obtain the judgment result of whether the urine sample is suspected non-albuminuria.
5. The detection method according to claim 4, characterized in that: The machine learning model is a support vector machine model, a decision tree model, a random forest model, an AdaBoost model or an XGBoost model; the qualitative analysis is to perform qualitative analysis on a urine sample using a urine test strip method; and the urine sample is random urine or 24-hour urine.
6. A non-albuminuria detection system, characterized in that: The detection system includes a laboratory information system, an IS system and a urine quantitative analysis system; the laboratory information system receives and stores the qualitative detection results of the urine sample by the urine qualitative analyzer, and sends the qualitative detection results to the IS system, and the qualitative detection results include urine pH, urine protein PRO and urine occult blood BLD; The IS system receives the qualitative test results of the urine sample sent by the laboratory information system, and analyzes the qualitative test results of the urine sample. If the urine occult blood BLD in the qualitative test results is positive, the IS system determines that the urine sample is suspected non-albuminuria, and sends the judgment result to the laboratory information system; if the urine occult blood BLD in the qualitative test results is negative, the IS system selects the quantitative test mode of the urine sample according to the qualitative test results of the urine sample, and sends the selected urine sample quantitative test mode instruction to the urine quantitative analysis system; the urine quantitative analysis system performs urine total protein TPU and microalbumin MALB quantitative tests on the urine sample according to the received urine sample quantitative test mode instruction, obtains the quantitative test results of urine total protein TPU and microalbumin MALB of the urine sample, and sends the urine total protein TPU and microalbumin MALB quantitative test results to the IS system; The IS system jointly determines whether the urine sample is non-albuminuria based on the quantitative detection results of total urine protein TPU, microalbumin MALB and urine protein PRO of the received urine sample, and sends the judgment result to the laboratory information system.
7. The non-albuminuria detection system according to claim 6, characterized in that: The specific operation of the IS system to jointly determine whether the urine sample is non-albuminuria based on the urine total protein TPU quantitative detection result, microalbumin MALB quantitative detection result, and urine protein PRO qualitative detection result of the received urine sample is as follows: the urine total protein TPU quantitative detection result, microalbumin MALB quantitative detection result, and urine protein PRO qualitative detection result of the urine specimen are input into the trained machine learning model installed in the IS system for detection and judgment, and the judgment result of whether the urine sample is suspected to be non-albuminuria is obtained; the urine quantitative analysis system includes A middleware platform and a urine quantitative biochemical analyzer, wherein the middleware platform is used to receive a urine sample quantitative detection mode instruction sent by an IS system, and control the urine quantitative biochemical analyzer to perform urine total protein TPU and microalbumin MALB quantitative detection on the urine sample according to the urine sample quantitative detection mode instruction; the urine quantitative biochemical analyzer is used to perform urine total protein TPU and microalbumin MALB quantitative detection on the urine sample according to the instruction of the middleware platform, and after the detection is completed, the urine total protein TPU and microalbumin MALB quantitative detection results of the urine sample are sent to the IS system.
8. The non-albuminuria detection system according to claim 7, characterized in that: The machine learning model is a support vector machine model, a decision tree model, a random forest model, an AdaBoost model or an XGBoost model.
9. Use of a detection reagent and / or device for detecting non-albuminuria in a urine sample in the preparation of a product for auxiliary diagnosis of metabolic syndrome.
10. The use according to claim 9, characterized in that: The detection reagent and / or instrument are detection reagents and / or instruments used when a urine sample is detected using any non-albuminuria detection method according to any one of claims 1 to 5.