A Metabolomics-Guided Machine Learning Model for Diagnosis and Differential Diagnosis of Diabetic Kidney Disease: A Dual-Center Study

A Metabolomics-Guided Machine Learning Model for Diagnosis and Differential Diagnosis of Diabetic Kidney Disease: A Dual-Center Study | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article A Metabolomics-Guided Machine Learning Model for Diagnosis and Differential Diagnosis of Diabetic Kidney Disease: A Dual-Center Study Xing Zhou, Luhan Li, Yingxin Wang, Tingting Kuai, Feng Wang, Dan Zhao, and 4 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7347937/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted 6 You are reading this latest preprint version Abstract Background Early diagnosis and timely intervention are critical for delaying the progression of diabetic kidney disease (DKD) to end-stage renal disease (ESRD). This study integrated metabolomics profiling with machine learning algorithms to comprehensively identify blood-based biomarkers—including 1,5-anhydroglucitol (1,5-AG) and multiple fatty acids—associated with DKD, and to develop predictive models for both diagnosis and differential diagnosis. Methods Clinical data and serum samples were collected from 1,038 patients with DKD, diabetes mellitus, or non-DKD chronic kidney disease (CKD) at the First Affiliated Hospital of Dalian Medical University. Concentrations of fatty acids and 1,5-AG were quantified by HPLC–MS/MS. Candidate biomarkers were screened using regression analyses. Diagnostic models for DKD were developed using four algorithms—binary logistic regression, random forest, decision tree, and naive Bayes—while a logistic regression model was applied to differentiate DKD from non-diabetic CKD. Model performance was assessed by the area under the receiver operating characteristic curve (AUC), sensitivity, and specificity. External validation was conducted in an independent cohort of 236 patients (DKD and diabetes without renal insufficiency) from the Second Affiliated Hospital. Results Four biomarkers—C16:0, C18:0, estimated glomerular filtration rate (eGFR), and glucose—were identified for DKD diagnosis. Logistic regression achieved the best performance, with AUCs of 0.920 (training), 0.879 (internal validation), and 0.881 (external validation). For differential diagnosis between DKD and non-diabetic CKD, five biomarkers—1,5-AG, glucose, C18:0, body mass index (BMI), and diastolic blood pressure (DBP)—yielded AUCs of 0.873 (training) and 0.812 (internal validation). Correlation analyses revealed that 1,5-AG was negatively associated with glucose and eGFR, but positively associated with serum creatinine, uric acid, and urea. In contrast, C14:0, C20:0, and C24:1 were positively correlated with glucose and eGFR, but negatively with serum creatinine. Multivariate analysis identified C24:1, C20:0, and 1,5-AG as independent risk factors for DKD progression. Conclusion Fatty acids C24:1 and C20:0, along with 1,5-AG, may independently increase the risk of DKD progression, and renal function appears to influence 1,5-AG levels. Both the diagnostic and differential diagnostic models demonstrated robust predictive performance for DKD in independent cohorts. 1 5-anhydroglucitol Diabetic kidney disease fatty acids machine learning diagnostic model Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 1 Background Diabetic kidney disease (DKD) is a major microvascular complication of diabetes mellitus and a leading cause of mortality and end-stage renal disease. Approximately 30% of patients with type 1 diabetes mellitus (T1DM) and 40% of those with type 2 diabetes mellitus (T2DM) develop chronic kidney disease (CKD)( 1 ). CKD, defined by structural and/or functional kidney abnormalities, has a prevalence of 13.1% among Chinese adults according to a 2021 meta‐analysis( 2 ). Studies from 2009–2010 and 2018–2019 identified diabetes (36%) and chronic nephritis (35%) as the primary causes of CKD, underscoring the increasing impact of diabetes on kidney disease( 3 ). As DKD progresses, accumulation of urinary toxins, oxidative stress, chronic inflammation, and acid-base imbalances contribute to abnormal insulin metabolism, insulin resistance, and worsened hyperglycemia. Declining renal function further impairs insulin clearance, predisposing patients to hypoglycemia and pronounced glucose fluctuations that exacerbate oxidative stress and kidney injury. Consequently, accurate blood glucose monitoring is essential in DKD management, employing techniques such as fingertip glucose testing, glycosylated hemoglobin A1c (HbA1c), and 1,5-anhydroglucitol (1,5-AG) measurement( 4 ). As a sensitive indicator reflecting short-term glucose fluctuations over 1–2 weeks, 1,5-AG was approved by the U.S. FDA in 2003 for glucose monitoring( 5 – 7 ). Although serum 1,5-AG levels have been correlated with glucose variability and microalbuminuria, the influence of renal function on 1,5-AG remains controversial( 8 – 10 ). The lipotoxicity hypothesis in DKD is supported by findings of large lipid deposits and intracellular lipid droplets in renal biopsies. Alterations in lipid metabolites, including non-esterified fatty acids (NEFA), have been observed, with elevated plasma free fatty acids noted in early DKD ( 11 ). Metabolomics, first introduced by Pauling in 1971 ( 12 ), enables comprehensive analysis of small molecules (< 1500 Da) in biological systems, offering insights into molecular pathways underlying disease progression. Both non-targeted and targeted metabolomics—often employing mass spectrometry (MS) coupled with separation techniques—are increasingly used for biomarker discovery in diabetes and its complications. In this study, high-performance liquid chromatography tandem mass spectrometry (HPLC-MS/MS) was employed to accurately quantify serum 1,5-AG and fatty acids. Machine learning (ML) techniques, such as logistic regression, decision trees, random forests, and naive Bayes, have been widely applied to develop predictive models in medical research ( 13 – 15 ). Although previous studies have reported correlations between lipid profiles and DKD, data based on mass spectrometry platforms remain limited. The present investigation integrates ML algorithms with precise MS detection of small-molecule metabolites to identify potential DKD biomarkers and construct models for diagnosis and differential diagnosis. This approach is anticipated to enhance the clinical diagnosis and treatment of DKD. 2 Materials and methods 2.1 Subjects 2.2 Clinical data Clinical data comprised gender, age, height, weight, history of hypertension, diastolic and systolic blood pressure, and body mass index (BMI). Laboratory parameters included serum creatinine (Scre), urea, uric acid (UA), glucose (Glu), glycated hemoglobin (HbA1c), urinary microalbumin (MAU), urine albumin-creatinine ratio (ACR), urinary creatinine (UCRE), high-density lipoprotein (HDL), low-density lipoprotein (LDL), total cholesterol (TC), triglycerides (TG), and estimated glomerular filtration rate (eGFR) calculated using the 2009 CKD-EPI formula. 2.3 Determination of Fatty Acids Nineteen serum fatty acids were quantified using HPLC-MS/MS (AB SCIEX Triple Quad™ 4500MD) with corresponding standards and internal standards. The targeted analytes included C14:0, C16:1, C16:0, C18:3, C18:2, C18:1, C18:0, C20:5, C20:4, C20:3, C20:2, C20:1, C20:0, C22:6, C22:5, C22:4, C22:0, C24:1, and C24:0. Additionally, eight indices were calculated: ω3/ω6, Triene/Tetraene, Total SFA, Total MUFA, Total PUFA, Tω3, Tω6, and Total FA. 2.4 Sample Preparation A 5 µL serum aliquot was transferred into a clean glass tube, followed by 50 µL of internal standard solution and 1000 µL of dissociation solution. After shaking at 1850 rpm for 10 s, the mixture was heated at 80°C for 20 min. Then, 80 µL of neutralization solution and 1 mL of extractant were added, and the sample was shaken at 1850 rpm for 5 min. Next, 700 µL of supernatant was evaporated under nitrogen at 50°C for 5 min, reconstituted in 400 µL of a methanol–acetonitrile solution, and shaken for 5 min. Finally, 100 µL of the supernatant was transferred to a 96-well plate for analysis. 2.5 Chromatographic Conditions Separation was performed on an ACQUITY UPLC BEH C18 (1.7 µm, Waters) column maintained at 40°C, with the injection chamber at 8°C and a flow rate of 0.3 mL/min. The mobile phase consisted of 0.01% ammonia in water (A) and an acetonitrile–isopropanol (1:1) mixture with 0.01% ammonia (B). Detection was carried out using negative electrospray ionization (ESI−) in multiple reaction monitoring (MRM) mode. 2.6 Statistical Analysis Data are expressed as mean ± SD for normally distributed variables, median (IQR) for non-normal data, and counts (percentages) for categorical variables. Group comparisons used χ² tests for categorical data, independent t-tests for normal continuous variables, and the Wilcoxon rank-sum test for non-normal data. Pearson or Spearman correlations were applied as appropriate. Analyses were conducted in R (V4.4.1) with significance set at p < 0.05. Variables with 20% missing were excluded. Fatty acid values were standardized with the “tableone” package. Data were divided into training and validation sets (7:3 ratio). Variables with one-way p < 0.05 entered multivariate regression, and potential DKD biomarkers were selected via stepwise backward regression. These biomarkers were modeled using random forest, logistic regression, decision tree, and naive Bayes. ROC curves evaluated model performance (AUC, sensitivity, specificity) to select the optimal algorithm. A nomogram was constructed using the rms package, and calibration curves along with the Hosmer-Lemeshow test assessed model fit. Finally, the diagnosis model was validated externally using data from the Second Hospital of Dalian Medical University. 3 Results 3.1 Modeling and Validation of DKD A total of 1038 patients from the First Hospital of Dalian Medical University were initially selected. After excluding cases with incomplete data or infections, 815 patients remained (271 with diabetes mellitus and 274 with diabetic kidney disease [DKD]). Significant differences (p<0.05; Supplementary Table S1) were observed between the diabetes mellitus and DKD groups in variables such as age, systolic blood pressure (SBP), 1,5-anhydroglucitol (1,5-AG), serum creatinine (Scre), urea, uric acid (UA), glucose (Glu), glycated hemoglobin (HbA1c), various fatty acids (e.g., C14:0, C16:0, C16:1, C18:0, C20:5, C20:3, C20:0, C22:0, C22:6, C22:5, C24:0), omega-3/omega-6 ratio, total saturated fatty acids (TotalSFA), total fatty acids (TotalFA), total omega-3 (Tω3), hypertension status, estimated glomerular filtration rate (eGFR), microalbuminuria (MAU), and albumin-to-creatinine ratio (ACR). The dataset was divided into training (70%) and validation (30%) sets. Univariate and multivariate analyses, using stepwise backward regression, identified four potential DKD biomarkers: C16:0, C18:0, eGFR, and Glu ( Figure 2，Table 1). Variance inflation factors (1.11, 1.24, 4.61, 4.34) were all below the threshold of 5 (19), indicating no significant multicollinearity. Diagnostic models were constructed using random forest, logistic regression, decision tree, and naive Bayes algorithms. Logistic regression was selected due to its stability, achieving an area under the receiver operating characteristic curve (AUC) of 0.920 (sensitivity 89.58%, specificity 81.77%) in the training set and 0.879 (sensitivity 81.01%, specificity 74.39%) in the validation set (Figure 3A-B， Table 2). A nomogram was constructed (Figure 3C) in which each variable in the diagnostic model was assigned a score; the total score corresponded to the predicted diagnostic outcome for DKD. Calibration curves confirmed the nomogram's reliability (Figure 3D). ROC analysis showed that the model achieved an AUC of 0.920 (Figure 3E). Decision curve analysis (DCA) demonstrated that the model significantly increased net benefit (NB) (Figure 3F). Internal validation yielded an AUC of 0.879 (Figure 3E), with the Hosmer-Lemeshow test indicating good fit (p = 0.181 > 0.05). Table 1 Stepwise regression analysis of diagnostic model for DKD Std.Error z OR 95%CI p β (Intercept) 0.788 6.944 238.203 57.476-1280.711 <0.01 5.473 Glu(mmol/L) 0.048 2.380 1.120 1.022-1.233 0.017 0.113 eGFR (ml/min*1.73m 2 ) 0.008 -8.994 0.929 0.913-0.943 <0.01 -0.074 C18:0(nmol/ml) 0.552 -4.192 0.099 0.032-0.282 <0.01 -2.313 C16:0(nmol/ml) 0.496 2.121 2.863 1.097-7.699 0.034 1.052 Abbreviations: eGFR, estimated glomerular filtration rate; Glu, glucose. Table 2 Comparison of diagnostic performance of constructing classification models based on different machine learning algorithms Model AUC Accuracy Precision Sensitivity Specificity Logistic Training 0.920 0.8568 0.8309 0.8958 0.8177 validation 0.879 0.7764 0.7529 0.8101 0.7439 RF Training 1 1.0000 1.0000 1.0000 1.0000 validation 0.866 0.8137 0.7882 0.8481 0.7805 DecisionTree Training 0.919 0.8880 0.8670 0.9167 0.8594 validation 0.841 0.7888 0.7647 0.8228 0.7561 NaiveBayes Training 0.877 0.7917 0.8146 0.7552 0.8281 validation 0.835 0.7267 0.7692 0.6329 0.8171 To compare the predictive performance of the four biomarker models with individual biomarkers, ROC curves were generated. The models outperformed individual biomarkers, with the model achieving an AUC of 0.920, significantly higher than eGFR (AUC 0.876, p < 0.05). The models also demonstrated better diagnosis of DKD than eGFR (Figure 4A). To further validate the DKD diagnostic model, 236 patients with diabetes and diabetic kidney disease were collected from the Second Affiliated Hospital of Dalian Medical University. After excluding cases with infections, tumors, liver disease, or incomplete data, 120 patients remained (60 with diabetes and 60 with DKD), forming the external validation cohort. Missing values for variables with 20% missing were excluded. Fatty acid data were standardized and analyzed using the “tableone” package (Supplementary Table S2). Significant differences (p < 0.05) were observed in age, systolic blood pressure, ACR, MAU, UCRE, eGFR, C18:0, C20:3, C22:6, C22:5, C22:0, C24:1, Triene/Tetraene, urea, UA, Scre, and hypertension history. In the external cohort, ROC analysis showed an AUC of 0.881, sensitivity of 0.867, and specificity of 0.750 (Figure 4B). Calibration curves (Figure 4C) and decision curve analysis (Figure 4D) confirmed the model’s reliability, with a Hosmer-Lemeshow test p-value of 0.620. 3.2 DKD differential diagnosis modeling A total of 1038 patients from the First Affiliated Hospital of Dalian Medical University were screened, and 744 met the inclusion criteria. Among them, 199 had non-DKD chronic kidney disease. After further controlling for gender, age, eGFR and Scre confounders, 199 patients with non-DKD chronic kidney disease (121 males and 78 females) and 156 patients with DKD (95 males and 61 females) were finally retained for differential diagnosis modeling. Significant differences (p<0.05) were observed between DKD and CKD patients in weight, BMI, SDP, 1,5-AG, TG, TC, Glu, and multiple fatty acids (C14:0, C16:1, C18:3, C18:2, C18:0, C20:5, C20:4, C20:1, C20:0, C22:6, C22:5, C22:4, ω3/ω6, TotalSFA , TotalPUFA, TotalFA, Tω3, Tω6, Triene /Tetraene). Additionally, TotalSFA , TotalPUFA, TotalFA, Tω3, Tω6, Triene /Tetraene ratio also differed significantly (p<0.05, Supplementary Table S3). The data were split into a training set (70%) and a validation set (30%). Univariate and multivariate analyses, along with stepwise backward regression, identified five potential biomarkers for diabetic kidney disease (DKD): 1,5-AG, Glu, C18:0, BMI, diastolic blood pressure(DBP) (Figure 5，Table 3). Multicollinearity was assessed using the variance inflation factor (VIF). The VIF values for these biomarkers were 1.05, 1.09, 1.06, 1.07 and 1.10, respectively. Given the VIF threshold of ≥ 5 (19), these results suggest that none of the biomarkers exhibited significant multicollinearity. A logistic regression-based diagnostic model was constructed using five diagnostic markers, and a column-line diagram was generated (Figure 6A). Each biomarker in the model corresponds to a score, summed to reflect DKD diagnostic ability. The ROC curve showed an AUC of 0.873, sensitivity of 0.801, and specificity of 0.722 (Figure 6B). The calibration curve confirmed the model’s reliability (Figure 6C). Decision curve analysis (DCA) indicated a significant net benefit (Figure 6D). The Hosmer-Lemeshow (H-L) test (p = 0.183>0.05) suggested a good fit. In the validation set, the model achieved an AUC of 0.812, sensitivity of 0.689, and specificity of 0.707 (Figure 6B), with an H-L test p-value of 0.057>0.05. These results demonstrate strong predictive performance in both training and validation sets. Table 3 Stepwise regression analysis of differential diagnosis model for DKD Std.Error z OR 95%CI p β (Intercept) 2.30 -1.62 0.11 0.02 0.00-2.08 -3.73 DBP 0.02 -3.00 <0.01 0.95 0.91-0.98 -0.05 C22:4(nmol/ml) 0.51 1.96 0.05 2.71 1.02-7.59 1.00 C20:3(nmol/ml) 0.44 -1.81 0.07 0.45 0.19-1.05 -0.79 C20:4(nmol/ml) 0.42 1.74 0.08 2.07 0.96-5.02 0.73 C18:0(nmol/ml) 0.24 -3.01 <0.01 0.48 0.29-0.76 -0.72 Glu(mmol/l) 0.17 4.13 <0.01 2.06 1.50-3.01 0.72 1,5-AG(μg/ml) 0.04 -4.70 <0.01 0.81 0.74-0.88 -0.21 age（years） 0.01 1.77 0.08 1.03 1.00-1.06 0.03 BMI (kg/m 2 ) 0.05 3.27 <0.01 1.18 1.07-1.31 0.17 Abbreviations: BMI, body mass index; DBP, diastolic blood pressure ;Glu,glucose; 1,5-AG, 1,5-anhydroglucitol. Table 4 logistic regression analysis Std.Error z p OR 95%CI β (Intercept) 0.22 -4.71 <0.01 0.35 0.22-0.53 -1.05 C24:1(nmol/ml) 0.22 3.62 <0.01 2.23 1.48-3.52 0.80 C20:0(nmol/ml) 0.21 3.49 <0.01 2.10 1.39-3.22 0.74 C20:5(nmol/ml) 0.20 -2.42 0.02 0.62 0.40-0.89 -0.48 C14:0(nmol/ml) 0.24 -4.84 <0.01 0.31 0.19-0.48 -1.16 1,5-AG(μg/ml) 0.05 3.13 <0.01 1.18 1.07-1.36 0.16 Abbreviations: 1,5-AG, 1,5-anhydroglucitol 3.3 Analysis of the correlation between 1,5-AG, fatty acids and DKD DKD was categorized into stages 1-5 according to eGFR, with stages 1-3 as early DKD and stages 4-5 as advanced DKD, of which 162 cases were in early stage, 97 males and 65 females, with a mean age of 62.17, and 112 cases were in advanced stage, 46 males and 66 females, with a mean age of 64.42, and 1,5-AG, UCRE, Scre, Urea, UA, HDL were found, Glu, HbA1c, C14:0, C16:1, C18:3, C18:0, C20:5, C20:3, C20:0, C22:6, C24:1, ω3/ω6, eGFR, Tω3, ACR, MAU, and hypertension were found to be statistically significant differences between the early and late stages of DKD（p<0.05, Supplementary Table S4）. 1,5-AG was found to be negatively correlated with Glu and eGFR, positively correlated with Scre、UA and UREA, and the correlation with lipids was not significant (Supplementary Table S5). Using binary logistic regression analysis, C24:1（p<0.01, OR 2.23，β 0.80）, C20:0（p<0.01, OR 2.10，β 0.74）and 1,5-AG（p<0.01, OR 1.18，β 0.16） were found to be possible independent risk factors for the progression of DKD (Table 4). 4 Discussion DKD is a common diabetes complication and a major cause of ESRD, severely impacting patient health and quality of life. Early identification of clinical biomarkers may slow or even halt DKD progression, and recent studies have focused on discovering such markers for timely intervention. 1,5-AG is a short-term glucose monitoring indicator with a structure similar to glucose, allowing it to freely cross cell membranes while remaining metabolically inert and stable (20). Under normal conditions, 99.9% of filtered 1,5-AG is reabsorbed in the renal tubules, maintaining stable blood levels. However, when blood glucose exceeds the renal threshold, glucose competitively inhibits 1,5-AG reabsorption, leading to its urinary excretion and a reduction in serum levels. Hasslacher et al. reported that in T2DM patients with CKD stages 1–3, 1,5-AG levels are influenced by both decreased renal function and glomerular hyperfiltration(21). Tavares et al. also found a correlation between 1,5-AG and the progression of massive proteinuria in DKD, independent of eGFR(22). In our study, 1,5-AG was negatively correlated with glucose and eGFR and positively correlated with serum creatinine and uric acid, consistent with previous findings (21). Although some studies report a strong negative correlation between serum 1,5-AG and total cholesterol (23), our data did not show a significant association. Fatty acids—key components of blood lipids—are classified as saturated (SFA), monounsaturated (MUFA), or polyunsaturated (PUFA) based on their hydrocarbon chains. Their role in DKD has recently attracted much attention. A meta-analysis of 3089 T2DM patients demonstrated that eGFR is negatively correlated with omega-6 fatty acids, linoleic acid, and non-HDL cholesterol, and positively correlated with certain HDL components (24). Moreover, lipid accumulation has been shown to induce tubular damage and glomerulosclerosis (25, 26). Case-control studies have associated palmitic (C16:0) and linoleic (C18:2) acids with DKD, suggesting their potential as biomarkers (27). Additionally, serum analysis of 326 type 1 diabetes patients revealed that MUFAs (e.g., C16:1, C18:1) and ω6 and ω9 fatty acids are linked to DKD development (28). Our findings further validate the role of fatty acids (C24:1, C20:0) as independent risk factors for the progression of DKD, underscoring their close relationship with the disease. Several studies have applied machine learning (ML) to evaluate DKD risk. Zou et al. compared four ML algorithms—gradient boosting, support vector machine, logistic regression, and random forest (RF)—and found that RF achieved the highest predictive performance (AUC 0.90), followed by SVM and GBM (AUC 0.88) and logistic regression (AUC 0.83), ultimately selecting RF for ESRD risk prediction (29). Dagliati et al. also used ML to predict diabetic complications (30). In another study, longitudinal data from 1365 participants (Chinese, Malaysian, and Indian, aged 40–80) indicated that a neural network achieved an AUC of 0.851—7.0% higher than logistic regression (AUC 0.795)—with sensitivities of 88.2% versus 73.0% and specificities of 65.9% versus 72.8% (29, 31) Jiang et al.’s meta-analysis of 20 cohorts with 41,271 T2DM patients reported an external validation AUC of 0.765 for early DKD risk prediction(32). Compared with previous studies, our DKD diagnostic model has several advantages. First, it integrates both macromolecular indices (e.g., total cholesterol, triglycerides) and small-molecule metabolites (e.g., 1,5-AG, fatty acids). Second, we developed not only an diagnosis model based on large samples but also a differential diagnostic model to distinguish DKD from non-DKD cases, reflecting differences in prognosis and treatment. Third, our model underwent independent external validation, achieving an AUC of 0.881. Fourth, although four ML algorithms were evaluated, logistic regression was ultimately chosen for its superior interpretability and clinical applicability. Our logistic model performed consistently in both the training set (AUC = 0.920, sensitivity 89.58%, specificity 81.77%) and the validation set (AUC = 0.879, sensitivity 81.01%, specificity 74.39%). Furthermore, combining metabolite data with ML improved prediction accuracy and helped uncover new risk factors. While RF models may yield higher AUCs with balanced datasets, their complexity and sensitivity to data imbalance limit clinical interpretability, making logistic regression the more practical choice. In summary, we developed a logistic regression–based model for DKD diagnostic that demonstrates high diagnostic efficacy (sensitivity 81.01%, specificity 74.39%, AUC 0.879) and robust external validation (AUC 0.881). The accompanying nomogram further aids clinicians in formulating diagnosis and treatment plans. However, limitations include the lack of pathological biopsies for all DKD patients, limited interpretability inherent to some ML models, and the exclusion of 24-hour urinary protein, glycosylated hemoglobin, urinary microalbumin, and ACR data. Future studies should expand sample sizes, include biopsy-confirmed cases, and integrate additional biomarkers to further optimize the model and enhance prediction accuracy. 5 Conclusion In conclusion, We developed an DKD diagnostic model using four biomarkers (C16:0, C18:0, eGFR, Glu) with AUCs of 0.920 (training), 0.879 (internal validation), and 0.881 (external validation). A logistic regression model for differentiating DKD from non-diabetic CKD using five biomarkers (1,5-AG, Glu, C18:0, BMI, DBP) achieved AUCs of 0.873 (training), and 0.812(validation). Notably, 1,5-AG was negatively correlated with eGFR and positively with creatinine and UA, while fatty acids C24:1, C20:0 and 1,5-AG may be independent risk factors. Our findings offer valuable guidance for diagnosis and differential DKD diagnosis. Abbreviations DKD diabetic kidney disease ESRD end-stage renal disease T2DM type 2 diabetes mellitus CKD chronic kidney disease AUC area under the receiver operating characteristic curve SBP blood pressure 1,5-AG 1,5-anhydroglucitol DBP diastolic blood pressure Scre serum creatinine UA uric acid Glu glucose HbA1c glycated hemoglobin HPLC-MS/MS high-performance liquid chromatography tandem mass spectrometry ML Machine learning eGFR estimated glomerular filtration rate DCA Decision curve analysis NB net benefit SFA saturated fatty acids MUFA monounsaturated fatty acids PUFA polyunsaturated fatty acids Declarations Data availability statement The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation. Author contributions XZ: Conceptualization, Data curation, Methodology, Writing-original draft. LL: Software, Data curation, Methodology. YW: Conceptualization, Methodology, Validation. TK: Validation, FW and DZ. Formal Analysis. MF and HC: Investigation. SL: Conceptualization, Methodology, Validation. PC: Conceptualization, Funding acquisition, Writing-review & editing. Funding information The author(s) declare financial support was received for the research, authorship, and/or publication of this article. This work was supported by the Liaoning Provincial Department of Education Basic Research Project，Development of a Mass Spectrometry-Based 1,5-Anhydroglucitol Project and Its Application in the Diagnosis and Treatment of Diabetes（NO.LJ212410161051）. Acknowledgments The authors sincerely thank all the doctors, nurses, and research staff at the First Affiliated Hospital of Dalian Medical University and the Second Affiliated Hospital of Dalian Medical University for their participation in this study. Conflict of interest The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. Publisher’s note All claims expressed in this article are solely those of the authors and do not necessarily reflect the views of their affiliated institutions, the publisher, the editors, or the reviewers. Any product evaluated in this article or any claim made by its manufacturer is neither guaranteed nor endorsed by the publisher. Ethics approval and consent to participate This study was reviewed and approved by the Institutional Ethical Review Boards of the First and Second Affiliated Hospitals of Dalian Medical University (Approval No. YJ-KS-KY-2024-715). The requirement for informed consent was waived by the Ethics Committees in accordance with the Measures for the Ethical Review of Biomedical Research Involving Humans (National Health and Family Planning Commission of the People’s Republic of China, 2016) and the principles of the Declaration of Helsinki (2013 revision), as the study involved minimal risk to participants and made use of existing clinical data without adversely affecting their rights or welfare. Clinical trial number Not applicable. Consent for publication Not applicable. Supplementary material The Supplementary Material for this article can be found online. References Bunny Y, Kai Y, Yan D. Advances in metabolomics in the early diagnosis of diabetic nephropathy. Western China University of Medicine. 2023;38(04):603-7. Xuelian B, Chia-Yi Z, Guoliang X, Xiaoqing C. Meta-analysis of the prevalence of chronic kidney disease in Chinese adults. Chinese Pharmaceutical Sciences. 2022;12(09):49-53. Yu W. Evolution of the etiology of chronic kidney disease in the elderly and the combined treatment of traditional Chinese and Western medicine [master]: Beijing University of Chinese Medicine; 2021. Diabetes CMA. China Guidelines for Clinical Application of Glucose Monitoring (2021 ). Chinese Diabetes Journal. 2021;13(10):926-48. Seok H, Huh JH, Kim HM, Lee BW, Kang ES, Lee HC, et al. 1,5-anhydroglucitol as a useful marker for assessing short-term glycemic excursions in type 1 diabetes. Diabetes Metab J. 2015;39(2):164-70. Wang Y, Yuan Y, Zhang Y, Lei C, Zhou Y, He J, et al. Serum 1,5-anhydroglucitol level as a screening tool for diabetes mellitus in a community-based population at high risk of diabetes. Acta Diabetol. 2017;54(5):425-31. Shuzhen W, Ting W, Yumei Z. Advances in clinical studies of 1,5-anhydroglucitol in diabetes mellitus. The New World of Diabetes. 2022;25(24):193-8. Tursun BB, Jinlian G, Yan J, Mingchen Z, Wenyan C. A study on the relationship between serum 1,5-anhydroglucitol levels and diabetic nephropathy. Journal of Xinjiang Medical University. 2016;39(06):750-2+6. Wei X, Weihua S, Zhao L. Correlation of serum 1,5-anhydroglucitol with urinary microclearbones in patients with type 2 diabetes mellitus. Journal of Clinical Laboratory Medicine. 2021;39(07):511-3. Bai Y, Yang R, Song Y, Wang Y. Serum 1,5-Anhydroglucitol Concentrations Remain Valid as a Glycemic Control Marker In Diabetes with Earlier Chronic Kidney Disease Stages. Exp Clin Endocrinol Diabetes. 2019;127(4):220-5. Han LD, Xia JF, Liang QL, Wang Y, Wang YM, Hu P, et al. Plasma esterified and non-esterified fatty acids metabolic profiling using gas chromatography-mass spectrometry and its application in the study of diabetic mellitus and diabetic nephropathy. Anal Chim Acta. 2011;689(1):85-91. Pauling L RAB, Teranishi R , Cary P. Quantitative Analysis of Urine Vapor and Breath by Gas-Liquid Partition Chromatography. Proceedings of the National Academy of Sciences of the United States of America 1971;68(10):2374-6. Choi RY, Coyner AS, Kalpathy-Cramer J, Chiang MF, Campbell JP. Introduction to Machine Learning, Neural Networks, and Deep Learning. Transl Vis Sci Technol. 2020;9(2):14. Zhu Y, Zhang Y, Yang M, Tang N, Liu L, Wu J, et al. Machine Learning-Based Predictive Modeling of Diabetic Nephropathy in Type 2 Diabetes Using Integrated Biomarkers: A Single-Center Retrospective Study. Diabetes Metab Syndr Obes. 2024;17:1987-97. Han H, Chen Y, Yang H, Cheng W, Zhang S, Liu Y, et al. Identification and Verification of Diagnostic Biomarkers for Glomerular Injury in Diabetic Nephropathy Based on Machine Learning Algorithms. Front Endocrinol (Lausanne). 2022;13:876-960. Kidney Disease: Improving Global Outcomes Diabetes Work G. KDIGO 2022 Clinical Practice Guideline for Diabetes Management in Chronic Kidney Disease. Kidney Int. 2022;102(5S):S1-S127. American Diabetes Association Professional Practice C. 2. Diagnosis and Classification of Diabetes: Standards of Care in Diabetes-2024. Diabetes Care. 2024;47(Suppl 1):S20-S42. Group KDIGOKCW. KDIGO 2024 Clinical Practice Guideline for the Evaluation and Management of Chronic Kidney Disease. Kidney International. 2024;105(4):S177-S314. Yue S, Li S, Huang X, Liu J, Hou X, Wang Y, et al. Construction and Validation of a Risk Prediction Model for Acute Kidney Injury in Patients Suffering from Septic Shock. Dis Markers. 2022;2022:9367873. Ying L, Ma X, Yin J, Wang Y, He X, Peng J, et al. The metabolism and transport of 1,5-anhydroglucitol in cells. Acta Diabetol. 2018;55(3):279-86. Hasslacher C, Kulozik F. Effect of renal function on serum concentration of 1,5-anhydroglucitol in type 2 diabetic patients in chronic kidney disease stages I-III: A comparative study with HbA1c and glycated albumin. J Diabetes. 2016;8(5):712-9. Tavares G, Venturini G, Padilha K, Zatz R, Pereira AC, Thadhani RI, et al. 1,5-Anhydroglucitol predicts CKD progression in macroalbuminuric diabetic kidney disease: results from non-targeted metabolomics. Metabolomics. 2018;14(4):39. Yuexing Y, Yao W. Analysis of the correlation between 1,5-anhydroglucitol and blood lipids in patients with type 2 diabetes mellitus. Journal of Southeast University. 2014;33(6):774-7. Tofte N, Vogelzangs N, Mook-Kanamori D, Brahimaj A, Nano J, Ahmadizar F, et al. Plasma Metabolomics Identifies Markers of Impaired Renal Function: A Meta-analysis of 3089 Persons with Type 2 Diabetes. J Clin Endocrinol Metab. 2020;105(7). Herman-Edelstein M, Scherzer P, Tobar A, Levi M, Gafter U. Altered renal lipid metabolism and renal lipid accumulation in human diabetic nephropathy. J Lipid Res. 2014;55(3):561-72. Yang W, Luo Y, Yang S, Zeng M, Zhang S, Liu J, et al. Ectopic lipid accumulation: potential role in tubular injury and inflammation in diabetic kidney disease. Clin Sci (Lond). 2018;132(22):2407-22. Zhang H, Zuo JJ, Dong SS, Lan Y, Wu CW, Mao GY, et al. Identification of Potential Serum Metabolic Biomarkers of Diabetic Kidney Disease: A Widely Targeted Metabolomics Study. J Diabetes Res. 2020;2020:3049098. Makinen VP, Tynkkynen T, Soininen P, Forsblom C, Peltola T, Kangas AJ, et al. Sphingomyelin is associated with kidney disease in type 1 diabetes (The FinnDiane Study). Metabolomics. 2012;8(3):369-75. Zou Y, Zhao L, Zhang J, Wang Y, Wu Y, Ren H, et al. Development and internal validation of machine learning algorithms for end-stage renal disease risk prediction model of people with type 2 diabetes mellitus and diabetic kidney disease. Ren Fail. 2022;44(1):562-70. Dagliati A, Marini S, Sacchi L, Cogni G, Teliti M, Tibollo V, et al. Machine Learning Methods to Predict Diabetes Complications. J Diabetes Sci Technol. 2018;12(2):295-302. Sabanayagam C, He F, Nusinovici S, Li J, Lim C, Tan G, et al. Prediction of diabetic kidney disease risk using machine learning models: A population-based cohort study of Asian adults. Elife. 2023;12. Jiang W, Wang J, Shen X, Lu W, Wang Y, Li W, et al. Establishment and Validation of a Risk Prediction Model for Early Diabetic Kidney Disease Based on a Systematic Review and Meta-Analysis of 20 Cohorts. Diabetes Care. 2020;43(4):925-33. Additional Declarations No competing interests reported. Supplementary Files Supplementarymaterial.docx Cite Share Download PDF Status: Under Review Version 1 posted Reviewers agreed at journal 24 Sep, 2025 Reviewers invited by journal 15 Sep, 2025 Editor assigned by journal 10 Sep, 2025 Editor invited by journal 19 Aug, 2025 Submission checks completed at journal 19 Aug, 2025 First submitted to journal 19 Aug, 2025 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-7347937","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":516438981,"identity":"ad53e461-9d39-4815-b690-ff0bdb2360d7","order_by":0,"name":"Xing Zhou","email":"","orcid":"","institution":"First Affiliated Hospital of Dalian Medical University","correspondingAuthor":false,"prefix":"","firstName":"Xing","middleName":"","lastName":"Zhou","suffix":""},{"id":516438982,"identity":"693ecd00-670f-4245-90ad-f0f5b78c9d40","order_by":1,"name":"Luhan Li","email":"","orcid":"","institution":"First Affiliated Hospital of Dalian Medical University","correspondingAuthor":false,"prefix":"","firstName":"Luhan","middleName":"","lastName":"Li","suffix":""},{"id":516438983,"identity":"b66cb1d6-448c-4852-96f4-cabd5669a34f","order_by":2,"name":"Yingxin Wang","email":"","orcid":"","institution":"Second Affiliated Hospital of Dalian Medical University","correspondingAuthor":false,"prefix":"","firstName":"Yingxin","middleName":"","lastName":"Wang","suffix":""},{"id":516438984,"identity":"1d4dfcfc-8f94-4f5a-a42a-c3032b626051","order_by":3,"name":"Tingting Kuai","email":"","orcid":"","institution":"Dalian Municipal Central Hospital affiliated with Dalian University of Technology","correspondingAuthor":false,"prefix":"","firstName":"Tingting","middleName":"","lastName":"Kuai","suffix":""},{"id":516438985,"identity":"c9cb9836-fd92-42fb-9adf-e47a57188583","order_by":4,"name":"Feng Wang","email":"","orcid":"","institution":"First Affiliated Hospital of Dalian Medical University","correspondingAuthor":false,"prefix":"","firstName":"Feng","middleName":"","lastName":"Wang","suffix":""},{"id":516438986,"identity":"baf1e221-88c6-4ce3-9bcb-b81dbd00c7aa","order_by":5,"name":"Dan Zhao","email":"","orcid":"","institution":"First Affiliated Hospital of Dalian Medical University","correspondingAuthor":false,"prefix":"","firstName":"Dan","middleName":"","lastName":"Zhao","suffix":""},{"id":516438989,"identity":"e7de7a81-640e-4979-ad6f-b6b34edfebef","order_by":6,"name":"Mengjie Feng","email":"","orcid":"","institution":"First Affiliated Hospital of Dalian Medical University","correspondingAuthor":false,"prefix":"","firstName":"Mengjie","middleName":"","lastName":"Feng","suffix":""},{"id":516438990,"identity":"0f3fcde7-4674-4909-9e7b-8ea1ff405255","order_by":7,"name":"Haoqian Cui","email":"","orcid":"","institution":"First Affiliated Hospital of Dalian Medical University","correspondingAuthor":false,"prefix":"","firstName":"Haoqian","middleName":"","lastName":"Cui","suffix":""},{"id":516438991,"identity":"f71963a5-fa89-46a9-b874-5ad813d061b3","order_by":8,"name":"Shuxin Liu","email":"","orcid":"","institution":"Dalian Municipal Central Hospital affiliated with Dalian University of Technology","correspondingAuthor":false,"prefix":"","firstName":"Shuxin","middleName":"","lastName":"Liu","suffix":""},{"id":516438992,"identity":"5f49e32b-eba3-4b6f-9523-00f189cd60ba","order_by":9,"name":"Penglong Cao","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA7klEQVRIiWNgGAWjYBADGSBmfMADZicQpwWkmNmAZC1sEkRp4W8/fOzBxx21PPzS7dcq3uYcZuBnzzFg+LkDtxaJM2nphjPPHOeRnHOm7ObcbYcZJHveGDD2nsGtxYAhx0yat+0Yj8GNnLTbvEAtQIYBM2MbHi38b8yk/0K1FIO02BPUIgG0hbGtBqgl/Rgz2BYJAlokbjxLN+xtO8AjOSOHWXLutnQeiTPPCg724tHC35987MHPtjo5fon0hx/ebrOW429P3ggUwa0FCNiA+DAQ8xiAeOCoOYBXA0RLHRCzPyCgcBSMglEwCkYqAADBj1ANhG8uiAAAAABJRU5ErkJggg==","orcid":"","institution":"First Affiliated Hospital of Dalian Medical University","correspondingAuthor":true,"prefix":"","firstName":"Penglong","middleName":"","lastName":"Cao","suffix":""}],"badges":[],"createdAt":"2025-08-11 15:38:17","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-7347937/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7347937/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":91897711,"identity":"ec292718-5601-4be1-862e-b9cca4dd1240","added_by":"auto","created_at":"2025-09-22 18:54:07","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":35871614,"visible":true,"origin":"","legend":"","description":"","filename":"RevisedManuscript.docx","url":"https://assets-eu.researchsquare.com/files/rs-7347937/v1/8c80ecb7c29fec2ee805d66b.docx"},{"id":91897696,"identity":"634516ca-42ed-4ba9-a4f7-bb6577e23e1c","added_by":"auto","created_at":"2025-09-22 18:54:07","extension":"json","order_by":1,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":11540,"visible":true,"origin":"","legend":"","description":"","filename":"1d34d2a93bd24060bea8742633cd9fae.json","url":"https://assets-eu.researchsquare.com/files/rs-7347937/v1/f0e60da92b405a67ef9d2a20.json"},{"id":91897692,"identity":"fcf36816-4399-448a-b441-104a39de0b5d","added_by":"auto","created_at":"2025-09-22 18:54:07","extension":"docx","order_by":2,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":48942,"visible":true,"origin":"","legend":"","description":"","filename":"Supplementarymaterial.docx","url":"https://assets-eu.researchsquare.com/files/rs-7347937/v1/20818f3aa268adc071639f52.docx"},{"id":91897694,"identity":"fbbef4b9-4de6-4194-94f1-453d625de2a6","added_by":"auto","created_at":"2025-09-22 18:54:07","extension":"xml","order_by":3,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":109684,"visible":true,"origin":"","legend":"","description":"","filename":"1d34d2a93bd24060bea8742633cd9fae1enriched.xml","url":"https://assets-eu.researchsquare.com/files/rs-7347937/v1/b96141ae3e5599793dbe9259.xml"},{"id":91897700,"identity":"072ad46a-f449-418f-81f4-2869f9f9b1d6","added_by":"auto","created_at":"2025-09-22 18:54:07","extension":"jpeg","order_by":6,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":1222899,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage1.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7347937/v1/42434d2c3aa6812175974f23.jpeg"},{"id":91898033,"identity":"0d9fbbb0-3331-4b63-92c2-445e979e2dea","added_by":"auto","created_at":"2025-09-22 19:02:07","extension":"jpeg","order_by":7,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":587000,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage2.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7347937/v1/4d085d994c2c236d2eb20778.jpeg"},{"id":91898257,"identity":"d8126580-fea8-4f74-9fb1-903075084964","added_by":"auto","created_at":"2025-09-22 19:10:07","extension":"jpeg","order_by":8,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":6196199,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage3.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7347937/v1/ea6f8ad89247538290b7a953.jpeg"},{"id":91898043,"identity":"c774a2bb-9204-4fd8-a3f6-7e428d3f2086","added_by":"auto","created_at":"2025-09-22 19:02:07","extension":"jpeg","order_by":9,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":6258033,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage4.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7347937/v1/16dbd0155c43938464836ea7.jpeg"},{"id":91898037,"identity":"48b4dcc2-f214-4b80-9fe3-7210318331a6","added_by":"auto","created_at":"2025-09-22 19:02:07","extension":"jpeg","order_by":10,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":573251,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage5.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7347937/v1/9da4d402ba359769a9a306c8.jpeg"},{"id":91898253,"identity":"fac57dc5-cbcf-411b-9819-a4a009c23d12","added_by":"auto","created_at":"2025-09-22 19:10:07","extension":"jpeg","order_by":11,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":363904,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage6.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7347937/v1/a73f1edf39675a61ef651ec5.jpeg"},{"id":91898256,"identity":"771e9f79-122c-4db1-b3dd-28ed67f361df","added_by":"auto","created_at":"2025-09-22 19:10:07","extension":"jpeg","order_by":12,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":1016894,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage7.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7347937/v1/8bead10c3df9f852790fe694.jpeg"},{"id":91898035,"identity":"77e0bb38-afed-44f9-bc82-fe6337e47d28","added_by":"auto","created_at":"2025-09-22 19:02:07","extension":"png","order_by":13,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":360369,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-7347937/v1/89814c6202162ffdefc2f63e.png"},{"id":91898251,"identity":"07e7ccd4-efff-4248-92e6-56f150c0910b","added_by":"auto","created_at":"2025-09-22 19:10:07","extension":"png","order_by":14,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":110034,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-7347937/v1/512bacc3ff85f44612f08815.png"},{"id":91897702,"identity":"ccffa826-0ea7-49a4-83a4-485a180caf7d","added_by":"auto","created_at":"2025-09-22 18:54:07","extension":"png","order_by":15,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":1019307,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-7347937/v1/69740d3f59766caae1fc9dd0.png"},{"id":91897704,"identity":"dd74997a-c379-4952-936e-db37e87545ae","added_by":"auto","created_at":"2025-09-22 18:54:07","extension":"png","order_by":16,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":538495,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage4.png","url":"https://assets-eu.researchsquare.com/files/rs-7347937/v1/369df5bd8b5592a4b1c3434c.png"},{"id":91898393,"identity":"7c5351b3-ba3e-46a9-b2aa-22b1a9b26d88","added_by":"auto","created_at":"2025-09-22 19:18:07","extension":"png","order_by":17,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":343746,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage5.png","url":"https://assets-eu.researchsquare.com/files/rs-7347937/v1/957bc74e9123da03dd150d1a.png"},{"id":91898038,"identity":"c53ba511-626f-409d-b158-c5807b2dfb17","added_by":"auto","created_at":"2025-09-22 19:02:07","extension":"png","order_by":18,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":91893,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage6.png","url":"https://assets-eu.researchsquare.com/files/rs-7347937/v1/217941b5f386a722099e7a5c.png"},{"id":91897707,"identity":"f2ebd0d4-b605-40c5-8643-6ef277d3b63e","added_by":"auto","created_at":"2025-09-22 18:54:07","extension":"png","order_by":19,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":138457,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage7.png","url":"https://assets-eu.researchsquare.com/files/rs-7347937/v1/dc1ef40a8ccdd96af609f807.png"},{"id":91897714,"identity":"8215fd22-daf7-4c72-bd19-8f853b4b48db","added_by":"auto","created_at":"2025-09-22 18:54:07","extension":"xml","order_by":20,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":107019,"visible":true,"origin":"","legend":"","description":"","filename":"1d34d2a93bd24060bea8742633cd9fae1structuring.xml","url":"https://assets-eu.researchsquare.com/files/rs-7347937/v1/8483ba61d70c5e7b1155b98e.xml"},{"id":91897709,"identity":"698f400a-f05b-44d5-8ddb-a32c8da73d60","added_by":"auto","created_at":"2025-09-22 18:54:07","extension":"html","order_by":21,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":117600,"visible":true,"origin":"","legend":"","description":"","filename":"earlyproof.html","url":"https://assets-eu.researchsquare.com/files/rs-7347937/v1/a87f230e6da38cfd510ed520.html"},{"id":91897689,"identity":"beb4da98-d3b3-4588-89b4-35202d3616dc","added_by":"auto","created_at":"2025-09-22 18:54:07","extension":"jpg","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":521662,"visible":true,"origin":"","legend":"\u003cp\u003eFlowchart of the sample selection\u003c/p\u003e","description":"","filename":"Picture1.jpg","url":"https://assets-eu.researchsquare.com/files/rs-7347937/v1/3fcf369a5d2dd0cddce053a4.jpg"},{"id":91897687,"identity":"85322b41-3b7e-4a46-8659-661cd9915212","added_by":"auto","created_at":"2025-09-22 18:54:07","extension":"jpg","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":28723,"visible":true,"origin":"","legend":"\u003cp\u003eModel variable screening. (A、B) Single factor forest map and (C) multi-factor forest map.\u003c/p\u003e","description":"","filename":"Picture2.jpg","url":"https://assets-eu.researchsquare.com/files/rs-7347937/v1/4cab06796baea3ca612e6130.jpg"},{"id":91898032,"identity":"22756674-fc26-4985-82e8-68d0dc86c684","added_by":"auto","created_at":"2025-09-22 19:02:07","extension":"jpg","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":51643,"visible":true,"origin":"","legend":"\u003cp\u003eComparison of ROC for 4 machine learning training and validation sets. (A) is the training set ROC comparison and (B) is the validation set ROC comparison. Construction of an diagnostic model for DKD. (C) DKD diagnostic model column line diagram based on 4 diagnostic markers. (D, E) Calibration curve (D) and ROC curve (E) are used to assess the diagnostic effect of the DKD diagnostic nomogram. (F) DCA description of NB assessment results.\u003c/p\u003e","description":"","filename":"Picture3.jpg","url":"https://assets-eu.researchsquare.com/files/rs-7347937/v1/d64f1fc8529c1ecb25b4d2ba.jpg"},{"id":91898392,"identity":"7982dc09-5061-4d42-a54a-cd413c65c5cd","added_by":"auto","created_at":"2025-09-22 19:18:07","extension":"jpg","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":542880,"visible":true,"origin":"","legend":"\u003cp\u003eInternal validation (A) and external validation (B) of the DKD diagnosis models. Calibration curves (C) and ROC curves (A) are used to assess the diagnostic effectiveness of the DKD diagnostic column chart. (D) DCA illustrates the results of NB assessment.\u003c/p\u003e","description":"","filename":"Picture4.jpg","url":"https://assets-eu.researchsquare.com/files/rs-7347937/v1/882323a6a027ff7775a11428.jpg"},{"id":91898034,"identity":"5e3e9797-945e-4e9d-9fc6-aaba7093977c","added_by":"auto","created_at":"2025-09-22 19:02:07","extension":"jpg","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":27680,"visible":true,"origin":"","legend":"\u003cp\u003eModel variable screening. (A、B) Single factor forest map and (C) multi-factor forest map.\u003c/p\u003e","description":"","filename":"Picture5.jpg","url":"https://assets-eu.researchsquare.com/files/rs-7347937/v1/72fb2ee212bc4f96ea2d72e8.jpg"},{"id":91897691,"identity":"a9c81c71-6dd3-436f-8244-69c0dcc87245","added_by":"auto","created_at":"2025-09-22 18:54:07","extension":"jpg","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":31681,"visible":true,"origin":"","legend":"\u003cp\u003eConstruction of a differential diagnostic model for DKD. (A) DKD diagnostic model column line diagram based on 5 diagnostic markers. (B, C) ROC curve (B) and Calibration curve (C) and are used to assess the diagnostic effect of the DKD diagnostic column line diagram. (D) DCA illustrating the results of NB assessment.\u003c/p\u003e","description":"","filename":"Picture6.jpg","url":"https://assets-eu.researchsquare.com/files/rs-7347937/v1/2d162a80281f0ddf6c6b907d.jpg"},{"id":91898662,"identity":"fb0d85df-6303-4c96-8c18-a515032f869f","added_by":"auto","created_at":"2025-09-22 19:26:07","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":2055331,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7347937/v1/416c3d72-bcf2-4672-9061-80b4621e476a.pdf"},{"id":91897690,"identity":"3b36e303-00ac-499c-a045-138f9a65e4a5","added_by":"auto","created_at":"2025-09-22 18:54:07","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"supplement","size":48942,"visible":true,"origin":"","legend":"","description":"","filename":"Supplementarymaterial.docx","url":"https://assets-eu.researchsquare.com/files/rs-7347937/v1/c370f5f61e4fdf3e57711eed.docx"}],"financialInterests":"No competing interests reported.","formattedTitle":"A Metabolomics-Guided Machine Learning Model for Diagnosis and Differential Diagnosis of Diabetic Kidney Disease: A Dual-Center Study","fulltext":[{"header":"1 Background","content":"\u003cp\u003eDiabetic kidney disease (DKD) is a major microvascular complication of diabetes mellitus and a leading cause of mortality and end-stage renal disease. Approximately 30% of patients with type 1 diabetes mellitus (T1DM) and 40% of those with type 2 diabetes mellitus (T2DM) develop chronic kidney disease (CKD)(\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e). CKD, defined by structural and/or functional kidney abnormalities, has a prevalence of 13.1% among Chinese adults according to a 2021 meta‐analysis(\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e). Studies from 2009\u0026ndash;2010 and 2018\u0026ndash;2019 identified diabetes (36%) and chronic nephritis (35%) as the primary causes of CKD, underscoring the increasing impact of diabetes on kidney disease(\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e). As DKD progresses, accumulation of urinary toxins, oxidative stress, chronic inflammation, and acid-base imbalances contribute to abnormal insulin metabolism, insulin resistance, and worsened hyperglycemia. Declining renal function further impairs insulin clearance, predisposing patients to hypoglycemia and pronounced glucose fluctuations that exacerbate oxidative stress and kidney injury. Consequently, accurate blood glucose monitoring is essential in DKD management, employing techniques such as fingertip glucose testing, glycosylated hemoglobin A1c (HbA1c), and 1,5-anhydroglucitol (1,5-AG) measurement(\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e). As a sensitive indicator reflecting short-term glucose fluctuations over 1\u0026ndash;2 weeks, 1,5-AG was approved by the U.S. FDA in 2003 for glucose monitoring(\u003cspan additionalcitationids=\"CR6\" citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e). Although serum 1,5-AG levels have been correlated with glucose variability and microalbuminuria, the influence of renal function on 1,5-AG remains controversial(\u003cspan additionalcitationids=\"CR9\" citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e).\u003c/p\u003e\u003cp\u003eThe lipotoxicity hypothesis in DKD is supported by findings of large lipid deposits and intracellular lipid droplets in renal biopsies. Alterations in lipid metabolites, including non-esterified fatty acids (NEFA), have been observed, with elevated plasma free fatty acids noted in early DKD (\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e). Metabolomics, first introduced by Pauling in 1971 (\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e), enables comprehensive analysis of small molecules (\u0026lt;\u0026thinsp;1500 Da) in biological systems, offering insights into molecular pathways underlying disease progression. Both non-targeted and targeted metabolomics\u0026mdash;often employing mass spectrometry (MS) coupled with separation techniques\u0026mdash;are increasingly used for biomarker discovery in diabetes and its complications. In this study, high-performance liquid chromatography tandem mass spectrometry (HPLC-MS/MS) was employed to accurately quantify serum 1,5-AG and fatty acids.\u003c/p\u003e\u003cp\u003eMachine learning (ML) techniques, such as logistic regression, decision trees, random forests, and naive Bayes, have been widely applied to develop predictive models in medical research (\u003cspan additionalcitationids=\"CR14\" citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e). Although previous studies have reported correlations between lipid profiles and DKD, data based on mass spectrometry platforms remain limited. The present investigation integrates ML algorithms with precise MS detection of small-molecule metabolites to identify potential DKD biomarkers and construct models for diagnosis and differential diagnosis. This approach is anticipated to enhance the clinical diagnosis and treatment of DKD.\u003c/p\u003e"},{"header":"2 Materials and methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e\u003ch2\u003e2.1 Subjects\u003c/h2\u003e\u003cp\u003e\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec4\" class=\"Section2\"\u003e\u003ch2\u003e2.2 Clinical data\u003c/h2\u003e\u003cp\u003eClinical data comprised gender, age, height, weight, history of hypertension, diastolic and systolic blood pressure, and body mass index (BMI). Laboratory parameters included serum creatinine (Scre), urea, uric acid (UA), glucose (Glu), glycated hemoglobin (HbA1c), urinary microalbumin (MAU), urine albumin-creatinine ratio (ACR), urinary creatinine (UCRE), high-density lipoprotein (HDL), low-density lipoprotein (LDL), total cholesterol (TC), triglycerides (TG), and estimated glomerular filtration rate (eGFR) calculated using the 2009 CKD-EPI formula.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec5\" class=\"Section2\"\u003e\u003ch2\u003e2.3 Determination of Fatty Acids\u003c/h2\u003e\u003cp\u003eNineteen serum fatty acids were quantified using HPLC-MS/MS (AB SCIEX Triple Quad\u0026trade; 4500MD) with corresponding standards and internal standards. The targeted analytes included C14:0, C16:1, C16:0, C18:3, C18:2, C18:1, C18:0, C20:5, C20:4, C20:3, C20:2, C20:1, C20:0, C22:6, C22:5, C22:4, C22:0, C24:1, and C24:0. Additionally, eight indices were calculated: ω3/ω6, Triene/Tetraene, Total SFA, Total MUFA, Total PUFA, Tω3, Tω6, and Total FA.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec6\" class=\"Section2\"\u003e\u003ch2\u003e2.4 Sample Preparation\u003c/h2\u003e\u003cp\u003eA 5 \u0026micro;L serum aliquot was transferred into a clean glass tube, followed by 50 \u0026micro;L of internal standard solution and 1000 \u0026micro;L of dissociation solution. After shaking at 1850 rpm for 10 s, the mixture was heated at 80\u0026deg;C for 20 min. Then, 80 \u0026micro;L of neutralization solution and 1 mL of extractant were added, and the sample was shaken at 1850 rpm for 5 min. Next, 700 \u0026micro;L of supernatant was evaporated under nitrogen at 50\u0026deg;C for 5 min, reconstituted in 400 \u0026micro;L of a methanol\u0026ndash;acetonitrile solution, and shaken for 5 min. Finally, 100 \u0026micro;L of the supernatant was transferred to a 96-well plate for analysis.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec7\" class=\"Section2\"\u003e\u003ch2\u003e2.5 Chromatographic Conditions\u003c/h2\u003e\u003cp\u003eSeparation was performed on an ACQUITY UPLC BEH C18 (1.7 \u0026micro;m, Waters) column maintained at 40\u0026deg;C, with the injection chamber at 8\u0026deg;C and a flow rate of 0.3 mL/min. The mobile phase consisted of 0.01% ammonia in water (A) and an acetonitrile\u0026ndash;isopropanol (1:1) mixture with 0.01% ammonia (B). Detection was carried out using negative electrospray ionization (ESI\u0026minus;) in multiple reaction monitoring (MRM) mode.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec8\" class=\"Section2\"\u003e\u003ch2\u003e2.6 Statistical Analysis\u003c/h2\u003e\u003cp\u003eData are expressed as mean\u0026thinsp;\u0026plusmn;\u0026thinsp;SD for normally distributed variables, median (IQR) for non-normal data, and counts (percentages) for categorical variables. Group comparisons used χ\u0026sup2; tests for categorical data, independent t-tests for normal continuous variables, and the Wilcoxon rank-sum test for non-normal data. Pearson or Spearman correlations were applied as appropriate. Analyses were conducted in R (V4.4.1) with significance set at p\u0026thinsp;\u0026lt;\u0026thinsp;0.05. Variables with \u0026lt;\u0026thinsp;20% missing data were imputed using the \u0026ldquo;mice\u0026rdquo; package; those with \u0026gt;\u0026thinsp;20% missing were excluded. Fatty acid values were standardized with the \u0026ldquo;tableone\u0026rdquo; package. Data were divided into training and validation sets (7:3 ratio). Variables with one-way p\u0026thinsp;\u0026lt;\u0026thinsp;0.05 entered multivariate regression, and potential DKD biomarkers were selected via stepwise backward regression. These biomarkers were modeled using random forest, logistic regression, decision tree, and naive Bayes. ROC curves evaluated model performance (AUC, sensitivity, specificity) to select the optimal algorithm. A nomogram was constructed using the rms package, and calibration curves along with the Hosmer-Lemeshow test assessed model fit. Finally, the diagnosis model was validated externally using data from the Second Hospital of Dalian Medical University.\u003c/p\u003e\u003c/div\u003e"},{"header":"3 Results","content":"\u003cp\u003e\u003cstrong\u003e3.1 Modeling and Validation of DKD\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eA total of 1038 patients from the First Hospital of Dalian Medical University were initially selected. After excluding cases with incomplete data or infections, 815 patients remained (271 with diabetes mellitus and 274 with diabetic kidney disease [DKD]). Significant differences (p\u0026lt;0.05; Supplementary Table S1) were observed between the diabetes mellitus and DKD groups in variables such as age, systolic blood pressure (SBP), 1,5-anhydroglucitol (1,5-AG), serum creatinine (Scre), urea, uric acid (UA), glucose (Glu), glycated hemoglobin (HbA1c), various fatty acids (e.g., C14:0, C16:0, C16:1, C18:0, C20:5, C20:3, C20:0, C22:0, C22:6, C22:5, C24:0), omega-3/omega-6 ratio, total saturated fatty acids (TotalSFA), total fatty acids (TotalFA), total omega-3 (T\u0026omega;3), hypertension status, estimated glomerular filtration rate (eGFR), microalbuminuria (MAU), and albumin-to-creatinine ratio (ACR).\u003c/p\u003e\n\u003cp\u003eThe dataset was divided into training (70%) and validation (30%) sets. Univariate and multivariate analyses, using stepwise backward regression, identified four potential DKD biomarkers: C16:0, C18:0, eGFR, and Glu ( Figure 2，Table 1). Variance inflation factors (1.11, 1.24, 4.61, 4.34) were all below the threshold of 5 (19), indicating no significant multicollinearity.\u003c/p\u003e\n\u003cp\u003eDiagnostic models were constructed using random forest, logistic regression, decision tree, and naive Bayes algorithms. Logistic regression was selected due to its stability, achieving an area under the receiver operating characteristic curve (AUC) of 0.920 (sensitivity 89.58%, specificity 81.77%) in the training set and 0.879 (sensitivity 81.01%, specificity 74.39%) in the validation set (Figure 3A-B， Table 2). A nomogram was constructed (Figure 3C) in which each variable in the diagnostic model was assigned a score; the total score corresponded to the predicted diagnostic outcome for DKD. Calibration curves confirmed the nomogram\u0026apos;s reliability (Figure 3D). ROC analysis showed that the model achieved an AUC of 0.920 (Figure 3E). Decision curve analysis (DCA) demonstrated that the model significantly increased net benefit (NB) (Figure 3F). Internal validation yielded an AUC of 0.879 (Figure 3E), with the Hosmer-Lemeshow test indicating good fit (p = 0.181 \u0026gt; 0.05).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTable 1\u003c/strong\u003e Stepwise regression analysis of diagnostic model for DKD\u003c/p\u003e\n\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\" align=\"left\" width=\"100%\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 114px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 82px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eStd.Error\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 65px;\"\u003e\n \u003cp\u003e\u003cstrong\u003ez\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 70px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eOR\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 128px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e95%CI\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 51px;\"\u003e\n \u003cp\u003e\u003cstrong\u003ep\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 59px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026beta;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 114px;\"\u003e\n \u003cp\u003e(Intercept)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 82px;\"\u003e\n \u003cp\u003e0.788\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 65px;\"\u003e\n \u003cp\u003e6.944\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 70px;\"\u003e\n \u003cp\u003e238.203\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 128px;\"\u003e\n \u003cp\u003e57.476-1280.711\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 51px;\"\u003e\n \u003cp\u003e\u0026lt;0.01\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 59px;\"\u003e\n \u003cp\u003e5.473\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 114px;\"\u003e\n \u003cp\u003eGlu(mmol/L)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 82px;\"\u003e\n \u003cp\u003e0.048\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 65px;\"\u003e\n \u003cp\u003e2.380\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 70px;\"\u003e\n \u003cp\u003e1.120\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 128px;\"\u003e\n \u003cp\u003e1.022-1.233\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.017\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 59px;\"\u003e\n \u003cp\u003e0.113\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 114px;\"\u003e\n \u003cp\u003eeGFR (ml/min*1.73m\u003csup\u003e2\u003c/sup\u003e)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 82px;\"\u003e\n \u003cp\u003e0.008\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 65px;\"\u003e\n \u003cp\u003e-8.994\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 70px;\"\u003e\n \u003cp\u003e0.929\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 128px;\"\u003e\n \u003cp\u003e0.913-0.943\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 51px;\"\u003e\n \u003cp\u003e\u0026lt;0.01\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 59px;\"\u003e\n \u003cp\u003e-0.074\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 114px;\"\u003e\n \u003cp\u003eC18:0(nmol/ml)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 82px;\"\u003e\n \u003cp\u003e0.552\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 65px;\"\u003e\n \u003cp\u003e-4.192\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 70px;\"\u003e\n \u003cp\u003e0.099\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 128px;\"\u003e\n \u003cp\u003e0.032-0.282\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 51px;\"\u003e\n \u003cp\u003e\u0026lt;0.01\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 59px;\"\u003e\n \u003cp\u003e-2.313\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 114px;\"\u003e\n \u003cp\u003eC16:0(nmol/ml)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 82px;\"\u003e\n \u003cp\u003e0.496\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 65px;\"\u003e\n \u003cp\u003e2.121\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 70px;\"\u003e\n \u003cp\u003e2.863\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 128px;\"\u003e\n \u003cp\u003e1.097-7.699\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.034\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 59px;\"\u003e\n \u003cp\u003e1.052\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003eAbbreviations: eGFR, estimated glomerular filtration rate; Glu, glucose.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTable 2\u003c/strong\u003e Comparison of diagnostic performance of constructing classification models based on different machine learning algorithms\u003c/p\u003e\n\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\" width=\"100%\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 89px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eModel\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 72px;\"\u003e\u0026nbsp;\u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 72px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eAUC\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 72px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eAccuracy\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 72px;\"\u003e\n \u003cp\u003e\u003cstrong\u003ePrecision\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eSensitivity\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 75px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eSpecificity\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 89px;\"\u003e\n \u003cp\u003eLogistic\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 72px;\"\u003e\n \u003cp\u003eTraining\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 72px;\"\u003e\n \u003cp\u003e0.920\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 72px;\"\u003e\n \u003cp\u003e0.8568\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 72px;\"\u003e\n \u003cp\u003e0.8309\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003e0.8958\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 75px;\"\u003e\n \u003cp\u003e0.8177\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 89px;\"\u003e\u0026nbsp;\u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 72px;\"\u003e\n \u003cp\u003evalidation\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 72px;\"\u003e\n \u003cp\u003e0.879\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 72px;\"\u003e\n \u003cp\u003e0.7764\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 72px;\"\u003e\n \u003cp\u003e0.7529\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003e0.8101\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 75px;\"\u003e\n \u003cp\u003e0.7439\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 89px;\"\u003e\n \u003cp\u003eRF\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 72px;\"\u003e\n \u003cp\u003eTraining\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 72px;\"\u003e\n \u003cp\u003e1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 72px;\"\u003e\n \u003cp\u003e1.0000\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 72px;\"\u003e\n \u003cp\u003e1.0000\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003e1.0000\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 75px;\"\u003e\n \u003cp\u003e1.0000\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 89px;\"\u003e\u0026nbsp;\u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 72px;\"\u003e\n \u003cp\u003evalidation\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 72px;\"\u003e\n \u003cp\u003e0.866\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 72px;\"\u003e\n \u003cp\u003e0.8137\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 72px;\"\u003e\n \u003cp\u003e0.7882\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003e0.8481\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 75px;\"\u003e\n \u003cp\u003e0.7805\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 89px;\"\u003e\n \u003cp\u003eDecisionTree\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 72px;\"\u003e\n \u003cp\u003eTraining\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 72px;\"\u003e\n \u003cp\u003e0.919\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 72px;\"\u003e\n \u003cp\u003e0.8880\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 72px;\"\u003e\n \u003cp\u003e0.8670\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003e0.9167\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 75px;\"\u003e\n \u003cp\u003e0.8594\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 89px;\"\u003e\u0026nbsp;\u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 72px;\"\u003e\n \u003cp\u003evalidation\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 72px;\"\u003e\n \u003cp\u003e0.841\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 72px;\"\u003e\n \u003cp\u003e0.7888\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 72px;\"\u003e\n \u003cp\u003e0.7647\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003e0.8228\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 75px;\"\u003e\n \u003cp\u003e0.7561\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 89px;\"\u003e\n \u003cp\u003eNaiveBayes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 72px;\"\u003e\n \u003cp\u003eTraining\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 72px;\"\u003e\n \u003cp\u003e0.877\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 72px;\"\u003e\n \u003cp\u003e0.7917\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 72px;\"\u003e\n \u003cp\u003e0.8146\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003e0.7552\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 75px;\"\u003e\n \u003cp\u003e0.8281\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 89px;\"\u003e\u0026nbsp;\u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 72px;\"\u003e\n \u003cp\u003evalidation\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 72px;\"\u003e\n \u003cp\u003e0.835\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 72px;\"\u003e\n \u003cp\u003e0.7267\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 72px;\"\u003e\n \u003cp\u003e0.7692\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003e0.6329\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 75px;\"\u003e\n \u003cp\u003e0.8171\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003eTo compare the predictive performance of the four biomarker models with individual biomarkers, ROC curves were generated. The models outperformed individual biomarkers, with the model achieving an AUC of 0.920, significantly higher than eGFR (AUC 0.876, p \u0026lt; 0.05). The models also demonstrated better diagnosis of DKD than eGFR (Figure 4A). To further validate the DKD diagnostic model, 236 patients with diabetes and diabetic kidney disease were collected from the Second Affiliated Hospital of Dalian Medical University. After excluding cases with infections, tumors, liver disease, or incomplete data, 120 patients remained (60 with diabetes and 60 with DKD), forming the external validation cohort. Missing values for variables with \u0026lt;20% missing data were imputed using the R package \u0026ldquo;mice,\u0026rdquo; while those with \u0026gt;20% missing were excluded. Fatty acid data were standardized and analyzed using the \u0026ldquo;tableone\u0026rdquo; package (Supplementary Table S2). Significant differences (p \u0026lt; 0.05) were observed in age, systolic blood pressure, ACR, MAU, UCRE, eGFR, C18:0, C20:3, C22:6, C22:5, C22:0, C24:1, Triene/Tetraene, urea, UA, Scre, and hypertension history. In the external cohort, ROC analysis showed an AUC of 0.881, sensitivity of 0.867, and specificity of 0.750 (Figure 4B). Calibration curves (Figure 4C) and decision curve analysis (Figure 4D) confirmed the model\u0026rsquo;s reliability, with a Hosmer-Lemeshow test p-value of 0.620.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e3.2 DKD differential diagnosis modeling\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eA total of 1038 patients from the First Affiliated Hospital of Dalian Medical University were screened, and 744 met the inclusion criteria. Among them, 199 had non-DKD chronic kidney disease. \u0026nbsp;After further controlling for gender, age, eGFR and Scre confounders, 199 patients with non-DKD chronic kidney disease (121 males and 78 females) and 156 patients with DKD (95 males and 61 females) were finally retained for differential diagnosis modeling. Significant differences (p\u0026lt;0.05) were observed between DKD and CKD patients in weight, BMI, SDP, 1,5-AG, TG, TC, Glu, and multiple fatty acids (C14:0, C16:1, C18:3, C18:2, C18:0, C20:5, C20:4, C20:1, C20:0, C22:6, C22:5, C22:4, \u0026omega;3/\u0026omega;6, TotalSFA , TotalPUFA, TotalFA, T\u0026omega;3, T\u0026omega;6, Triene /Tetraene). Additionally, TotalSFA , TotalPUFA, TotalFA, T\u0026omega;3, T\u0026omega;6, Triene /Tetraene ratio also differed significantly (p\u0026lt;0.05, Supplementary Table S3).\u003c/p\u003e\n\u003cp\u003eThe data were split into a training set (70%) and a validation set (30%). Univariate and multivariate analyses, along with stepwise backward regression, identified five potential biomarkers for diabetic kidney disease (DKD): 1,5-AG, Glu, C18:0, BMI, diastolic blood pressure(DBP) (Figure 5，Table 3). Multicollinearity was assessed using the variance inflation factor (VIF). The VIF values for these biomarkers were 1.05, 1.09, 1.06, 1.07 and 1.10, respectively. Given the VIF threshold of \u0026ge; 5 (19), these results suggest that none of the biomarkers exhibited significant multicollinearity.\u003c/p\u003e\n\u003cp\u003eA logistic regression-based diagnostic model was constructed using five diagnostic markers, and a column-line diagram was generated (Figure 6A). Each biomarker in the model corresponds to a score, summed to reflect DKD diagnostic ability. The ROC curve showed an AUC of 0.873, sensitivity of 0.801, and specificity of 0.722 (Figure 6B). The calibration curve confirmed the model\u0026rsquo;s reliability (Figure 6C). Decision curve analysis (DCA) indicated a significant net benefit (Figure 6D). The Hosmer-Lemeshow (H-L) test (p = 0.183\u0026gt;0.05) suggested a good fit. In the validation set, the model achieved an AUC of 0.812, sensitivity of 0.689, and specificity of 0.707 (Figure 6B), with an H-L test p-value of 0.057\u0026gt;0.05. These results demonstrate strong predictive performance in both training and validation sets.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTable 3\u003c/strong\u003e Stepwise regression analysis of differential diagnosis model for DKD\u003c/p\u003e\n\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\" width=\"100%\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 111px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eStd.Error\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003e\u003cstrong\u003ez\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eOR\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e95%CI\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003e\u003cstrong\u003ep\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026beta;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 111px;\"\u003e\n \u003cp\u003e(Intercept)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003e2.30\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003e-1.62\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003e0.11\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e0.02\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003e0.00-2.08\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003e-3.73\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 111px;\"\u003e\n \u003cp\u003eDBP\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003e0.02\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003e-3.00\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003e\u0026lt;0.01\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e0.95\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003e0.91-0.98\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003e-0.05\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 111px;\"\u003e\n \u003cp\u003eC22:4(nmol/ml)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003e0.51\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003e1.96\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003e0.05\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e2.71\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003e1.02-7.59\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003e1.00\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 111px;\"\u003e\n \u003cp\u003eC20:3(nmol/ml)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003e0.44\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003e-1.81\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003e0.07\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e0.45\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003e0.19-1.05\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003e-0.79\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 111px;\"\u003e\n \u003cp\u003eC20:4(nmol/ml)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003e0.42\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003e1.74\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003e0.08\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e2.07\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003e0.96-5.02\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003e0.73\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 111px;\"\u003e\n \u003cp\u003eC18:0(nmol/ml)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003e0.24\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003e-3.01\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003e\u0026lt;0.01\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e0.48\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003e0.29-0.76\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003e-0.72\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 111px;\"\u003e\n \u003cp\u003eGlu(mmol/l)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003e0.17\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003e4.13\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003e\u0026lt;0.01\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e2.06\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003e1.50-3.01\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003e0.72\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 111px;\"\u003e\n \u003cp\u003e1,5-AG(\u0026mu;g/ml)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003e0.04\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003e-4.70\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003e\u0026lt;0.01\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e0.81\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003e0.74-0.88\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003e-0.21\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 111px;\"\u003e\n \u003cp\u003eage（years）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003e0.01\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003e1.77\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003e0.08\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e1.03\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003e1.00-1.06\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003e0.03\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 111px;\"\u003e\n \u003cp\u003eBMI (kg/m\u003csup\u003e2\u003c/sup\u003e)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003e0.05\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003e3.27\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003e\u0026lt;0.01\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003e1.18\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003e1.07-1.31\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 74px;\"\u003e\n \u003cp\u003e0.17\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003eAbbreviations: BMI, body mass index; DBP, diastolic blood pressure ;Glu,glucose; 1,5-AG, 1,5-anhydroglucitol.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTable 4\u003c/strong\u003e logistic regression analysis\u003c/p\u003e\n\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\" width=\"100%\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 118px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 82px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eStd.Error\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 55px;\"\u003e\n \u003cp\u003e\u003cstrong\u003ez\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 69px;\"\u003e\n \u003cp\u003e\u003cstrong\u003ep\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 69px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eOR\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 97px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e95%CI\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 78px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026beta;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 118px;\"\u003e\n \u003cp\u003e(Intercept)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 82px;\"\u003e\n \u003cp\u003e0.22\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 55px;\"\u003e\n \u003cp\u003e-4.71\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 69px;\"\u003e\n \u003cp\u003e\u0026lt;0.01\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 69px;\"\u003e\n \u003cp\u003e0.35\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 97px;\"\u003e\n \u003cp\u003e0.22-0.53\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 78px;\"\u003e\n \u003cp\u003e-1.05\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 118px;\"\u003e\n \u003cp\u003eC24:1(nmol/ml)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 82px;\"\u003e\n \u003cp\u003e0.22\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 55px;\"\u003e\n \u003cp\u003e3.62\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 69px;\"\u003e\n \u003cp\u003e\u0026lt;0.01\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 69px;\"\u003e\n \u003cp\u003e2.23\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 97px;\"\u003e\n \u003cp\u003e1.48-3.52\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 78px;\"\u003e\n \u003cp\u003e0.80\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 118px;\"\u003e\n \u003cp\u003eC20:0(nmol/ml)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 82px;\"\u003e\n \u003cp\u003e0.21\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 55px;\"\u003e\n \u003cp\u003e3.49\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 69px;\"\u003e\n \u003cp\u003e\u0026lt;0.01\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 69px;\"\u003e\n \u003cp\u003e2.10\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 97px;\"\u003e\n \u003cp\u003e1.39-3.22\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 78px;\"\u003e\n \u003cp\u003e0.74\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 118px;\"\u003e\n \u003cp\u003eC20:5(nmol/ml)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 82px;\"\u003e\n \u003cp\u003e0.20\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 55px;\"\u003e\n \u003cp\u003e-2.42\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 69px;\"\u003e\n \u003cp\u003e0.02\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 69px;\"\u003e\n \u003cp\u003e0.62\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 97px;\"\u003e\n \u003cp\u003e0.40-0.89\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 78px;\"\u003e\n \u003cp\u003e-0.48\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 118px;\"\u003e\n \u003cp\u003eC14:0(nmol/ml)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 82px;\"\u003e\n \u003cp\u003e0.24\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 55px;\"\u003e\n \u003cp\u003e-4.84\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 69px;\"\u003e\n \u003cp\u003e\u0026lt;0.01\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 69px;\"\u003e\n \u003cp\u003e0.31\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 97px;\"\u003e\n \u003cp\u003e0.19-0.48\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 78px;\"\u003e\n \u003cp\u003e-1.16\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 118px;\"\u003e\n \u003cp\u003e1,5-AG(\u0026mu;g/ml)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 82px;\"\u003e\n \u003cp\u003e0.05\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 55px;\"\u003e\n \u003cp\u003e3.13\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 69px;\"\u003e\n \u003cp\u003e\u0026lt;0.01\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 69px;\"\u003e\n \u003cp\u003e1.18\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 97px;\"\u003e\n \u003cp\u003e1.07-1.36\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 78px;\"\u003e\n \u003cp\u003e0.16\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003eAbbreviations: 1,5-AG, 1,5-anhydroglucitol\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e3.3 Analysis of the correlation between 1,5-AG, fatty acids and DKD\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eDKD was categorized into stages 1-5 according to eGFR, with stages 1-3 as early DKD and stages 4-5 as advanced DKD, of which 162 cases were in early stage, 97 males and 65 females, with a mean age of 62.17, and 112 cases were in advanced stage, 46 males and 66 females, with a mean age of 64.42, and 1,5-AG, UCRE, Scre, Urea, UA, HDL were found, Glu, HbA1c, C14:0, C16:1, C18:3, C18:0, C20:5, C20:3, C20:0, C22:6, C24:1, \u0026omega;3/\u0026omega;6, eGFR, T\u0026omega;3, ACR, MAU, and hypertension were found to be statistically significant differences between the early and late stages of DKD（p\u0026lt;0.05, Supplementary Table S4）. 1,5-AG was found to be negatively correlated with Glu and eGFR, positively correlated with Scre、UA and UREA, and the correlation with lipids was not significant (Supplementary Table S5). Using binary logistic regression analysis, C24:1（p\u0026lt;0.01, OR 2.23，\u0026beta; 0.80）, C20:0（p\u0026lt;0.01, OR 2.10，\u0026beta; 0.74）and 1,5-AG（p\u0026lt;0.01, OR 1.18，\u0026beta; 0.16） were found to be possible independent risk factors for the progression of DKD (Table 4).\u003c/p\u003e"},{"header":"4 Discussion","content":"\u003cp\u003eDKD is a common diabetes complication and a major cause of ESRD, severely impacting patient health and quality of life. Early identification of clinical biomarkers may slow or even halt DKD progression, and recent studies have focused on discovering such markers for timely intervention.\u003c/p\u003e\n\u003cp\u003e1,5-AG is a short-term glucose monitoring indicator with a structure similar to glucose, allowing it to freely cross cell membranes while remaining metabolically inert and stable (20). Under normal conditions, 99.9% of filtered 1,5-AG is reabsorbed in the renal tubules, maintaining stable blood levels. However, when blood glucose exceeds the renal threshold, glucose competitively inhibits 1,5-AG reabsorption, leading to its urinary excretion and a reduction in serum levels. Hasslacher et al. reported that in T2DM patients with CKD stages 1–3, 1,5-AG levels are influenced by both decreased renal function and glomerular hyperfiltration(21). Tavares et al. also found a correlation between 1,5-AG and the progression of massive proteinuria in DKD, independent of eGFR(22). In our study, 1,5-AG was negatively correlated with glucose and eGFR and positively correlated with serum creatinine and uric acid, consistent with previous findings (21).\u003c/p\u003e\n\u003cp\u003eAlthough some studies report a strong negative correlation between serum 1,5-AG and total cholesterol (23), our data did not show a significant association. Fatty acids—key components of blood lipids—are classified as saturated (SFA), monounsaturated (MUFA), or polyunsaturated (PUFA) based on their hydrocarbon chains. Their role in DKD has recently attracted much attention. A meta-analysis of 3089 T2DM patients demonstrated that eGFR is negatively correlated with omega-6 fatty acids, linoleic acid, and non-HDL cholesterol, and positively correlated with certain HDL components (24). Moreover, lipid accumulation has been shown to induce tubular damage and glomerulosclerosis (25, 26). Case-control studies have associated palmitic (C16:0) and linoleic (C18:2) acids with DKD, suggesting their potential as biomarkers (27). Additionally, serum analysis of 326 type 1 diabetes patients revealed that MUFAs (e.g., C16:1, C18:1) and ω6 and ω9 fatty acids are linked to DKD development (28). Our findings further validate the role of fatty acids (C24:1, C20:0) as independent risk factors for the progression of DKD, underscoring their close relationship with the disease.\u003c/p\u003e\n\u003cp\u003eSeveral studies have applied machine learning (ML) to evaluate DKD risk. Zou et al. compared four ML algorithms—gradient boosting, support vector machine, logistic regression, and random forest (RF)—and found that RF achieved the highest predictive performance (AUC 0.90), followed by SVM and GBM (AUC 0.88) and logistic regression (AUC 0.83), ultimately selecting RF for ESRD risk prediction (29). Dagliati et al. also used ML to predict diabetic complications (30). In another study, longitudinal data from 1365 participants (Chinese, Malaysian, and Indian, aged 40–80) indicated that a neural network achieved an AUC of 0.851—7.0% higher than logistic regression (AUC 0.795)—with sensitivities of 88.2% versus 73.0% and specificities of 65.9% versus 72.8% (29, 31) Jiang et al.’s meta-analysis of 20 cohorts with 41,271 T2DM patients reported an external validation AUC of 0.765 for early DKD risk prediction(32).\u003c/p\u003e\n\u003cp\u003eCompared with previous studies, our DKD diagnostic model has several advantages. First, it integrates both macromolecular indices (e.g., total cholesterol, triglycerides) and small-molecule metabolites (e.g., 1,5-AG, fatty acids). Second, we developed not only an diagnosis model based on large samples but also a differential diagnostic model to distinguish DKD from non-DKD cases, reflecting differences in prognosis and treatment. Third, our model underwent independent external validation, achieving an AUC of 0.881. Fourth, although four ML algorithms were evaluated, logistic regression was ultimately chosen for its superior interpretability and clinical applicability. Our logistic model performed consistently in both the training set (AUC = 0.920, sensitivity 89.58%, specificity 81.77%) and the validation set (AUC = 0.879, sensitivity 81.01%, specificity 74.39%). Furthermore, combining metabolite data with ML improved prediction accuracy and helped uncover new risk factors. While RF models may yield higher AUCs with balanced datasets, their complexity and sensitivity to data imbalance limit clinical interpretability, making logistic regression the more practical choice.\u003c/p\u003e\n\u003cp\u003eIn summary, we developed a logistic regression–based model for DKD diagnostic that demonstrates high diagnostic efficacy (sensitivity 81.01%, specificity 74.39%, AUC 0.879) and robust external validation (AUC 0.881). The accompanying nomogram further aids clinicians in formulating diagnosis and treatment plans. However, limitations include the lack of pathological biopsies for all DKD patients, limited interpretability inherent to some ML models, and the exclusion of 24-hour urinary protein, glycosylated hemoglobin, urinary microalbumin, and ACR data. Future studies should expand sample sizes, include biopsy-confirmed cases, and integrate additional biomarkers to further optimize the model and enhance prediction accuracy.\u003c/p\u003e"},{"header":"5 Conclusion","content":"\u003cp\u003eIn conclusion, We developed an DKD diagnostic model using four biomarkers (C16:0, C18:0, eGFR, Glu) with AUCs of 0.920 (training), 0.879 (internal validation), and 0.881 (external validation). A logistic regression model for differentiating DKD from non-diabetic CKD using five biomarkers (1,5-AG, Glu, C18:0, BMI, DBP) achieved AUCs of 0.873 (training), and 0.812(validation). Notably, 1,5-AG was negatively correlated with eGFR and positively with creatinine and UA, while fatty acids C24:1, C20:0 and 1,5-AG may be independent risk factors. Our findings offer valuable guidance for diagnosis and differential DKD diagnosis.\u003c/p\u003e"},{"header":"Abbreviations","content":"\u003cp\u003eDKD \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;diabetic kidney disease\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eESRD \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; end-stage renal disease\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eT2DM \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; type 2 diabetes mellitus\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eCKD \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;chronic kidney disease\u003c/p\u003e\n\u003cp\u003eAUC \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;area under the receiver operating characteristic curve\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eSBP \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;blood pressure\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e1,5-AG \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;1,5-anhydroglucitol\u003c/p\u003e\n\u003cp\u003eDBP \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; diastolic blood pressure\u003c/p\u003e\n\u003cp\u003eScre \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;serum creatinine\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eUA \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; uric acid\u003c/p\u003e\n\u003cp\u003eGlu \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; glucose\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eHbA1c \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;glycated hemoglobin\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eHPLC-MS/MS high-performance liquid chromatography tandem mass spectrometry\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eML \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;Machine learning\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eeGFR \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;estimated glomerular filtration rate\u003c/p\u003e\n\u003cp\u003eDCA \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; Decision curve analysis\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eNB \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;net benefit\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eSFA \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;saturated fatty acids\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eMUFA \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;monounsaturated fatty acids \u0026nbsp;\u003c/p\u003e\n\u003cp\u003ePUFA \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;polyunsaturated fatty acids\u0026nbsp;\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eData availability statement\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthor\u003c/strong\u003e\u003cstrong\u003econtributions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eXZ: Conceptualization, Data curation, Methodology, Writing-original draft. LL: Software, Data curation, Methodology. YW: Conceptualization, Methodology, Validation. TK: Validation, FW and DZ. Formal Analysis. MF and HC: Investigation. SL: Conceptualization, Methodology, Validation. PC: Conceptualization, Funding acquisition, Writing-review \u0026amp; editing.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding information\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe author(s) declare financial support was received for the research, authorship, and/or publication of this article. This work was supported by the Liaoning Provincial Department of Education Basic Research Project，Development of a Mass Spectrometry-Based 1,5-Anhydroglucitol Project and Its Application in the Diagnosis and Treatment of Diabetes（NO.LJ212410161051）.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAcknowledgments\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors sincerely thank all the doctors, nurses, and research staff at the First Affiliated Hospital of Dalian Medical University and the Second Affiliated Hospital of Dalian Medical University for their participation in this study.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConflict of interest\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ePublisher’s note\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAll claims expressed in this article are solely those of the authors and do not necessarily reflect the views of their affiliated institutions, the publisher, the editors, or the reviewers. Any product evaluated in this article or any claim made by its manufacturer is neither guaranteed nor endorsed by the publisher.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eEthics approval and consent to participate\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis study was reviewed and approved by the Institutional Ethical Review Boards of the First and Second Affiliated Hospitals of Dalian Medical University (Approval No. YJ-KS-KY-2024-715). The requirement for informed consent was waived by the\u0026nbsp;Ethics Committees in accordance with the\u003cem\u003e\u0026nbsp;Measures for the Ethical Review of Biomedical Research Involving Humans\u0026nbsp;\u003c/em\u003e(National Health and Family Planning Commission of the People’s Republic of China, 2016) and the principles of the Declaration of Helsinki (2013 revision), as the study involved minimal risk to participants and made use of existing clinical data without adversely affecting their rights or welfare.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eClinical trial number\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNot applicable.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConsent for publication\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNot applicable.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eSupplementary material\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe Supplementary Material for this article can be found online.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eBunny Y, Kai Y, Yan D. Advances in metabolomics in the early diagnosis of diabetic nephropathy. Western China University of Medicine. 2023;38(04):603-7.\u003c/li\u003e\n\u003cli\u003eXuelian B, Chia-Yi Z, Guoliang X, Xiaoqing C. Meta-analysis of the prevalence of chronic kidney disease in Chinese adults. Chinese Pharmaceutical Sciences. 2022;12(09):49-53.\u003c/li\u003e\n\u003cli\u003eYu W. Evolution of the etiology of chronic kidney disease in the elderly and the combined treatment of traditional Chinese and Western medicine [master]: Beijing University of Chinese Medicine; 2021.\u003c/li\u003e\n\u003cli\u003eDiabetes CMA. China Guidelines for Clinical Application of Glucose Monitoring (2021 ). Chinese Diabetes Journal. 2021;13(10):926-48.\u003c/li\u003e\n\u003cli\u003eSeok H, Huh JH, Kim HM, Lee BW, Kang ES, Lee HC, et al. 1,5-anhydroglucitol as a useful marker for assessing short-term glycemic excursions in type 1 diabetes. Diabetes Metab J. 2015;39(2):164-70.\u003c/li\u003e\n\u003cli\u003eWang Y, Yuan Y, Zhang Y, Lei C, Zhou Y, He J, et al. Serum 1,5-anhydroglucitol level as a screening tool for diabetes mellitus in a community-based population at high risk of diabetes. Acta Diabetol. 2017;54(5):425-31.\u003c/li\u003e\n\u003cli\u003eShuzhen W, Ting W, Yumei Z. Advances in clinical studies of 1,5-anhydroglucitol in diabetes mellitus. The New World of Diabetes. 2022;25(24):193-8.\u003c/li\u003e\n\u003cli\u003eTursun BB, Jinlian G, Yan J, Mingchen Z, Wenyan C. A study on the relationship between serum 1,5-anhydroglucitol levels and diabetic nephropathy. Journal of Xinjiang Medical University. 2016;39(06):750-2+6.\u003c/li\u003e\n\u003cli\u003eWei X, Weihua S, Zhao L. Correlation of serum 1,5-anhydroglucitol with urinary microclearbones in patients with type 2 diabetes mellitus. Journal of Clinical Laboratory Medicine. 2021;39(07):511-3.\u003c/li\u003e\n\u003cli\u003eBai Y, Yang R, Song Y, Wang Y. Serum 1,5-Anhydroglucitol Concentrations Remain Valid as a Glycemic Control Marker In Diabetes with Earlier Chronic Kidney Disease Stages. Exp Clin Endocrinol Diabetes. 2019;127(4):220-5.\u003c/li\u003e\n\u003cli\u003eHan LD, Xia JF, Liang QL, Wang Y, Wang YM, Hu P, et al. Plasma esterified and non-esterified fatty acids metabolic profiling using gas chromatography-mass spectrometry and its application in the study of diabetic mellitus and diabetic nephropathy. Anal Chim Acta. 2011;689(1):85-91.\u003c/li\u003e\n\u003cli\u003ePauling L RAB, Teranishi R , Cary P. Quantitative Analysis of Urine Vapor and Breath by Gas-Liquid Partition Chromatography. Proceedings of the National Academy of Sciences of the United States of America 1971;68(10):2374-6.\u003c/li\u003e\n\u003cli\u003eChoi RY, Coyner AS, Kalpathy-Cramer J, Chiang MF, Campbell JP. Introduction to Machine Learning, Neural Networks, and Deep Learning. Transl Vis Sci Technol. 2020;9(2):14.\u003c/li\u003e\n\u003cli\u003eZhu Y, Zhang Y, Yang M, Tang N, Liu L, Wu J, et al. Machine Learning-Based Predictive Modeling of Diabetic Nephropathy in Type 2 Diabetes Using Integrated Biomarkers: A Single-Center Retrospective Study. Diabetes Metab Syndr Obes. 2024;17:1987-97.\u003c/li\u003e\n\u003cli\u003eHan H, Chen Y, Yang H, Cheng W, Zhang S, Liu Y, et al. Identification and Verification of Diagnostic Biomarkers for Glomerular Injury in Diabetic Nephropathy Based on Machine Learning Algorithms. Front Endocrinol (Lausanne). 2022;13:876-960.\u003c/li\u003e\n\u003cli\u003eKidney Disease: Improving Global Outcomes Diabetes Work G. KDIGO 2022 Clinical Practice Guideline for Diabetes Management in Chronic Kidney Disease. Kidney Int. 2022;102(5S):S1-S127.\u003c/li\u003e\n\u003cli\u003eAmerican Diabetes Association Professional Practice C. 2. Diagnosis and Classification of Diabetes: Standards of Care in Diabetes-2024. Diabetes Care. 2024;47(Suppl 1):S20-S42.\u003c/li\u003e\n\u003cli\u003eGroup KDIGOKCW. KDIGO 2024 Clinical Practice Guideline for the Evaluation and Management of Chronic Kidney Disease. Kidney International. 2024;105(4):S177-S314.\u003c/li\u003e\n\u003cli\u003eYue S, Li S, Huang X, Liu J, Hou X, Wang Y, et al. Construction and Validation of a Risk Prediction Model for Acute Kidney Injury in Patients Suffering from Septic Shock. Dis Markers. 2022;2022:9367873.\u003c/li\u003e\n\u003cli\u003eYing L, Ma X, Yin J, Wang Y, He X, Peng J, et al. The metabolism and transport of 1,5-anhydroglucitol in cells. Acta Diabetol. 2018;55(3):279-86.\u003c/li\u003e\n\u003cli\u003eHasslacher C, Kulozik F. Effect of renal function on serum concentration of 1,5-anhydroglucitol in type 2 diabetic patients in chronic kidney disease stages I-III: A comparative study with HbA1c and glycated albumin. J Diabetes. 2016;8(5):712-9.\u003c/li\u003e\n\u003cli\u003eTavares G, Venturini G, Padilha K, Zatz R, Pereira AC, Thadhani RI, et al. 1,5-Anhydroglucitol predicts CKD progression in macroalbuminuric diabetic kidney disease: results from non-targeted metabolomics. Metabolomics. 2018;14(4):39.\u003c/li\u003e\n\u003cli\u003eYuexing Y, Yao W. Analysis of the correlation between 1,5-anhydroglucitol and blood lipids in patients with type 2 diabetes mellitus. Journal of Southeast University. 2014;33(6):774-7.\u003c/li\u003e\n\u003cli\u003eTofte N, Vogelzangs N, Mook-Kanamori D, Brahimaj A, Nano J, Ahmadizar F, et al. Plasma Metabolomics Identifies Markers of Impaired Renal Function: A Meta-analysis of 3089 Persons with Type 2 Diabetes. J Clin Endocrinol Metab. 2020;105(7).\u003c/li\u003e\n\u003cli\u003eHerman-Edelstein M, Scherzer P, Tobar A, Levi M, Gafter U. Altered renal lipid metabolism and renal lipid accumulation in human diabetic nephropathy. J Lipid Res. 2014;55(3):561-72.\u003c/li\u003e\n\u003cli\u003eYang W, Luo Y, Yang S, Zeng M, Zhang S, Liu J, et al. Ectopic lipid accumulation: potential role in tubular injury and inflammation in diabetic kidney disease. Clin Sci (Lond). 2018;132(22):2407-22.\u003c/li\u003e\n\u003cli\u003eZhang H, Zuo JJ, Dong SS, Lan Y, Wu CW, Mao GY, et al. Identification of Potential Serum Metabolic Biomarkers of Diabetic Kidney Disease: A Widely Targeted Metabolomics Study. J Diabetes Res. 2020;2020:3049098.\u003c/li\u003e\n\u003cli\u003eMakinen VP, Tynkkynen T, Soininen P, Forsblom C, Peltola T, Kangas AJ, et al. Sphingomyelin is associated with kidney disease in type 1 diabetes (The FinnDiane Study). Metabolomics. 2012;8(3):369-75.\u003c/li\u003e\n\u003cli\u003eZou Y, Zhao L, Zhang J, Wang Y, Wu Y, Ren H, et al. Development and internal validation of machine learning algorithms for end-stage renal disease risk prediction model of people with type 2 diabetes mellitus and diabetic kidney disease. Ren Fail. 2022;44(1):562-70.\u003c/li\u003e\n\u003cli\u003eDagliati A, Marini S, Sacchi L, Cogni G, Teliti M, Tibollo V, et al. Machine Learning Methods to Predict Diabetes Complications. J Diabetes Sci Technol. 2018;12(2):295-302.\u003c/li\u003e\n\u003cli\u003eSabanayagam C, He F, Nusinovici S, Li J, Lim C, Tan G, et al. Prediction of diabetic kidney disease risk using machine learning models: A population-based cohort study of Asian adults. Elife. 2023;12.\u003c/li\u003e\n\u003cli\u003eJiang W, Wang J, Shen X, Lu W, Wang Y, Li W, et al. Establishment and Validation of a Risk Prediction Model for Early Diabetic Kidney Disease Based on a Systematic Review and Meta-Analysis of 20 Cohorts. Diabetes Care. 2020;43(4):925-33.\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"bmc-endocrine-disorders","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"bend","sideBox":"Learn more about [BMC Endocrine Disorders](http://bmcendocrdisord.biomedcentral.com/)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/bend/default.aspx","title":"BMC Endocrine Disorders","twitterHandle":"BMC_series","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"em","reportingPortfolio":"BMC Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"1,5-anhydroglucitol, Diabetic kidney disease, fatty acids, machine learning, diagnostic model","lastPublishedDoi":"10.21203/rs.3.rs-7347937/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7347937/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003ch2\u003eBackground\u003c/h2\u003e\u003cp\u003eEarly diagnosis and timely intervention are critical for delaying the progression of diabetic kidney disease (DKD) to end-stage renal disease (ESRD). This study integrated metabolomics profiling with machine learning algorithms to comprehensively identify blood-based biomarkers\u0026mdash;including 1,5-anhydroglucitol (1,5-AG) and multiple fatty acids\u0026mdash;associated with DKD, and to develop predictive models for both diagnosis and differential diagnosis.\u003c/p\u003e\u003ch2\u003eMethods\u003c/h2\u003e\u003cp\u003eClinical data and serum samples were collected from 1,038 patients with DKD, diabetes mellitus, or non-DKD chronic kidney disease (CKD) at the First Affiliated Hospital of Dalian Medical University. Concentrations of fatty acids and 1,5-AG were quantified by HPLC\u0026ndash;MS/MS. Candidate biomarkers were screened using regression analyses. Diagnostic models for DKD were developed using four algorithms\u0026mdash;binary logistic regression, random forest, decision tree, and naive Bayes\u0026mdash;while a logistic regression model was applied to differentiate DKD from non-diabetic CKD. Model performance was assessed by the area under the receiver operating characteristic curve (AUC), sensitivity, and specificity. External validation was conducted in an independent cohort of 236 patients (DKD and diabetes without renal insufficiency) from the Second Affiliated Hospital.\u003c/p\u003e\u003ch2\u003eResults\u003c/h2\u003e\u003cp\u003eFour biomarkers\u0026mdash;C16:0, C18:0, estimated glomerular filtration rate (eGFR), and glucose\u0026mdash;were identified for DKD diagnosis. Logistic regression achieved the best performance, with AUCs of 0.920 (training), 0.879 (internal validation), and 0.881 (external validation). For differential diagnosis between DKD and non-diabetic CKD, five biomarkers\u0026mdash;1,5-AG, glucose, C18:0, body mass index (BMI), and diastolic blood pressure (DBP)\u0026mdash;yielded AUCs of 0.873 (training) and 0.812 (internal validation). Correlation analyses revealed that 1,5-AG was negatively associated with glucose and eGFR, but positively associated with serum creatinine, uric acid, and urea. In contrast, C14:0, C20:0, and C24:1 were positively correlated with glucose and eGFR, but negatively with serum creatinine. Multivariate analysis identified C24:1, C20:0, and 1,5-AG as independent risk factors for DKD progression.\u003c/p\u003e\u003ch2\u003eConclusion\u003c/h2\u003e\u003cp\u003eFatty acids C24:1 and C20:0, along with 1,5-AG, may independently increase the risk of DKD progression, and renal function appears to influence 1,5-AG levels. Both the diagnostic and differential diagnostic models demonstrated robust predictive performance for DKD in independent cohorts.\u003c/p\u003e","manuscriptTitle":"A Metabolomics-Guided Machine Learning Model for Diagnosis and Differential Diagnosis of Diabetic Kidney Disease: A Dual-Center Study","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-09-22 18:54:02","doi":"10.21203/rs.3.rs-7347937/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"reviewerAgreed","content":"242545184391649463206761986843389039636","date":"2025-09-25T02:56:21+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2025-09-15T05:21:45+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2025-09-10T05:53:59+00:00","index":"","fulltext":""},{"type":"editorInvited","content":"","date":"2025-08-19T06:42:12+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2025-08-19T04:47:56+00:00","index":"","fulltext":""},{"type":"submitted","content":"BMC Endocrine Disorders","date":"2025-08-19T04:43:33+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"bmc-endocrine-disorders","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"bend","sideBox":"Learn more about [BMC Endocrine Disorders](http://bmcendocrdisord.biomedcentral.com/)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/bend/default.aspx","title":"BMC Endocrine Disorders","twitterHandle":"BMC_series","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"em","reportingPortfolio":"BMC Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"610f8a18-8436-432d-b145-a5c593fb35cd","owner":[],"postedDate":"September 22nd, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[],"tags":[],"updatedAt":"2025-09-22T18:54:02+00:00","versionOfRecord":[],"versionCreatedAt":"2025-09-22 18:54:02","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-7347937","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7347937","identity":"rs-7347937","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}