Predicting three-month fasting blood glucose and glycated hemoglobin of patients with type 2 diabetes based on multiple machine learning algorithms

Predicting three-month fasting blood glucose and glycated hemoglobin of patients with type 2 diabetes based on multiple machine learning algorithms | 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 Predicting three-month fasting blood glucose and glycated hemoglobin of patients with type 2 diabetes based on multiple machine learning algorithms Xue Tao, Min Jiang, Yumeng Liu, Qi Hu, Baoqiang Zhu, Jiaqiang Hu, and 7 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-1868105/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Background Type 2 diabetes is the type with the largest proportion of people with diabetes.With the progression of the disease, patients with type 2 diabetes mellitus will have different degrees of complications, which will seriously reduce the quality of life of the patients and bring a heavy economic burden to the patient's families. Therefore, establishing a predictive model for glycemic control in patients with type 2 diabetes mellitus is of great help in optimizing the treatment of type 2 diabetes mellitus and delaying disease progression. Design and Methods: A retrospective study was conducted on type 2 diabetes mellitus real-world medical data from 4 cities in Sichuan Province, China from January 2015 to December 2020, including basic patient information, medication status, laboratory results, dietary habits, exercise status, and the actual follow-up of the patient after treatment. After data preprocessing, data inputting, data sampling, and feature screening, 16 kinds of machine learning methods were used to construct fasting blood glucose prediction models and glycated hemoglobin prediction models for type 2 diabetes mellitus patients, and 5 prediction models with the best prediction performance were screened respectively. Results A total of 375,723 cases of type 2 diabetes mellitus patients were collected, 10,000 cases were included to establish the fasting blood glucose model, and 2,169 cases were established to establish the HbA1c model. The best prediction model both of fasting blood glucose and HbA1c finally obtained are realized by ensemble learning and modified random forest inputting, the AUC value are 0.819 and 0.970, respectively. The most important indicators of the fasting blood glucose and glycated hemoglobin prediction model were fasting blood glucose and glycated hemoglobin. Medication compliance, follow-up outcome, dietary habits, BMI, and waist circumference also had a greater impact onfasting blood glucose levels. But on the glycated hemoglobin level, laboratory indicators such as platelets, Serum creatinine, Aspartate Transaminase, Hemoglobin, etc. had more impact. Conclusion The prediction accuracy of the models of the two blood glucose control indicators is high and has certain clinical applicability. Glycated hemoglobin and fasting blood glucose are mutually important predictors, and there is a close relationship between them. fasting blood glucose glycated hemoglobin type 2 diabetes mellitus prediction model machine learning Figures Figure 1 Figure 2 Figure 3 Figure 4 Introduction Diabetes mellitus is a chronic progressive disease characterized by disorders of glucose metabolism [1] . In recent years, the global incidence of diabetes has been increasing year by year. According to the latest epidemiological data from the International Diabetes Federation(IDF): the global diabetes prevalence in 20-79 years old in 2021 was estimated to be 10.5% (536.6 million people), rising to 12.2% (783.2 million) in 2045 [2] . And in 2021, almost one in two adults (20-79 years old) with diabetes were unaware of their diabetes status (44.7%, 239.7 million) [3] . Type 2 diabetes mellitus (T2DM) patients accounted for more than 90.0% of the total diabetic patients [4] . With the development of the disease, most T2DM patients will have different degrees of complications, which will seriously reduce the quality of life of the patients and bring a heavy economic burden to the patients' families [5] . And the severity of complications is inseparable from glycemic control. Therefore, active, safe, and effective blood glucose control has positive significance for preventing complications, improving the quality of life of T2DM patients, and reducing the economic burden on patients and society. The global age-standardized point prevalence and death rates for T2DM were 5282.9 and 18.5 per 100 000, an increase of 49% and 10.8%, respectively, since 1990 [6] . Serious challenges exist in diabetes prevention and glycemic control. The reason may be that the individual differences of patients, such as physiology and pathology, diet structure, and lifestyle, are not fully considered in the treatment process. Therefore, it is urgent to establish an efficient, accurate, and economical T2DM prediction model to improve the treatment rate of T2DM in medical institutions at all levels. With the continuous development of database and data mining technology, data mining is more and more used to mine medical databases efficiently [7] . The existing data mining technology application research shows that the model established by data mining has high accuracy [8] . Fasting blood glucose (FBG) is the blood glucose value detected in the plasma collected before breakfast after an overnight fast (at least 8-10 hours without any food, except for drinking water) [9] , which can be used to reflect the secretion function of basal insulin. Glycated hemoglobin (HbA1c) is the product of the combination of hemoglobin in red blood cells and blood glucose, and its content depends on the blood glucose concentration and the contact time between blood glucose and, and has nothing to do with factors such as blood drawing time and whether the patient is fasting [10] . They are used to detect blood glucose and both are important indicators for diagnosing diabetes and reflecting the prognosis of diabetes [11] . At present, prediction models based on machine learning algorithms are mainly used for the prediction of diabetes and its complications, and there are few studies on the prediction models of patients' glycemic control after medication [12.13] . Based on this, this study intends to establish artificial intelligence prediction models for the compliance of two blood glucose indicators in T2DM patients after 3 months of treatment through data mining, to explore potential predictive relationships between FBG and HbA1c, improve the treatment rate and control rate of T2DM, reduce the incidence of adverse reactions, and prevent and reduce the occurrence of complications. METHODS Study design and data source The data of this study were obtained from the Public Health Service System and the Medical Record Homepage Management System of the Health Information Center of Sichuan Province, China (including personal basic information form, health check-up form, and follow-up service record form), and the overall data were derived from patients who received anti-diabetic drugs or had the International Classification of Diseases Tenth Revision (ICD-10) code[ 14] for type 2 diabetes between January 2015 and December 2020. A total of 375,723 T2DM patients' related diagnosis and treatment data were collected in this study, and the available data for constructing the FBG prediction model and the HbA1c prediction model were screened according to the following criteria: if the same patient had 2 or more registration data within 3±1 months, the patient's data was available, If there were multiple sets of data within 3±1 months, the data closest to 3 months from the baseline should be taken, if there were multiple sets of consistent data longitudinally for the same patient (a patient had multiple sets of data that met the requirements of having 2 or more data within 3±1 months), a group was randomly selected for inclusion. FBG control outcomes judgment standard: (1)well-controlled, FBG was 4.4-7.0 mmol/L, (2)poorly controlled, FBG＞7.0 mmol/L or FBG＜4.4 mmol/L. Judgment criteria for HbAlc value: (1)well-controlled, HbA1c<7% , (2)poorly controlled, the value of HbA1c was not in this range. The data included the patient's basic information, drug use, test indicators, and living and diet, as well as the actual follow-up of the patient after treatment. This study used a unique ID to identify patient connection information, and all research operations carried out would not be traced to the individual patient, and the patient's sensitive personal information (such as name, phone number, address, work unit, responsible doctor, etc.) would be deleted. All files were encrypted during transmission and use, and documents were received by a password. This study has passed the ethical review, the approval document in Supplementary Fig 1. In this study, a total of 511 variables were included, which were named X1–X511 for statistical convenience(Detailed variables are shown in Supplemental Table 1). Data analysis was performed using named variables, and the variable names were restored after the model evaluation process was complete. Data cleaning We deleted variables with a missing ratio of 90%, a single category ratio of 90%, and variables with a coefficient of variation less than 0.1. These variables had little impact on the establishment of the model, and the analysis was meaningless, so they were deleted. Data inputting was done using not inputting and random forest inputting. After the data were inputted, if there was a large difference between the positive and negative sample sizes, the data was balanced by sampling. And we modified outliers to the maximum or minimum value of the norm. The method of “not inputting” was to delete the missing columns and the missing rows in the data in turn, and finally, we got the data without missing values. The “modified random forest inputting” meant that by continuously introducing the inputted columns into the model, as the amount of data continued to accumulate, the obtained values had a higher accuracy rate, which could achieve a more accurate prediction of missing values. After the data were inputted, the data were divided into training and test sets for machine learning. And the number of training sets accounted for 80% of the total sample size, and the number of test sets accounted for 20% of the total sample size. Feature screening The data were screened using three methods: Not screening, Lasso screening, and Boruta screening. Feature screening is an important aspect of model building, which helps to exclude relevant variables, biases, and limitations of unnecessary noise, making the final analysis results closer to reality. Lasso are a useful atheoretical approach for both developing predictive models and selecting key indicators within an often substantially larger pool of available indicators by inputting all latent variables at the same time, reducing bias caused by unimportant variables, and selecting only the most important variables from a potentially large initial pool[ 15] . Model training 16 or 18 kinds of machine learning algorithms were used for model training, and the data after feature screening were modeled respectively. The specific machine learning algorithm models used included: Logistic regression, Stochastic Gradient Descent（SGD）, Decision Tree, Gaussian Naive Bayes, Bernoulli Naive Bayes, Multinomial Naive Bayes, Quadratic Discriminant Analysis (QDA), Random Forest, Extra Tree, Linear Discriminant Analysis (LDA), Passive Aggressive, AdaBoost, Begging, Gradient Boosting, XGBoost, and Ensemble Learning. In 2011, Tianqi Chen and Carlos Guestrin first proposed the XGBoost algorithm, or the extreme gradient boosting algorithm. It is a machine learning model that achieves stronger learning effects by integrating multiple weak learners[ 16] , and has better flexibility and scalability[ 17] . Compared with general machine learning algorithms, the XGBoost model shows strong advantages. These machine learning algorithms have their strengths, among which, the ensemble learning model is an evaluation index based on the trained model, summarizing the best model and outputting according to the voting principle. The evaluation indicators of the prediction model included Area Under Curve(AUC), Accuracy, Precision, Recall, and F1 Score. According to the machine learning results, the 5 models with the best prediction performance were selected and their receiver operating characteristic curve（ROC）and P-R curves were drawn. Model verification Ten-fold cross-validation and bootstrapping sampling were used to verify the impact of different preprocessing algorithms and different machine learning algorithms on the prediction of building FBG and HbAlc models. The model with the largest AUC was selected and constructed using 10 subsets (randomly drawn 10%–100% of the total sample size) to assess the effect of different sample sizes on predictive power. Each subset was split 4:1 into a training set and a test set, and the AUC calculated from the test set was used for sample size checking. By transforming randomly sampled data, 10 independent replicates were generated for each model. A process framework of the data flow is shown in Figure 1. Data flowed through each node according to a predetermined schedule. Statistical Analysis Continuous variables were expressed as mean±standard deviation, and count variables were expressed as frequency. Differences between quantitative data were tested using a t-test and rank test. Hypothesis testing was used to investigate the influence of different data processing methods and algorithms on the model prediction performance. On the analysis results of bootstrapping sampling and validation set, hypothesis testing single factor analysis was performed. The analysis content included different data inputting methods, feature screening methods, and the corresponding mean±standard deviation and 95% confidence interval between the three dimensions of the machine learning model and the five evaluation indicators (AUC, Accuracy, Precision, Recall, and F1 Score) and p-value. Excel 2016 was used for summarizing data, and all statistical analyses were performed using Python 3.8. RESULT Baseline characteristics The FBG study cohort included 100,000 patients and the HbA1c study cohort included 2,169 patients. Baseline demographic, clinical, laboratory and medication details are shown in Table 1. The mean ages of the two cohorts were 64.0 ± 10.1 years and 63.1 ± 10.2 years, respectively. The most common comorbidities in both cohorts were hypertension, kidney disease, and heart disease. At baseline, FBG was 8.6±3.9 mmol/L and 8.9±3.8 mmol/L, respectively, and HbA1c were both 7.8±2.8%. Table 1 Baseline characteristics of participants Predictors FBG（N=100,000） HbA1c（N=2,169） Categorical variables,n(%) Gender Male 38,205 ( 38.2 ) 735(33.9) Female 61,795 (61.8) 1434(66.1) Education College and above 71,532 ( 71.5 ) 1631(75.2) Below college and other 28,648 (28.5) 538(24.8) Marital status Married/living as married/civil partnershi 84,413 ( 84.5 ) 1749(80.6) Single/never marrie 13,867 ( 13.9 ) 339(15.6) Widowed 981 ( 1.0 ) 28(1.3) Divorced or separate 687 ( 0.6 ) 6(0.3) Complication Complicated with diabetes-related complications 16,016 ( 16.0 ) 820(37.8) Without diabetes-related complications 83,984 (84.0) 1349(62.2) Comorbidity Cerebrovascular diseas 11,018 ( 11.0 ) 916(42.2) Kidney disease 15,752 ( 15.8 ) 262(12.1) Heart disease 13,298 ( 13.3 ) 889(41.0) Vascular disease 11,542 ( 11.5 ) 917(42.3) Ophthalmological disease 11,155 ( 11.2 ) 897(41.4) Hypertensio 54,363 ( 54.4 ) 1354(62.4) Drug metformin 16,978 ( 17.0 ) - Grezit 8,421 ( 15.7 ) - Continuous variables, mean (SD) FBG, mmol/L 8.6 ( 3.9 ) 8.9(3.8) HbA1c, % 7.8 ( 2.9 ) 7.8(2.9) Pulse rate, CPM 75.3 ( 10.6 ) 76(10.4) BMI,kg/m 24.8 ( 3.6 ) 25.4(3.7) Waist circumference, cm 84.3 ( 9.2 ) 85.6(9.2) Hemoglobin,g/L 133.9 ( 17.9 ) 135.1(15.9) Leukocyte,×10 9 / 6.5 ( 3.2 ) 6.5(1.8) Platelet,×10 9 /L 178.1 ( 67.8 ) 189.7(63.6) ALT, U/L 25.3 ( 16.2 ) 24.7(16.3) AST, U/L 24.1 ( 14.2 ) 20.8(14.2) Albumin,g/L 39.5 ( 13.4 ) - Total bilirubin,μmol/L 13.6 ( 10.7 ) 13.2(6.3) Conjugated bilirubin,μmol/L 4.5 ( 2.9 ) - Scr,μmol/L 75.6 ( 33.6 ) 62.8(29.9) Blood urea nitrogen,mmol/L ) 5.7 ( 2.2 ) 5.7(1.9) TC,mmol/L 4.7 ( 1.5 ) 4.8(1.8) TG,mmol/L 1.9 ( 2.4 ) 2.2(2.5) DBP,mmHg 80.1 ( 9.4 ) 80.2(8.9) SBP,mmHg 134.9 ( 16.4 ) 134.3(15.6) n (%), number of patients and percentage over the total number of patients, mean(SD), the mean and standard deviation of the variable Variable and feature screening 426 and 432 variables were removed from the FBG prediction model and the HbA1c prediction model, respectively, during data cleaning, and the specific variables are shown in Supplementary Table 2. Therefore, a total of 85 and 79 variables were finally used for modeling, respectively (Supplementary Table. 3). After inputting data with missing variables, the positive and negative samples are relatively balanced, so no sampling is required. The results of the feature screening are shown in Table 2. The results showed that the most important indicators of the FBG and HbA1c prediction model were the FBG value and HbA1c. The patient's medication compliance, follow-up, dietary habits, BMI, and waist circumference also had a greater impact on the FBG level. In addition, the feature selection results also showed that patients with hypertension or other comorbid diseases, laboratory indicators of related diseases such as Scr, serum alanine aminotransferase, blood urea nitrogen, platelets, and white blood cells, as well as age, smoking, all had a certain degree of influence on FBG. For the HbA1c prediction model, the feature screening results showed that laboratory indicators such as platelets, Scr, AST, Hb, AST, etc. accounted for a large proportion. According to the results of Not inputting and Lasso screening, the feature importance of the data is drawn（Figure 2）, and the feature importance bar chart drawn by other inputting and screening methods is shown in supplementary figures 2-7. Table 2 Feature screening results Inputting method Screening method Top 10 Variables FBG HbA1c Not Lasso Adverse reactions、FBG、Satisfaction with follow-up、Medical Compliance - Good、diet control、Concomitant disease、Medication Adherence - Not Taking Medication、Medical Compliance—general、T2DM、hypertension FBG、Platelets、Satisfaction with follow-up、Adverse reactions、Scr、BMI、SBP、Age、Pulse rate、ALT、AST Not Boruta FBG、platelets、Satisfaction with follow-up、Adverse reactions、Scr、BMI、high blood pressure、age、Pulse rate、ALT FBG、pulse rate、Scr、HBP、BMI、SBP、Waist circumference、AST、Age、daily staple food Modified random forest Lasso HbA1c、Adverse reactions、Satisfaction with follow-up、Medical Compliance - Good、FBG、diet control、Medication Adherence - Not Taking Medication、Concomitant disease、Outpatient follow-up、kidney disease HbA1c、Age、Medication Adherence - Not Taking、Adverse reactions 、Symmetry palpation of dorsalis pedis、pulse rate、SBP、FBG、Current the length of each exercise、BMI Modified random forest Boruta platelets、Adverse reactions、FBG、BMI、blood urea nitrogen、HbA1c、leukocyte、Satisfaction with follow-up、high blood pressure、Scr Age、Hb、HBP、BUN、Scr、AST、HbA1c、PLT、FBG、BMI Model performance The machine learning results are shown in Table 3. The inputting methods used by all best models were all modified random forest inputting: The optimal feature screening method was Not screening and Boruta screening The optimal machine learning methods were ensemble learning and XGBoost. The AUC value of the best model for FBG was model 1 (AUC=0.8190), the worse one was model 5 (AUC=0.8082). The AUC value of the best model for HbA1c was model 1 (AUC=0.9704), the worse one was model 5 (AUC=0.9674). The AUC values of the ten best models were all greater than 0.75, indicating that the prediction model had good prediction performance, and had the possibility of certain clinical application. The ROC curves and P-R curves of the five best models are shown in Figures 3 and 4. Tabel 3 Predictive model building results Model ID AUC Accuracy Precision Recall F1Score Imputing methods Screening methods Models FBG model 1 0.819 0.7439 0.7733 0.6901 0.7293 Modified random forest inputting Not Ensemble learning model 2 0.8163 0.7423 0.7674 0.6955 0.7297 Modified random forest inputting Not XGBoost model 3 0.8119 0.7415 0.7692 0.69 0.7275 Modified random forest inputting Boruta Ensemble learning model 4 0.8087 0.7404 0.769 0.6872 0.7258 Modified random forest inputting Lasso Ensemble learning model 5 0.8082 0.7388 0.7629 0.6929 0.7262 Modified random forest inputting Boruta XGBoost HbA1c model 1 0.9704 0.9217 0.894 0.9463 0.9194 Modified random forest inputting Boruta Ensemble learning model 2 0.9702 0.924 0.9135 0.9268 0.9201 Modified random forest inputting Not Ensemble learning model 3 0.9697 0.924 0.9095 0.9317 0.9205 Modified random forest inputting Lasso Ensemble learning model 4 0.9688 0.9263 0.9179 0.9268 0.9223 Modified random forest inputting Lasso XGBoost model 5 0.9674 0.9171 0.9043 0.922 0.913 Modified random forest inputting Not XGBoost The effect of different data processing methods on the results The results of the ten-fold cross-validation analysis of the data imputing method for FBG are shown in Supplemental Table 4. The modified random forest imputing had the greater impact on the model effect, the AUC values for the FBG and HbA1c models were 0.749±0.044(95%CI=0.745-0.753) and 0.901±0.078 (95%CI=0.894-0.907). The results of the ten-fold cross-validation analysis of the feature screening of the validation set showed that (Supplemental Table 5), Lasso screening had the greatest impact on the model effect, the AUC values for the FBG, and HbA1c models were 0.728±0.038(95%CI=0.725-0.731) and 0.776±0.130(95%CI=0.766-0.785). The analysis results of the data imputing method in the bootstrapping sampling showed that the modified random forest imputing had a greater impact on the model effect（Supplemental Table 6）, the AUC values for the FBG, and HbA1c models were 0.754±0.048(95%CI=0.754-0.755) and 0.902±0.083(95%CI=0.902-0.903). The results of feature screening analysis in bootstrapping sampling showed that Boruta screening had the greatest impact on the FBG model effect, with an AUC value of 0.732±0.040(95%CI=0.731-0.732). For HbA1c, the greatest impact analysis screening method in bootstrapping sampling was Lasso screening, with an AUC value of 0.772±0.131(95%CI=0.772-0.773) ( Supplemental Table 7 ).According to the results of the test set and the validation set, the prediction ability of the model using the modified random forest imputing method was better. The ten prediction models established in the best model all adopted three feature screening methods. Overall, the results of this study can provide a good methodological reference for the establishment of subsequent T2DM prediction models. The effect of different algorithms on the results Hypothesis testing was used to examine the impact of different algorithms on the model's predictive performance. The ten-fold cross-validation results for FBG and HbA1c prediction model are shown in Table 4. The results showed that XGBoost had the greatest impact on the model effect, the AUC values for the FBG and HbA1c models were 0.761±0.029 (95%CI=0.756-0.766) and 0.802±0.159 (95%CI=0.773-0.831). Passive Aggressive had the least impact on the FBG model effect, with an AUC value of 0.610 ±0.052 (95%CI=0.600-0.619) (P<0.0001), and Multinomial Naive Bayes had the least impact on the HbA1c model effect, with an AUC value of 0.662±0.076(95%CI=0.648-0.676). The results of bootstrapping sampling analysis (Supplemental Table 8) showed that among the 16 machine learning models used in this section, ensemble learning had the greatest impact on the model effect, the AUC values of FBG, and HbA1c are respectively 0.767±0.029 and 0.843±0.118. Table 4 Machine learning algorithm ten-fold cross-validation analysis results FBG AUC Accuracy Precision Recall F1Score Mean±SD 95%CI Mean±SD 95%CI Mean±SD 95%CI Mean±SD 95%CI Mean±SD 95%CI AdaBoost 0.741±0.015 0.738-0.744 0.682±0.012 0.680-0.684 0.679±0.019 0.675-0.682 0.677±0.020 0.674-0.681 0.678±0.016 0.675-0.681 Bagging 0.747±0.033 0.741-0.753 0.690±0.026 0.685-0.695 0.694±0.038 0.687-0.701 0.668±0.026 0.664-0.673 0.681±0.029 0.675-0.686 Bernoulli_Naive_Bayes 0.722±0.010 0.721-0.724 0.673±0.008 0.672-0.675 0.663±0.009 0.661-0.665 0.688±0.022 0.684-0.692 0.675±0.015 0.673-0.678 Decision_Tree 0.738±0.019 0.735-0.742 0.685±0.015 0.683-0.688 0.689±0.037 0.682-0.695 0.671±0.038 0.664-0.678 0.678±0.015 0.675-0.681 Extra_Tree 0.724±0.017 0.721-0.727 0.677±0.013 0.674-0.679 0.677±0.025 0.673-0.682 0.663±0.028 0.658-0.668 0.669±0.016 0.667-0.672 Gaussian_Naive_Bayes 0.696±0.014 0.693-0.698 0.647±0.012 0.645-0.649 0.630±0.010 0.628-0.631 0.694±0.030 0.689-0.700 0.660±0.017 0.657-0.663 Gradient_Boosting 0.750±0.018 0.747-0.753 0.690±0.015 0.687-0.692 0.695±0.029 0.689-0.700 0.666±0.017 0.663-0.669 0.680±0.015 0.677-0.682 LDA 0.737±0.012 0.734-0.739 0.678±0.007 0.676-0.679 0.668±0.007 0.666-0.669 0.692±0.028 0.687-0.697 0.679±0.016 0.676-0.682 Logistic_Regression 0.737±0.012 0.735-0.740 0.679±0.008 0.677-0.680 0.670±0.008 0.669-0.672 0.688±0.029 0.683-0.693 0.679±0.017 0.676-0.682 Multinomial_Naive_Bayes 0.707±0.009 0.705-0.708 0.666±0.006 0.665-0.667 0.657±0.006 0.656-0.658 0.678±0.035 0.672-0.684 0.667±0.017 0.664-0.670 Passive_Aggressive 0.610±0.052 0.600-0.619 0.581±0.040 0.574-0.588 0.575±0.044 0.567-0.583 0.579±0.076 0.565-0.593 0.575±0.054 0.565-0.585 QDA 0.723±0.010 0.721-0.725 0.670±0.008 0.669-0.672 0.660±0.008 0.659-0.662 0.686±0.023 0.682-0.690 0.673±0.015 0.670-0.675 Random_Forest 0.756±0.027 0.751-0.761 0.696±0.022 0.692-0.700 0.702±0.033 0.696-0.708 0.669±0.025 0.664-0.674 0.685±0.025 0.680-0.690 SGD 0.736±0.012 0.734-0.739 0.670±0.005 0.669-0.670 0.652±0.004 0.651-0.652 0.712±0.034 0.706-0.718 0.680±0.016 0.677-0.683 XGBoost 0.761±0.029 0.756-0.766 0.698±0.023 0.694-0.702 0.702±0.035 0.696-0.708 0.679±0.021 0.675-0.683 0.690±0.024 0.686-0.694 P value P<0.0001 P<0.0001 P<0.0001 P<0.0001 P<0.0001 HbA1c AUC Accuracy Precision Recall F1Score Mean±SD 95%CI Mean±SD 95%CI Mean±SD 95%CI Mean±SD 95%CI Mean±SD 95%CI AdaBoost 0.782±0.141 0.756-0.807 0.737±0.120 0.716-0.759 0.762±0.130 0.738-0.786 0.737±0.150 0.710-0.764 0.739±0.111 0.719-0.759 Bagging 0.796±0.161 0.767-0.825 0.747±0.143 0.721-0.772 0.763±0.131 0.739-0.786 0.741±0.145 0.715-0.767 0.749±0.133 0.725-0.773 Bernoulli_Naive_Bayes 0.727±0.103 0.708-0.745 0.682±0.101 0.664-0.701 0.683±0.093 0.666-0.700 0.692±0.123 0.670-0.715 0.685±0.100 0.667-0.703 Decision_Tree 0.773±0.142 0.747-0.799 0.742±0.125 0.719-0.764 0.750±0.133 0.726-0.774 0.773±0.101 0.755-0.791 0.756±0.101 0.738-0.775 Extra_Tree 0.740±0.124 0.718-0.763 0.710±0.104 0.691-0.729 0.718±0.116 0.697-0.739 0.740±0.140 0.715-0.765 0.719±0.095 0.702-0.736 Gaussian_Naive_Bayes 0.699±0.114 0.679-0.720 0.606±0.114 0.585-0.626 0.602±0.108 0.582-0.622 0.808±0.176 0.776-0.840 0.671±0.090 0.655-0.687 Gradient_Boosting 0.801±0.151 0.774-0.829 0.747±0.135 0.723-0.772 0.765±0.131 0.742-0.789 0.754±0.130 0.731-0.778 0.755±0.117 0.734-0.776 LDA 0.766±0.132 0.742-0.790 0.716±0.118 0.695-0.737 0.715±0.103 0.696-0.734 0.743±0.126 0.720-0.765 0.726±0.106 0.707-0.745 Logistic_Regression 0.779±0.144 0.753-0.805 0.743±0.129 0.720-0.767 0.749±0.120 0.727-0.771 0.761±0.138 0.736-0.786 0.751±0.118 0.730-0.773 Multinomial_Naive_Bayes 0.662±0.076 0.648-0.676 0.603±0.054 0.593-0.613 0.602±0.051 0.592-0.611 0.613±0.161 0.584-0.642 0.599±0.086 0.583-0.614 Passive_Aggressive 0.694±0.148 0.668-0.721 0.649±0.131 0.625-0.672 0.646±0.148 0.620-0.673 0.648±0.186 0.615-0.682 0.641±0.156 0.612-0.669 QDA 0.730±0.111 0.710-0.750 0.680±0.100 0.662-0.698 0.674±0.088 0.658-0.690 0.719±0.109 0.699-0.738 0.694±0.090 0.677-0.710 Random_Forest 0.791±0.152 0.763-0.818 0.740±0.135 0.715-0.764 0.762±0.131 0.738-0.786 0.731±0.129 0.708-0.754 0.743±0.121 0.721-0.765 SGD 0.785±0.140 0.760-0.810 0.740±0.132 0.716-0.764 0.744±0.125 0.721-0.767 0.767±0.127 0.745-0.790 0.752±0.116 0.731-0.773 XGBoost 0.802±0.159 0.773-0.831 0.753±0.146 0.727-0.779 0.768±0.138 0.743-0.793 0.754±0.143 0.728-0.780 0.758±0.134 0.734-0.782 P value P<0.0001 P<0.0001 P<0.0001 P<0.0001 P<0.0001 Discussion Our research is based on machine learning methods and aims to build artificial intelligence prediction models for FBG and HbAlc in patients with T2DM. By analyzing the influencing factors of related blood glucose control indicators in T2DM patients, we established two prediction models for FBG and HbAlc, and they can assist clinical treatment and T2DM patient management, to allow early adjustment of the treatment plan and improve the treatment rate and control rate of T2DM. Our analysis results show that: 1) FBG, HbA1c, medication compliance, and dietary habits have a greater impact on both prediction models, 2) The multi-parameter predictive risk models incorporate variables from different domains, including baseline demographics, complications, and laboratory tests, and can accurately predict three-month FBG values and HbAlc values, 3) In the machine learning-driven algorithm, the optimal models both adopt the ensemble learning algorithm. The influence of different algorithms and data processing methods on the results shows that the algorithms that have the greatest impact on the model effect are ensemble learning and XGBoost, and the best imputing method is the modified random forest imputing. Ensemble learning is an advanced machine learning strategy that can improve classification performance and generalization by combining multiple models [18] . Nemat H et al. utilized deep learning and ensemble learning to predict blood glucose levels, and compared the performance of the proposed ensemble model with the non-ensemble model, and the results showed that the developed ensemble model outperformed the non-ensemble baseline model [19] . In our research, we combine the remaining model indicators that have been trained, and the integrated model indicators have great advantages. From the feature screening results, this study is consistent with other studies, FBG and HbAlc are both the most important predictors. Del Parigi A et al. [20] used several machine learning algorithms to find predictors of glycemic control in diabetes and found that HbA1c and FPG were the strongest predictors of achieving glycemic control. This is consistent with our findings. For this result, we explain that current FBG and HbA1c values have important effects on future FBG and HbA1c values, respectively. In our study, FBG and HbAlc were mutually important predictors, indicating an important correlation between glycated hemoglobin and fasting blood glucose. The reason maybe is that once the glucose in human blood combines with hemoglobin to form glycosylated hemoglobin, it will age with the aging of red blood cells, which is the product of an irreversible glycation reaction. The contact time between blood glucose and hemoglobin and the content of blood glucose can determine the level of HbA1c, so the content of HbA1c is positively correlated with the blood glucose content of diabetic patients, which may have the ability to predict each other. Studies have shown that postprandial blood glucose and fasting blood glucose are closely related to glycosylated hemoglobin, and for poorly controlled diabetic patients, the greater the value of HbA1c, the greater the contribution of fasting blood glucose value [21] . Wang J et al [22] .established a blood glucose prediction model. After feature screening, the top six indicators were: age, fasting glutamate transaminase (ALT), blood urea nitrogen (BUN), total protein (TP), uric acid (UA), and BMI. BMI is also in the top ten important features in this study, and the rest of the indicators did not enter the top ten in importance. However, our study also found that ALT and BUN will have a certain degree of influence on blood glucose. Wang YS et al established a T2DM prediction model in western Xinjiang, China, and used Lasso screening for feature screening [23] . The study showed that age, family history of T2DM, waist circumference, TC, TG, BMI, HDL-C, and previous history of hypertension had a significant impact on FBG. These factors are included in the feature selection results of our study. Chien KL et al [24] used multiple logistic regression to predict HbA1c and found that both waist circumference and BMI were associated with abnormal glycated hemoglobin levels. Age, family history of diabetes, systolic blood pressure, and biochemical markers including C-reactive protein and triglycerides were significantly associated with higher glycated hemoglobin levels. In our study, waist circumference and BMI had important effects on HbA1c, as did age and hypertension, but the study did not take into account enough variables, and our accuracy is higher. In addition, our study, based on a large sample of physical examination and follow-up data, found that patients' medication compliance, follow-up conditions, and living habits (including dietary habits and smoking) had a greater impact on blood glucose control. The results better clarify the importance of primary prevention of T2DM, which is to focus on changing environmental factors and lifestyles, reducing calorie intake, maintaining a low-salt, low-sugar, high-fiber diet, quitting smoking, limiting alcohol, and getting daily moderate exercise. At the same time, the results of this study also show the importance of follow-up for secondary and tertiary prevention of T2DM. Pourat N et al [25] conducted an observational study and found that timely linking behavioral health patients to outpatient follow-up after hospitalization is an effective care transition strategy that may reduce readmission rates. Tong L et al [26] also concluded that follow-up was associated with a reduced risk of readmission. Patients benefited the most from outpatient follow-up because face-to-face conversations allowed more information (both therapeutic and emotional) to be exchanged with patients and better individualized care for patients. In addition, adverse reaction monitoring during the follow-up process can timely detect the risk of hypoglycemia in patients, which brings greater benefits to patients. Most studies tend to use machine learning algorithms such as decision trees, random forests, SVM, logistic regression, and neural networks to build T2DM prediction models, with AUC values ranging from 0.7 to 0.9 [27.28] . Wang J et al adopted three commonly used machine learning algorithms (RF, SVM, and BP-ANN) combined with the elastic network (EN) to simulate and predict blood glucose status in China. The AUCs of RF, SVM and BP were 0.75, 0.72 and 0.72, respectively [29] . In a study of T2DM prediction models in Australia by Zhang L et al [30] , the model built using XGBoost had the best prediction ability, with a 3-year prediction model AUC value of 0.78 and a 10-year AUC value of 0.75. Xue M et al [31] established a T2DM prediction model using algorithms such as decision trees, random forests, AdaBoost with decision trees (AdaBoost), and extreme gradient boosting decision trees (XGBoost), and XGBoost had the best performance (AUC = 0.968). Usually, the AUC value is above 0.8, showing a good classification effect [32] . The AUC values of the five optimal FBG models obtained in our study are all greater than 0.8 and the AUC values of the five optimal HbA1c models obtained in our study are all greater than 0.9. Based on incorporating 100,000 pieces of data, 85 variables, and 16 machine learning algorithms for research, we obtained the FBG prediction model with the best AUC value of 0.819 and the HbA1c prediction model with the best AUC value of 0.970, indicating that thes e prediction models have better performance and better clinical prediction ability. The establishment of thes e two models can imput the current gap in the prediction model of individualized treatment of T2DM patients, provide new ideas and methods for T2DM treatment, and provide T2DM patients with efficient and accurate individualized treatment plans to solve the real health problems of patients. In addition, this study explores the T2DM prediction model based on real-world medical data mining, uses multiple classifiers for comparative research, and selects the optimal model to ensure the optimization of the model, which effectively makes up for the current shortcomings of using a single classifier. Therefore, this study comprehensively and completely demonstrated the process of predicting the outcome of T2DM drug treatment with the help of data mining technology under the background of real-world research and provided a good methodological reference for the management of other chronic diseases. Strengths The data in this study came from the Public Health Service System and the Medical Record Homepage Management System of the Health Information Center of Sichuan Province. The data quality is reliable enough to meet the needs of modeling. In the process of data cleaning, this study is not limited to a single data preprocessing method but uses a variety of imputing methods, feature screening methods, and various data preprocessing methods and applies them to the data cleaning of each predictive indicator. The process avoids the possible impact of a certain data preprocessing method on the modeling effect. The modeling method has been improved in this study. Different from the previous use of one or several algorithms to build predictive models, our study used more than ten machine learning algorithms for modeling and selected the optimal five models. The model results are more reliable and the prediction performance is better. The information included in the study is more comprehensive, including the basic information about patients, disease-related factors, treatment factors, metabolic index factors, and lifestyle-related factors. Limitations 1) This study only uses medical data from Sichuan Province, China for modeling. Differences in lifestyle and ethnicity in different regions may lead to a limited scope of application of the model. 2) Although the data set used for validation in this study is independent of the data set used for model development, the two are derived from practice records in the same database, and no more rigorous prospective external validation has been performed. 3) In this study, the classification of some variables may be wrong because the system automatically recognizes variables with more than 10 categories as continuous variables, but this operation will not affect the prediction effect of the final prediction model. 4) In this study, the AUC values of the two optimal prediction models differed by 0.1, possibly due to too little HbA1c data for modeling. In future studies, if the amount of data used to build the model can be increased, the difference in values may be reduced. 5) Some predictive factors, such as the course of diabetes, are not recorded in detail in the original database, so these variables are not included in the modeling, which may have an impact on the prediction results. In the following research, we will use more extensive and detailed data for modeling to obtain a more accurate prediction model. Conclusion In this study, the three-month FBG prediction model and three-month HbA1c prediction model for T2DM patients were constructed using population data from Sichuan Province, China. The patient's FBG and HbA1c are both the most important predictors of two kinds of prediction models. This research can provide a methodological reference for other prediction models. The AUC values of the five best FBG prediction models finally established are all greater than 0.8, and the AUC values of the five best HbA1c prediction models finally established are all greater than 0.9, which could accurately predict the FBG and HbA1c for clinical applications. Abbreviations ALT Alanine aminotransferase AST Aspartate aminotransferase AUC Area under the curve BMI Body mass index BP-ANN Back-propagation artificial neural network BUN Blood urea nitrogen CI Confidence intervals CPM Cycles per minute DBP Diastolic blood pressure EN Elastic network FBG Fasting blood glucose Fig Figure Hb Hemoglobin HbA1c Glycosylated hemoglobin HBP High blood pressure HDL-C High density lipoprotein cholesterol LDA Linear discriminant analysis P-R curve Precision-recall curve QDA Quadratic discriminant analysis RF Random forest ROC Receiver operating characteristic curve SBP Systolic blood pressure Scr Serum creatinine SD Standard deviation SGD Stochastic Gradient Descent SVM Support vector machine TC Total cholesterol TG Triglyceride TP Total protein UA Uric acid Declarations Ethics approval and consent to participate This study was approved by the Medical Ethics Committee of the Sichuan Academy of Medical Sciences (Sichuan Provincial People's Hospital). Details are in the supplementary material. A statement to confirm that all methods were carried out in accordance with relevant guidelines and regulations (declaration of helsinki). We collected patient data retrospectively, so informed consent could not be obtained from all participants. However, we used a unique ID to identify patient connection information, and all research operations carried out would not be traced to the individual patient, and the patient's sensitive personal information (such as name, phone number, address, work unit, responsible doctor, etc.) would be deleted. The study has passed the ethical review, and patient privacy will not be disclosed. Consent for publication Not applicable Availability of data and materials The data that support the findings of this study are available from Sichuan Provincial Health Commission, but restrictions apply to the availability of these data, which were used under license for the current study, and so are not publicly available. Data are however available from the authors upon reasonable request and with permission of Sichuan Provincial Health Commission. Competing interests The authors declare that they have no competing interests. Funding EWL was funded by the Sichuan Provincial Department of Science and Technology (Grant No. 2021YFS0197). MJ is supported by the Personalized Drug Therapy Key Laboratory of Sichuan Provincial People’s Hospital. XWW was supported by the National Natural Science Foundation of China (grant number: 72004020). Authors' contributions EWL and XT contributed to the idea and design of the study. MJ and BQZ conducted the literature search. YML and QH conducted the data analysis and created the figures. YML wrote the manuscript with support from WMG, MJ, XT, and XWW. YX, JQH, XS, RST and EWL provided critical feedback on the analysis and its interpretation and commented on the drafted manuscript. WYL, RST, XWW and EWL provided final approval of the manuscript for publication. XT is responsible for the integrity of the work as a whole. Acknowledgements Not applicable Authors' information (optional) Not applicable References American Diabetes Association. Introduction: Standards of Medical Care in Diabetes-2022. Diabetes Care. 2022;45(Suppl 1):S1-S2. Sun H, Saeedi P, Karuranga S, et al. IDF Diabetes Atlas: Global, regional and country-level diabetes prevalence estimates for 2021 and projections for 2045. Diabetes Res Clin Pract. 2022 Jan;183:109119. Ogurtsova K, Guariguata L, Barengo NC, et al. IDF diabetes Atlas: Global estimates of undiagnosed diabetes in adults for 2021. Diabetes Res Clin Pract. 2022;183:109118. Zheng Y, Ley SH, Hu FB. Global aetiology and epidemiology of type 2 diabetes mellitus and its complications. Nat Rev Endocrinol. 2018 Feb;14(2):88-98. Harding JL, Pavkov ME, Magliano DJ,et al. Global trends in diabetes complications: a review of current evidence. Diabetologia. 2019 Jan;62(1):3-16. Safiri S, Karamzad N, Kaufman JS, et al. Prevalence, Deaths and Disability-Adjusted-Life-Years (DALYs) Due to Type 2 Diabetes and Its Attributable Risk Factors in 204 Countries and Territories, 1990-2019: Results From the Global Burden of Disease Study 2019. Front Endocrinol (Lausanne). 2022 Feb 25;13:838027. Xiao W, Jing L, Xu Y, et al. Different Data Mining Approaches Based Medical Text Data. J Healthc Eng. 2021 Dec 6;2021:1285167. Makridakis S, Spiliotis E, Assimakopoulos V. Statistical and Machine Learning forecasting methods: Concerns and ways forward. PLoS One. 2018 Mar 27;13(3):e0194889. Liao WC, Tu YK, Wu MS, et al. Blood glucose concentration and risk of pancreatic cancer: systematic review and dose-response meta-analysis. BMJ. 2015 Jan 2;350:g7371. Nagalakshmi CS, Santhosh NU, Krishnamurthy N, et al. Role of Altered Venous Blood Lactate and HbA1c in Women with Gestational Diabetes Mellitus. J Clin Diagn Res. 2016 Dec;10(12):BC18-BC20. Jia W, Weng J, Zhu D, et al; Chinese Diabetes Society. Standards of medical care for type 2 diabetes in China 2019. Diabetes Metab Res Rev. 2019 Sep;35(6):e3158. Fregoso-Aparicio L, Noguez J, Montesinos L, et al. Machine learning and deep learning predictive models for type 2 diabetes: a systematic review. Diabetol Metab Syndr. 2021 Dec 20;13(1):148. Zhu T, Li K, Herrero P, Georgiou P. Deep Learning for Diabetes: A Systematic Review. IEEE J Biomed Health Inform. 2021 Jul;25(7):2744-2757. WHO. World Health Organization. ICD-10 version:2010 [Internet], 2010. Available from https://icd.who.int/browse10/2019/en. Accessed 16 March 2021. Greenwood CJ, Youssef GJ, Letcher P, et al. A comparison of penalised regression methods for informing the selection of predictive markers. PLoS One. 2020;15(11):e0242730. Published 2020 Nov 20. Zhao Z, Yang W, Zhai Y, et al. Identify DNA-Binding Proteins Through the Extreme Gradient Boosting Algorithm. Front Genet. 2022;12:821996. Published 2022 Jan 28. Yang H., Luo Y., Ren X., et al. Risk Prediction of Diabetes: Big Data Mining with Fusion of Multifarious Physical Examination Indicators. Inf. 2021; 75:140–149. Jin LP, Dong J. Ensemble Deep Learning for Biomedical Time Series Classification. Comput Intell Neurosci. 2016;2016:6212684. doi: 10.1155/2016/6212684. Epub 2016 Sep 20. Nemat H, Khadem H, Eissa MR, et al. Blood Glucose Level Prediction: Advanced Deep-Ensemble Learning Approach. IEEE J Biomed Health Inform. 2022 Jan 25;PP. Del Parigi A, Tang W, Liu D, et al. Machine Learning to Identify Predictors of Glycemic Control in Type 2 Diabetes: An Analysis of Target HbA1c Reduction Using Empagliflozin/Linagliptin Data. Pharmaceut Med. 2019 Jun;33(3):209-217. Makris K, Spanou L, Rambaouni-Antoneli A, et al. Relationship between mean blood glucose and glycated haemoglobin in Type 2 diabetic patients. Diabet Med. 2008 Feb;25(2):174-8. Wang J, Wang F, Liu Y, et al. Multiple Linear Regression and Artificial Neural Network to Predict Blood Glucose in Overweight Patients. Exp Clin Endocrinol Diabetes. 2016 Jan;124(1):34-8. Wang YS, Zhang YS, Wang K, et al. Nomogram Model for Screening the Risk of Type II Diabetes in Western Xinjiang, China.[J] .Diabetes Metab Syndr Obes, 2021, 14: 3541-3553. Chien KL, Lin HJ, Lee BC, et al. Prediction model for high glycated hemoglobin concentration among ethnic Chinese in Taiwan. Cardiovasc Diabetol. 2010 Sep 27;9:59. Pourat N, Chen X, Wu SH, Davis AC. Timely Outpatient Follow-up Is Associated with Fewer Hospital Readmissions among Patients with Behavioral Health Conditions. J Am Board Fam Med. 2019 May-Jun;32(3):353-361. Tong L, Arnold T, Yang J, et al. The association between outpatient follow-up visits and all-cause non-elective 30-day readmissions: A retrospective observational cohort study. PLoS One. 2018 Jul 17;13(7):e0200691. De Silva K, Enticott J, Barton C, et al. Use and performance of machine learning models for type 2 diabetes prediction in clinical and community care settings: Protocol for a systematic review and meta-analysis of predictive modeling studies. Digit Health. 2021 Sep 28;7:20552076211047390. Kim H, Lim DH, Kim Y. Classification and Prediction on the Effects of Nutritional Intake on Overweight/Obesity, Dyslipidemia, Hypertension and Type 2 Diabetes Mellitus Using Deep Learning Model: 4-7th Korea National Health and Nutrition Examination Survey. Int J Environ Res Public Health. 2021 May 24;18(11):5597. Wang J, Wang MY, Wang H, et al. Status of glycosylated hemoglobin and prediction of glycemic control among patients with insulin-treated type 2 diabetes in North China: a multicenter observational study. Chin Med J (Engl). 2020 Jan 5;133(1):17-24. Zhang L, Shang X, Sreedharan S, et al. Predicting the Development of Type 2 Diabetes in a Large Australian Cohort Using Machine-Learning Techniques: Longitudinal Survey Study. JMIR Med Inform. 2020 Jul 28;8(7):e16850. Xue M, Su Y, Li C, et al. Identification of Potential Type II Diabetes in a Large-Scale Chinese Population Using a Systematic Machine Learning Framework. J Diabetes Res. 2020 Sep 24;2020:6873891. Mandrekar JN. Receiver operating characteristic curve in diagnostic test assessment. J Thorac Oncol. 2010 Sep;5(9):1315-6. Additional Declarations No competing interests reported. Supplementary Files Supplementary.docx Cite Share Download PDF Status: Posted Version 1 posted 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-1868105","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":129886036,"identity":"a8788804-d9c0-448a-9035-c9eeb8b48535","order_by":0,"name":"Xue Tao","email":"","orcid":"","institution":"Sichuan Academy of Medical Sciences \u0026 Sichuan Provincial People's Hospital","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Xue","middleName":"","lastName":"Tao","suffix":""},{"id":129886042,"identity":"9fc7db69-c83f-4c53-b57c-b9089818e021","order_by":1,"name":"Min Jiang","email":"","orcid":"","institution":"Sichuan Academy of Medical Sciences \u0026 Sichuan Provincial People's Hospital","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Min","middleName":"","lastName":"Jiang","suffix":""},{"id":129886046,"identity":"1128f25f-3d2c-405f-8668-d06b8d9f36b3","order_by":2,"name":"Yumeng Liu","email":"","orcid":"","institution":"Southwest Medical University","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Yumeng","middleName":"","lastName":"Liu","suffix":""},{"id":129886047,"identity":"45d35ffa-4d12-413f-9f34-a2813ca6c723","order_by":3,"name":"Qi Hu","email":"","orcid":"","institution":"Sichuan Academy of Medical Sciences \u0026 Sichuan Provincial People's Hospital","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Qi","middleName":"","lastName":"Hu","suffix":""},{"id":129886048,"identity":"e774180e-5205-4827-a934-734f0012f9a6","order_by":4,"name":"Baoqiang Zhu","email":"","orcid":"","institution":"Southwest Medical University","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Baoqiang","middleName":"","lastName":"Zhu","suffix":""},{"id":129886050,"identity":"0ab6c657-54fc-4352-a156-22530a895ea7","order_by":5,"name":"Jiaqiang Hu","email":"","orcid":"","institution":"Sichuan Academy of Medical Sciences \u0026 Sichuan Provincial People's Hospital","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Jiaqiang","middleName":"","lastName":"Hu","suffix":""},{"id":129886052,"identity":"93ebbad4-dadb-4597-ae89-b0c946e7baea","order_by":6,"name":"Wenmei Guo","email":"","orcid":"","institution":"Sichuan Academy of Medical Sciences \u0026 Sichuan Provincial People's Hospital","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Wenmei","middleName":"","lastName":"Guo","suffix":""},{"id":129886054,"identity":"f92b12a4-e00f-440a-abf0-f3dda9b78bdc","order_by":7,"name":"Xingwei Wu","email":"","orcid":"","institution":"Sichuan Academy of Medical Sciences \u0026 Sichuan Provincial People's Hospital","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Xingwei","middleName":"","lastName":"Wu","suffix":""},{"id":129886056,"identity":"6ffbd16c-9b88-4f7b-817b-a783322ebb79","order_by":8,"name":"Yu Xiong","email":"","orcid":"","institution":"Chinese Academy of Medical Sciences, Peking Union Medical College","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Yu","middleName":"","lastName":"Xiong","suffix":""},{"id":129886058,"identity":"e0540f87-2c45-444f-928e-3cae9c8f74ed","order_by":9,"name":"Xia Shi","email":"","orcid":"","institution":"Sichuan Academy of Medical Sciences \u0026 Sichuan Provincial People's Hospital","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Xia","middleName":"","lastName":"Shi","suffix":""},{"id":129886061,"identity":"182d8987-b3d1-4f98-b9d1-1f605bca76e6","order_by":10,"name":"Wenyuan Li","email":"","orcid":"","institution":"Sichuan Academy of Medical Sciences \u0026 Sichuan Provincial People's Hospital","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Wenyuan","middleName":"","lastName":"Li","suffix":""},{"id":129886064,"identity":"eba4cf2b-c8c9-4cc2-89fb-55319f08608f","order_by":11,"name":"Rongsheng Tong","email":"","orcid":"","institution":"Sichuan Academy of Medical Sciences \u0026 Sichuan Provincial People's Hospital","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Rongsheng","middleName":"","lastName":"Tong","suffix":""},{"id":129886065,"identity":"01303317-07ed-473c-bdfd-b60301f6f2b3","order_by":12,"name":"Enwu Long","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA8ElEQVRIiWNgGAWjYDCCA2DSRo6xmfnggw8GNnLEakkzZm5nSzacUZBmTKyWw4nt/Txq0jwfDicS1MF3+/DjFx93HDbmbeZhk7YxYE5gYD98dAM+LZLn0swsZ55Jl5Ns5j1snWPAlsfAk5Z2A58WgzMMZsa8bdbGhs18ibdzDHiKGSR4zAhoYf8G1MKcuP8wj4G0hYFEYgNhLTzGj3nbnBMbm3mMpBkMDAhrkTzDU8Y4sy3NmLEZGMg9BgnGbIT8wneGffOHj23AqOw/fPDBjz//5fjZDx/DqwUI2CRQuQSUgwDzByIUjYJRMApGwUgGAJG3SsDEuqwuAAAAAElFTkSuQmCC","orcid":"","institution":"Sichuan Academy of Medical Sciences \u0026 Sichuan Provincial People's Hospital","correspondingAuthor":true,"submittingAuthor":false,"prefix":"","firstName":"Enwu","middleName":"","lastName":"Long","suffix":""}],"badges":[],"createdAt":"2022-07-18 03:44:13","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-1868105/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-1868105/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":26505007,"identity":"9425d1c2-db88-4942-bdaf-a40d4ecc3fe8","added_by":"auto","created_at":"2022-09-15 12:47:50","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":68755,"visible":true,"origin":"","legend":"\u003cp\u003eData analysis flow chart\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-1868105/v1/4222393b7fefbdc2bb01174c.png"},{"id":26505006,"identity":"94deba45-ce2a-4710-b01c-8c7dcdb872a0","added_by":"auto","created_at":"2022-09-15 12:47:50","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":611961,"visible":true,"origin":"","legend":"\u003cp\u003eFeature importance bar chart –A: FBG; B: HbA1c (Inputting method: Not; Screening method: Lasso Screening)\u003c/p\u003e","description":"","filename":"2.png","url":"https://assets-eu.researchsquare.com/files/rs-1868105/v1/998ba503f9590e3fc7dd36b4.png"},{"id":26505009,"identity":"64f507f8-dd49-4cad-b5c5-40eafcfcb107","added_by":"auto","created_at":"2022-09-15 12:47:50","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":137362,"visible":true,"origin":"","legend":"\u003cp\u003eROC curves of the five best predictive models(A:FBG:B: HbA1c)\u003c/p\u003e","description":"","filename":"3.png","url":"https://assets-eu.researchsquare.com/files/rs-1868105/v1/f0a16d0c38d7edb0206faf76.png"},{"id":26505008,"identity":"91fff741-2997-4c3e-a203-1eec4d34a53a","added_by":"auto","created_at":"2022-09-15 12:47:50","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":92643,"visible":true,"origin":"","legend":"\u003cp\u003eP-R curves of the five best predictive models(A:FBG:B: HbA1c)\u003c/p\u003e","description":"","filename":"4.png","url":"https://assets-eu.researchsquare.com/files/rs-1868105/v1/b169c4349b6ebc237e48bb3b.png"},{"id":27306432,"identity":"9c7286ff-01d9-4806-9c66-a22cf659ac14","added_by":"auto","created_at":"2022-10-04 07:44:31","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1689335,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-1868105/v1/3c1ea7bf-04bb-4ff9-9142-ba121740330d.pdf"},{"id":26505324,"identity":"397df4d2-94e4-413c-94ac-eb35d816fc3f","added_by":"auto","created_at":"2022-09-15 12:52:50","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":4350900,"visible":true,"origin":"","legend":"","description":"","filename":"Supplementary.docx","url":"https://assets-eu.researchsquare.com/files/rs-1868105/v1/11fb8f57797ce40df10aa1f7.docx"}],"financialInterests":"No competing interests reported.","formattedTitle":"Predicting three-month fasting blood glucose and glycated hemoglobin of patients with type 2 diabetes based on multiple machine learning algorithms","fulltext":[{"header":"Introduction","content":"\u003cp\u003eDiabetes mellitus is a chronic progressive disease characterized by disorders of glucose metabolism\u003csup\u003e[1]\u003c/sup\u003e. In recent years, the global incidence of diabetes has been increasing year by year. According to the latest epidemiological data from the International Diabetes Federation(IDF): the global diabetes prevalence in 20-79 years old in 2021 was estimated to be 10.5% (536.6 million people), rising to 12.2% (783.2 million) in 2045\u003csup\u003e[2]\u003c/sup\u003e. And\u0026nbsp;in 2021, almost one in two adults (20-79 years old) with diabetes were unaware of their diabetes status (44.7%, 239.7 million)\u003csup\u003e[3]\u003c/sup\u003e. Type 2 diabetes mellitus (T2DM) patients accounted for more than 90.0% of the total diabetic patients\u003csup\u003e[4]\u003c/sup\u003e. With the development of the disease, most T2DM patients will have different degrees of complications, which will seriously reduce the quality of life of the patients and bring a heavy economic burden to the patients\u0026apos; families\u003csup\u003e[5]\u003c/sup\u003e. And the severity of complications is inseparable from glycemic control. Therefore, active, safe, and effective blood glucose control has positive significance for preventing complications, improving the quality of life of T2DM patients, and reducing the economic burden on patients and society.\u003c/p\u003e\n\u003cp\u003eThe global age-standardized point prevalence and death rates for T2DM were 5282.9 and 18.5 per 100 000, an increase of 49% and 10.8%, respectively, since 1990\u003csup\u003e[6]\u003c/sup\u003e. Serious challenges exist in diabetes prevention and glycemic control. The reason may be that the individual differences of patients, such as physiology and pathology, diet structure, and lifestyle, are not fully considered in the treatment process. Therefore, it is urgent to establish an efficient, accurate, and economical T2DM prediction model to improve the treatment rate of T2DM in medical institutions at all levels.\u003c/p\u003e\n\u003cp\u003eWith the continuous development of database and data mining technology, data mining is more and more used to mine medical databases efficiently\u003csup\u003e[7]\u003c/sup\u003e. The existing data mining technology application research shows that the model established by data mining has high accuracy\u003csup\u003e[8]\u003c/sup\u003e. Fasting blood glucose (FBG) is the blood glucose value detected in the plasma collected before breakfast after an overnight fast (at least 8-10 hours without any food, except for drinking water)\u003csup\u003e[9]\u003c/sup\u003e, which can be used to reflect the secretion function of basal insulin. Glycated hemoglobin (HbA1c) is the product of the combination of hemoglobin in red blood cells and blood glucose, and its content depends on the blood glucose concentration and the contact time between blood glucose and, and has nothing to do with factors such as blood drawing time and whether the patient is fasting\u003csup\u003e[10]\u003c/sup\u003e. They are used to detect blood glucose and both are important indicators for diagnosing diabetes and reflecting the prognosis of diabetes\u003csup\u003e[11]\u003c/sup\u003e. At present, prediction models based on machine learning algorithms are mainly used for the prediction of diabetes and its complications, and there are few studies on the prediction models of patients\u0026apos; glycemic control after medication\u003csup\u003e[12.13]\u003c/sup\u003e. Based on this, this study intends to establish artificial intelligence prediction models for the compliance of two blood glucose indicators in T2DM patients after 3 months of treatment through data mining, to explore potential predictive relationships between FBG and HbA1c, improve the treatment rate and control rate of T2DM, reduce the incidence of adverse reactions, and prevent and reduce the occurrence of complications.\u003c/p\u003e"},{"header":"METHODS","content":"\u003ch2\u003eStudy design and data source\u003c/h2\u003e\n\u003cp\u003eThe data of this study were obtained from the Public Health Service System and the Medical Record Homepage Management System of the Health Information Center of Sichuan Province, China (including personal basic information form, health check-up form, and follow-up service record form), and the overall data were derived from patients who received anti-diabetic drugs or had the International Classification of Diseases Tenth Revision (ICD-10) code[\u003csup\u003e14]\u0026nbsp;\u003c/sup\u003efor type 2 diabetes between January 2015 and December 2020.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eA total of 375,723 T2DM patients\u0026apos; related diagnosis and treatment data were collected in this study, and the available data for constructing the FBG prediction model and the HbA1c prediction model were screened according to the following criteria: if the same patient had 2 or more registration data within 3\u0026plusmn;1 months, the patient\u0026apos;s data was available, If there were multiple sets of data within 3\u0026plusmn;1 months, the data closest to 3 months from the baseline should be taken, if there were multiple sets of consistent data longitudinally for the same patient (a patient had multiple sets of data that met the requirements of having 2 or more data within 3\u0026plusmn;1 months), a group was randomly selected for inclusion. FBG control outcomes judgment standard: (1)well-controlled, FBG was 4.4-7.0 mmol/L, (2)poorly controlled, FBG＞7.0 mmol/L or FBG＜4.4 mmol/L. Judgment criteria for HbAlc value: (1)well-controlled, HbA1c\u0026lt;7% , (2)poorly controlled, the value of HbA1c was not in this range.\u003c/p\u003e\n\u003cp\u003eThe data included the patient\u0026apos;s basic information, drug use, test indicators, and living and diet, as well as the actual follow-up of the patient after treatment. This study used a unique ID to identify patient connection information, and all research operations carried out would not be traced to the individual patient, and the patient\u0026apos;s sensitive personal information (such as name, phone number, address, work unit, responsible doctor, etc.) would be deleted. All files were encrypted during transmission and use, and documents were received by a password. This study has passed the ethical review, the approval document in Supplementary Fig 1.\u003c/p\u003e\n\u003cp\u003eIn this study, a total of 511 variables were included, which were named X1\u0026ndash;X511 for statistical convenience(Detailed variables are shown in Supplemental Table 1). Data analysis was performed using named variables, and the variable names were restored after the model evaluation process was complete.\u003c/p\u003e\n\u003ch3\u003eData cleaning\u003c/h3\u003e\n\u003cp\u003eWe deleted variables with a missing ratio of 90%, a single category ratio of 90%, and variables with a coefficient of variation less than 0.1. These variables had little impact on the establishment of the model, and the analysis was meaningless, so they were deleted. Data inputting was done using not inputting and random forest inputting. After the data were inputted, if there was a large difference between the positive and negative sample sizes, the data was balanced by sampling. And we modified outliers to the maximum or minimum value of the norm.\u003c/p\u003e\n\u003cp\u003eThe method of \u0026ldquo;not inputting\u0026rdquo; was to delete the missing columns and the missing rows in the data in turn, and finally, we got the data without missing values. The \u0026ldquo;modified random forest inputting\u0026rdquo; meant that by continuously introducing the inputted columns into the model, as the amount of data continued to accumulate, the obtained values had a higher accuracy rate, which could achieve a more accurate prediction of missing values.\u003c/p\u003e\n\u003cp\u003eAfter the data were inputted, the data were divided into training and test sets for machine learning. And the number of training sets accounted for 80% of the total sample size, and the number of test sets accounted for 20% of the total sample size.\u003c/p\u003e\n\u003ch3\u003eFeature screening\u003c/h3\u003e\n\u003cp\u003eThe data were screened using three methods: Not screening, Lasso screening, and Boruta screening. Feature screening is an important aspect of model building, which helps to exclude relevant variables, biases, and limitations of unnecessary noise, making the final analysis results closer to reality. Lasso are a useful atheoretical approach for both developing predictive models and selecting key indicators within an often substantially larger pool of available indicators by inputting all latent variables at the same time, reducing bias caused by unimportant variables, and selecting only the most important variables from a potentially large initial pool[\u003csup\u003e15]\u003c/sup\u003e.\u003c/p\u003e\n\u003ch3\u003eModel training\u003c/h3\u003e\n\u003cp\u003e16 or 18 kinds of machine learning algorithms were used for model training, and the data after feature screening were modeled respectively. The specific machine learning algorithm models used included: Logistic regression, Stochastic Gradient Descent（SGD）, Decision Tree, Gaussian Naive Bayes, Bernoulli Naive Bayes, Multinomial Naive Bayes, Quadratic Discriminant Analysis (QDA), Random Forest, Extra Tree, Linear Discriminant Analysis (LDA), Passive Aggressive, AdaBoost, Begging, Gradient Boosting, XGBoost, and Ensemble Learning.\u0026nbsp;In 2011, Tianqi Chen and Carlos Guestrin first proposed the XGBoost algorithm, or the extreme gradient boosting algorithm. It is a machine learning model that achieves stronger learning effects by integrating multiple weak learners[\u003csup\u003e16]\u003c/sup\u003e,\u0026nbsp;and has better flexibility and scalability[\u003csup\u003e17]\u003c/sup\u003e. Compared with general machine learning algorithms, the XGBoost model shows strong advantages.\u0026nbsp;These machine learning algorithms have their strengths, among which, the ensemble learning model is an evaluation index based on the trained model, summarizing the best model and outputting according to the voting principle. The evaluation indicators of the prediction model included Area Under Curve(AUC), Accuracy, Precision, Recall, and F1 Score. According to the machine learning results, the 5 models with the best prediction performance were selected and their receiver operating characteristic curve（ROC）and P-R curves were drawn.\u003c/p\u003e\n\u003ch3\u003eModel verification\u003c/h3\u003e\n\u003cp\u003eTen-fold cross-validation and bootstrapping sampling were used to verify the impact of different preprocessing algorithms and different machine learning algorithms on the prediction of building FBG and HbAlc models. The model with the largest AUC was selected and constructed using 10 subsets (randomly drawn 10%\u0026ndash;100% of the total sample size) to assess the effect of different sample sizes on predictive power. Each subset was split 4:1 into a training set and a test set, and the AUC calculated from the test set was used for sample size checking. By transforming randomly sampled data, 10 independent replicates were generated for each model.\u003c/p\u003e\n\u003cp\u003eA process framework of the data flow is shown in Figure 1. Data flowed through each node according to a predetermined schedule.\u003c/p\u003e\n\u003ch3\u003eStatistical Analysis\u003c/h3\u003e\n\u003cp\u003eContinuous variables were expressed as mean\u0026plusmn;standard deviation, and count variables were expressed as frequency. Differences between quantitative data were tested using a t-test and rank test. Hypothesis testing was used to investigate the influence of different data processing methods and algorithms on the model prediction performance. On the analysis results of bootstrapping sampling and validation set, hypothesis testing single factor analysis was performed. The analysis content included different data inputting methods, feature screening methods, and the corresponding mean\u0026plusmn;standard deviation and 95% confidence interval between the three dimensions of the machine learning model and the five evaluation indicators (AUC, Accuracy, Precision, Recall, and F1 Score) and p-value.\u003c/p\u003e\n\u003cp\u003eExcel 2016 was used for summarizing data, and all statistical analyses were performed using Python 3.8.\u003c/p\u003e"},{"header":"RESULT","content":"\u003ch3\u003eBaseline characteristics\u003c/h3\u003e\n\u003cp\u003eThe FBG study cohort included 100,000 patients and the HbA1c study cohort included 2,169 patients. Baseline demographic, clinical, laboratory and medication details are shown in Table 1. The mean ages of the two cohorts were 64.0 \u0026plusmn; 10.1 years and 63.1 \u0026plusmn; 10.2 years, respectively. The most common comorbidities in both cohorts were hypertension, kidney disease, and heart disease. At baseline, FBG was 8.6\u0026plusmn;3.9 mmol/L and 8.9\u0026plusmn;3.8 mmol/L, respectively, and HbA1c were both 7.8\u0026plusmn;2.8%. \u0026nbsp;\u003c/p\u003e\n\u003cp\u003eTable 1 Baseline characteristics of participants\u003c/p\u003e\n\u003ctable border=\"1\" cellpadding=\"0\" cellspacing=\"0\" width=\"100%\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd width=\"62.994350282485875%\"\u003e\n \u003cp\u003ePredictors\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"20.338983050847457%\"\u003e\n \u003cp\u003eFBG（N=100,000）\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"16.666666666666668%\"\u003e\n \u003cp\u003eHbA1c（N=2,169）\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"62.994350282485875%\"\u003e\n \u003cp\u003eCategorical variables,n(%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"20.338983050847457%\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"16.666666666666668%\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"62.994350282485875%\"\u003e\n \u003cp\u003e\u003cem\u003eGender\u0026nbsp;\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"20.338983050847457%\"\u003e\n \u003cp\u003e\u003cem\u003e\u0026nbsp;\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"16.666666666666668%\"\u003e\n \u003cp\u003e\u003cem\u003e\u0026nbsp;\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"62.994350282485875%\"\u003e\n \u003cp\u003e\u003cem\u003eMale\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"20.338983050847457%\"\u003e\n \u003cp\u003e\u003cem\u003e38,205\u003c/em\u003e\u003cem\u003e\u0026nbsp;(\u003c/em\u003e\u003cem\u003e38.2\u003c/em\u003e\u003cem\u003e)\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"16.666666666666668%\"\u003e\n \u003cp\u003e\u003cem\u003e735(33.9)\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"62.994350282485875%\"\u003e\n \u003cp\u003e\u003cem\u003eFemale\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"20.338983050847457%\"\u003e\n \u003cp\u003e\u003cem\u003e61,795 (61.8)\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"16.666666666666668%\"\u003e\n \u003cp\u003e\u003cem\u003e1434(66.1)\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"62.994350282485875%\"\u003e\n \u003cp\u003e\u003cem\u003eEducation\u0026nbsp;\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"20.338983050847457%\"\u003e\n \u003cp\u003e\u003cem\u003e\u0026nbsp;\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"16.666666666666668%\"\u003e\n \u003cp\u003e\u003cem\u003e\u0026nbsp;\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"62.994350282485875%\"\u003e\n \u003cp\u003e\u003cem\u003eCollege and above\u0026nbsp;\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"20.338983050847457%\"\u003e\n \u003cp\u003e\u003cem\u003e71,532\u003c/em\u003e\u003cem\u003e\u0026nbsp;(\u003c/em\u003e\u003cem\u003e71.5\u003c/em\u003e\u003cem\u003e)\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"16.666666666666668%\"\u003e\n \u003cp\u003e\u003cem\u003e1631(75.2)\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"62.994350282485875%\"\u003e\n \u003cp\u003e\u003cem\u003eBelow college and other\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"20.338983050847457%\"\u003e\n \u003cp\u003e\u003cem\u003e28,648 (28.5)\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"16.666666666666668%\"\u003e\n \u003cp\u003e\u003cem\u003e538(24.8)\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"62.994350282485875%\"\u003e\n \u003cp\u003e\u003cem\u003eMarital status\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"20.338983050847457%\"\u003e\n \u003cp\u003e\u003cem\u003e\u0026nbsp;\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"16.666666666666668%\"\u003e\n \u003cp\u003e\u003cem\u003e\u0026nbsp;\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"62.994350282485875%\"\u003e\n \u003cp\u003e\u003cem\u003eMarried/living as married/civil partnershi\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"20.338983050847457%\"\u003e\n \u003cp\u003e\u003cem\u003e84,413\u003c/em\u003e\u003cem\u003e\u0026nbsp;(\u003c/em\u003e\u003cem\u003e84.5\u003c/em\u003e\u003cem\u003e)\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"16.666666666666668%\"\u003e\n \u003cp\u003e\u003cem\u003e1749(80.6)\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"62.994350282485875%\"\u003e\n \u003cp\u003e\u003cem\u003eSingle/never marrie\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"20.338983050847457%\"\u003e\n \u003cp\u003e\u003cem\u003e13,867\u003c/em\u003e\u003cem\u003e\u0026nbsp;(\u003c/em\u003e\u003cem\u003e13.9\u003c/em\u003e\u003cem\u003e)\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"16.666666666666668%\"\u003e\n \u003cp\u003e\u003cem\u003e339(15.6)\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"62.994350282485875%\"\u003e\n \u003cp\u003e\u003cem\u003eWidowed\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"20.338983050847457%\"\u003e\n \u003cp\u003e\u003cem\u003e981\u003c/em\u003e\u003cem\u003e\u0026nbsp;(\u003c/em\u003e\u003cem\u003e1.0\u003c/em\u003e\u003cem\u003e)\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"16.666666666666668%\"\u003e\n \u003cp\u003e\u003cem\u003e28(1.3)\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"62.994350282485875%\"\u003e\n \u003cp\u003e\u003cem\u003eDivorced or separate\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"20.338983050847457%\"\u003e\n \u003cp\u003e\u003cem\u003e687\u003c/em\u003e\u003cem\u003e\u0026nbsp;(\u003c/em\u003e\u003cem\u003e0.6\u003c/em\u003e\u003cem\u003e)\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"16.666666666666668%\"\u003e\n \u003cp\u003e\u003cem\u003e6(0.3)\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"62.994350282485875%\"\u003e\n \u003cp\u003e\u003cem\u003eComplication\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"20.338983050847457%\"\u003e\n \u003cp\u003e\u003cem\u003e\u0026nbsp;\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"16.666666666666668%\"\u003e\n \u003cp\u003e\u003cem\u003e\u0026nbsp;\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"62.994350282485875%\"\u003e\n \u003cp\u003e\u003cem\u003eComplicated with diabetes-related complications\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"20.338983050847457%\"\u003e\n \u003cp\u003e\u003cem\u003e16,016\u003c/em\u003e\u003cem\u003e\u0026nbsp;(\u003c/em\u003e\u003cem\u003e16.0\u003c/em\u003e\u003cem\u003e)\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"16.666666666666668%\"\u003e\n \u003cp\u003e\u003cem\u003e820(37.8)\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"62.994350282485875%\"\u003e\n \u003cp\u003e\u003cem\u003eWithout diabetes-related complications\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"20.338983050847457%\"\u003e\n \u003cp\u003e\u003cem\u003e83,984 (84.0)\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"16.666666666666668%\"\u003e\n \u003cp\u003e\u003cem\u003e1349(62.2)\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" width=\"62.994350282485875%\"\u003e\n \u003cp\u003e\u003cem\u003eComorbidity\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"20.338983050847457%\"\u003e\n \u003cp\u003e\u003cem\u003e\u0026nbsp;\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"16.666666666666668%\"\u003e\n \u003cp\u003e\u003cem\u003e\u0026nbsp;\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" width=\"62.994350282485875%\"\u003e\n \u003cp\u003e\u003cem\u003eCerebrovascular diseas\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"20.338983050847457%\"\u003e\n \u003cp\u003e\u003cem\u003e11,018\u003c/em\u003e\u003cem\u003e\u0026nbsp;(\u003c/em\u003e\u003cem\u003e11.0\u003c/em\u003e\u003cem\u003e)\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"16.666666666666668%\"\u003e\n \u003cp\u003e\u003cem\u003e916(42.2)\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" width=\"62.994350282485875%\"\u003e\n \u003cp\u003e\u003cem\u003eKidney disease\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"20.338983050847457%\"\u003e\n \u003cp\u003e\u003cem\u003e15,752\u003c/em\u003e\u003cem\u003e\u0026nbsp;(\u003c/em\u003e\u003cem\u003e15.8\u003c/em\u003e\u003cem\u003e)\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"16.666666666666668%\"\u003e\n \u003cp\u003e\u003cem\u003e262(12.1)\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" width=\"62.994350282485875%\"\u003e\n \u003cp\u003e\u003cem\u003eHeart disease\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"20.338983050847457%\"\u003e\n \u003cp\u003e\u003cem\u003e13,298\u003c/em\u003e\u003cem\u003e\u0026nbsp;(\u003c/em\u003e\u003cem\u003e13.3\u003c/em\u003e\u003cem\u003e)\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"16.666666666666668%\"\u003e\n \u003cp\u003e\u003cem\u003e889(41.0)\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" width=\"62.994350282485875%\"\u003e\n \u003cp\u003e\u003cem\u003eVascular disease\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"20.338983050847457%\"\u003e\n \u003cp\u003e\u003cem\u003e11,542\u003c/em\u003e\u003cem\u003e\u0026nbsp;(\u003c/em\u003e\u003cem\u003e11.5\u003c/em\u003e\u003cem\u003e)\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"16.666666666666668%\"\u003e\n \u003cp\u003e\u003cem\u003e917(42.3)\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" width=\"62.994350282485875%\"\u003e\n \u003cp\u003e\u003cem\u003eOphthalmological disease\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"20.338983050847457%\"\u003e\n \u003cp\u003e\u003cem\u003e11,155\u003c/em\u003e\u003cem\u003e\u0026nbsp;(\u003c/em\u003e\u003cem\u003e11.2\u003c/em\u003e\u003cem\u003e)\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"16.666666666666668%\"\u003e\n \u003cp\u003e\u003cem\u003e897(41.4)\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" width=\"62.994350282485875%\"\u003e\n \u003cp\u003e\u003cem\u003eHypertensio\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"20.338983050847457%\"\u003e\n \u003cp\u003e\u003cem\u003e54,363\u003c/em\u003e\u003cem\u003e\u0026nbsp;(\u003c/em\u003e\u003cem\u003e54.4\u003c/em\u003e\u003cem\u003e)\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"16.666666666666668%\"\u003e\n \u003cp\u003e\u003cem\u003e1354(62.4)\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" width=\"62.994350282485875%\"\u003e\n \u003cp\u003e\u003cem\u003eDrug\u003c/em\u003e\u003cem\u003e\u0026nbsp;\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"20.338983050847457%\"\u003e\n \u003cp\u003e\u003cem\u003e\u0026nbsp;\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"16.666666666666668%\"\u003e\n \u003cp\u003e\u003cem\u003e\u0026nbsp;\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" width=\"62.994350282485875%\"\u003e\n \u003cp\u003e\u003cem\u003emetformin\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"20.338983050847457%\"\u003e\n \u003cp\u003e\u003cem\u003e16,978\u003c/em\u003e\u003cem\u003e\u0026nbsp;(\u003c/em\u003e\u003cem\u003e17.0\u003c/em\u003e\u003cem\u003e)\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"16.666666666666668%\"\u003e\n \u003cp\u003e\u003cem\u003e-\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" width=\"62.994350282485875%\"\u003e\n \u003cp\u003e\u003cem\u003eGrezit\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"20.338983050847457%\"\u003e\n \u003cp\u003e\u003cem\u003e8,421\u003c/em\u003e\u003cem\u003e\u0026nbsp;(\u003c/em\u003e\u003cem\u003e15.7\u003c/em\u003e\u003cem\u003e)\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"16.666666666666668%\"\u003e\n \u003cp\u003e\u003cem\u003e-\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" width=\"62.994350282485875%\"\u003e\n \u003cp\u003eContinuous variables,\u0026nbsp;mean\u0026nbsp;(SD)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"20.338983050847457%\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"16.666666666666668%\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"62.994350282485875%\"\u003e\n \u003cp\u003e\u003cem\u003eFBG,\u003c/em\u003e\u003cem\u003emmol/L\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"20.338983050847457%\"\u003e\n \u003cp\u003e\u003cem\u003e8.6\u0026nbsp;\u003c/em\u003e\u003cem\u003e(\u003c/em\u003e\u003cem\u003e3.9\u003c/em\u003e\u003cem\u003e)\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"16.666666666666668%\"\u003e\n \u003cp\u003e\u003cem\u003e8.9(3.8)\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"62.994350282485875%\"\u003e\n \u003cp\u003e\u003cem\u003eHbA1c, %\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"20.338983050847457%\"\u003e\n \u003cp\u003e\u003cem\u003e7.8\u003c/em\u003e\u003cem\u003e\u0026nbsp;(\u003c/em\u003e\u003cem\u003e2.9\u003c/em\u003e\u003cem\u003e)\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"16.666666666666668%\"\u003e\n \u003cp\u003e\u003cem\u003e7.8(2.9)\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" width=\"62.994350282485875%\"\u003e\n \u003cp\u003e\u003cem\u003ePulse rate, CPM\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"20.338983050847457%\"\u003e\n \u003cp\u003e\u003cem\u003e75.3\u003c/em\u003e\u003cem\u003e\u0026nbsp;(\u003c/em\u003e\u003cem\u003e10.6\u003c/em\u003e\u003cem\u003e)\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"16.666666666666668%\"\u003e\n \u003cp\u003e\u003cem\u003e76(10.4)\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" width=\"62.994350282485875%\"\u003e\n \u003cp\u003e\u003cem\u003eBMI,kg/m\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"20.338983050847457%\"\u003e\n \u003cp\u003e\u003cem\u003e24.8\u003c/em\u003e\u003cem\u003e\u0026nbsp;(\u003c/em\u003e\u003cem\u003e3.6\u003c/em\u003e\u003cem\u003e)\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"16.666666666666668%\"\u003e\n \u003cp\u003e\u003cem\u003e25.4(3.7)\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" width=\"62.994350282485875%\"\u003e\n \u003cp\u003e\u003cem\u003eWaist circumference, cm\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"20.338983050847457%\"\u003e\n \u003cp\u003e\u003cem\u003e84.3\u003c/em\u003e\u003cem\u003e\u0026nbsp;(\u003c/em\u003e\u003cem\u003e9.2\u003c/em\u003e\u003cem\u003e)\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"16.666666666666668%\"\u003e\n \u003cp\u003e\u003cem\u003e85.6(9.2)\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" width=\"62.994350282485875%\"\u003e\n \u003cp\u003e\u003cem\u003eHemoglobin,g/L\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"20.338983050847457%\"\u003e\n \u003cp\u003e\u003cem\u003e133.9\u003c/em\u003e\u003cem\u003e\u0026nbsp;(\u003c/em\u003e\u003cem\u003e17.9\u003c/em\u003e\u003cem\u003e)\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"16.666666666666668%\"\u003e\n \u003cp\u003e\u003cem\u003e135.1(15.9)\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" width=\"62.994350282485875%\"\u003e\n \u003cp\u003e\u003cem\u003eLeukocyte,\u0026times;10\u003csup\u003e9\u003c/sup\u003e /\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"20.338983050847457%\"\u003e\n \u003cp\u003e\u003cem\u003e6.5\u003c/em\u003e\u003cem\u003e\u0026nbsp;(\u003c/em\u003e\u003cem\u003e3.2\u003c/em\u003e\u003cem\u003e)\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"16.666666666666668%\"\u003e\n \u003cp\u003e\u003cem\u003e6.5(1.8)\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" width=\"62.994350282485875%\"\u003e\n \u003cp\u003e\u003cem\u003ePlatelet,\u0026times;10\u003csup\u003e9\u003c/sup\u003e /L\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"20.338983050847457%\"\u003e\n \u003cp\u003e\u003cem\u003e178.1\u003c/em\u003e\u003cem\u003e\u0026nbsp;(\u003c/em\u003e\u003cem\u003e67.8\u003c/em\u003e\u003cem\u003e)\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"16.666666666666668%\"\u003e\n \u003cp\u003e\u003cem\u003e189.7(63.6)\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" width=\"62.994350282485875%\"\u003e\n \u003cp\u003e\u003cem\u003eALT, U/L\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"20.338983050847457%\"\u003e\n \u003cp\u003e\u003cem\u003e25.3\u003c/em\u003e\u003cem\u003e\u0026nbsp;(\u003c/em\u003e\u003cem\u003e16.2\u003c/em\u003e\u003cem\u003e)\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"16.666666666666668%\"\u003e\n \u003cp\u003e\u003cem\u003e24.7(16.3)\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" width=\"62.994350282485875%\"\u003e\n \u003cp\u003e\u003cem\u003eAST, U/L\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"20.338983050847457%\"\u003e\n \u003cp\u003e\u003cem\u003e24.1\u003c/em\u003e\u003cem\u003e\u0026nbsp;(\u003c/em\u003e\u003cem\u003e14.2\u003c/em\u003e\u003cem\u003e)\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"16.666666666666668%\"\u003e\n \u003cp\u003e\u003cem\u003e20.8(14.2)\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" width=\"62.994350282485875%\"\u003e\n \u003cp\u003e\u003cem\u003eAlbumin,g/L\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"20.338983050847457%\"\u003e\n \u003cp\u003e\u003cem\u003e39.5\u003c/em\u003e\u003cem\u003e\u0026nbsp;(\u003c/em\u003e\u003cem\u003e13.4\u003c/em\u003e\u003cem\u003e)\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"16.666666666666668%\"\u003e\n \u003cp\u003e\u003cem\u003e-\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" width=\"62.994350282485875%\"\u003e\n \u003cp\u003e\u003cem\u003eTotal bilirubin,\u0026mu;mol/L\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"20.338983050847457%\"\u003e\n \u003cp\u003e\u003cem\u003e13.6\u003c/em\u003e\u003cem\u003e\u0026nbsp;(\u003c/em\u003e\u003cem\u003e10.7\u003c/em\u003e\u003cem\u003e)\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"16.666666666666668%\"\u003e\n \u003cp\u003e\u003cem\u003e13.2(6.3)\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" width=\"62.994350282485875%\"\u003e\n \u003cp\u003e\u003cem\u003eConjugated bilirubin,\u0026mu;mol/L\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"20.338983050847457%\"\u003e\n \u003cp\u003e\u003cem\u003e4.5\u003c/em\u003e\u003cem\u003e\u0026nbsp;(\u003c/em\u003e\u003cem\u003e2.9\u003c/em\u003e\u003cem\u003e)\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"16.666666666666668%\"\u003e\n \u003cp\u003e\u003cem\u003e-\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" width=\"62.994350282485875%\"\u003e\n \u003cp\u003e\u003cem\u003eScr,\u0026mu;mol/L\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"20.338983050847457%\"\u003e\n \u003cp\u003e\u003cem\u003e75.6\u003c/em\u003e\u003cem\u003e\u0026nbsp;(\u003c/em\u003e\u003cem\u003e33.6\u003c/em\u003e\u003cem\u003e)\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"16.666666666666668%\"\u003e\n \u003cp\u003e\u003cem\u003e62.8(29.9)\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" width=\"62.994350282485875%\"\u003e\n \u003cp\u003e\u003cem\u003eBlood urea nitrogen,mmol/L\u003c/em\u003e\u003cem\u003e)\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"20.338983050847457%\"\u003e\n \u003cp\u003e\u003cem\u003e5.7\u003c/em\u003e\u003cem\u003e\u0026nbsp;(\u003c/em\u003e\u003cem\u003e2.2\u003c/em\u003e\u003cem\u003e)\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"16.666666666666668%\"\u003e\n \u003cp\u003e\u003cem\u003e5.7(1.9)\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" width=\"62.994350282485875%\"\u003e\n \u003cp\u003e\u003cem\u003eTC,mmol/L\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"20.338983050847457%\"\u003e\n \u003cp\u003e\u003cem\u003e4.7\u003c/em\u003e\u003cem\u003e\u0026nbsp;(\u003c/em\u003e\u003cem\u003e1.5\u003c/em\u003e\u003cem\u003e)\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"16.666666666666668%\"\u003e\n \u003cp\u003e\u003cem\u003e4.8(1.8)\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" width=\"62.994350282485875%\"\u003e\n \u003cp\u003e\u003cem\u003eTG,mmol/L\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"20.338983050847457%\"\u003e\n \u003cp\u003e\u003cem\u003e1.9\u003c/em\u003e\u003cem\u003e\u0026nbsp;(\u003c/em\u003e\u003cem\u003e2.4\u003c/em\u003e\u003cem\u003e)\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"16.666666666666668%\"\u003e\n \u003cp\u003e\u003cem\u003e2.2(2.5)\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" width=\"62.994350282485875%\"\u003e\n \u003cp\u003e\u003cem\u003eDBP,mmHg\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"20.338983050847457%\"\u003e\n \u003cp\u003e\u003cem\u003e80.1\u003c/em\u003e\u003cem\u003e\u0026nbsp;(\u003c/em\u003e\u003cem\u003e9.4\u003c/em\u003e\u003cem\u003e)\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"16.666666666666668%\"\u003e\n \u003cp\u003e\u003cem\u003e80.2(8.9)\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" width=\"62.994350282485875%\"\u003e\n \u003cp\u003e\u003cem\u003eSBP,mmHg\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"20.338983050847457%\"\u003e\n \u003cp\u003e\u003cem\u003e134.9\u003c/em\u003e\u003cem\u003e\u0026nbsp;(\u003c/em\u003e\u003cem\u003e16.4\u003c/em\u003e\u003cem\u003e)\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" width=\"16.666666666666668%\"\u003e\n \u003cp\u003e\u003cem\u003e134.3(15.6)\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"3\" valign=\"top\" width=\"100%\"\u003e\n \u003cp\u003en (%), number of patients and percentage over the total number of patients, mean(SD), the mean and standard deviation of the variable\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e\u003cbr\u003e\u003c/p\u003e\n\u003ch3\u003eVariable and feature screening\u003c/h3\u003e\n\u003cp\u003e426 and 432 variables were removed from the FBG prediction model and the HbA1c prediction model, respectively, during data cleaning, and the specific variables are shown in Supplementary Table 2. Therefore, a total of 85 and 79 variables were finally used for modeling, respectively (Supplementary Table. 3). After inputting data with missing variables, the positive and negative samples are relatively balanced, so no sampling is required. The results of the feature screening are shown in Table 2. The results showed that the most important indicators of the FBG and HbA1c prediction model were the FBG value and HbA1c. The patient\u0026apos;s medication compliance, follow-up, dietary habits, BMI, and waist circumference also had a greater impact on the FBG level. In addition, the feature selection results also showed that patients with hypertension or other comorbid diseases, laboratory indicators of related diseases such as Scr, serum alanine aminotransferase, blood urea nitrogen, platelets, and white blood cells, as well as age, smoking, all had a certain degree of influence on FBG. For the HbA1c prediction model, the feature screening results showed that laboratory indicators such as platelets, Scr, AST, Hb, AST, etc. accounted for a large proportion. According to the results of Not inputting and Lasso screening, the feature importance of the data is drawn（Figure 2）, and the feature importance bar chart drawn by other inputting and screening methods is shown in supplementary figures 2-7.\u003c/p\u003e\n\u003cp\u003eTable 2 Feature screening results\u003c/p\u003e\n\u003ctable border=\"1\" cellpadding=\"0\" cellspacing=\"0\" width=\"99%\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd rowspan=\"2\" width=\"21.21212121212121%\"\u003e\n \u003cp\u003eInputting\u003c/p\u003e\n \u003cp\u003emethod\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" width=\"13.131313131313131%\"\u003e\n \u003cp\u003eScreening\u003c/p\u003e\n \u003cp\u003emethod\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" width=\"65.65656565656566%\"\u003e\n \u003cp\u003eTop 10 Variables\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 39.824%;\" width=\"58.46153846153846%\"\u003e\n \u003cp\u003eFBG\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 27.6002%;\" width=\"41.53846153846154%\"\u003e\n \u003cp\u003eHbA1c\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"21.21212121212121%\"\u003e\n \u003cp\u003eNot\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"13.131313131313131%\"\u003e\n \u003cp\u003eLasso\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 39.824%;\" width=\"38.38383838383838%\"\u003e\n \u003cp\u003eAdverse reactions、FBG、Satisfaction with follow-up、Medical Compliance - Good、diet control、Concomitant disease、Medication Adherence - Not Taking Medication、Medical Compliance\u0026mdash;general、T2DM、hypertension\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 27.6002%;\" valign=\"top\" width=\"27.272727272727273%\"\u003e\n \u003cp\u003eFBG、Platelets、Satisfaction with follow-up、Adverse reactions、Scr、BMI、SBP、Age、Pulse rate、ALT、AST\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"21.21212121212121%\"\u003e\n \u003cp\u003eNot\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"13.131313131313131%\"\u003e\n \u003cp\u003eBoruta\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 39.824%;\" width=\"38.38383838383838%\"\u003e\n \u003cp\u003eFBG、platelets、Satisfaction with follow-up、Adverse reactions、Scr、BMI、high blood pressure、age、Pulse rate、ALT\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 27.6002%;\" valign=\"top\" width=\"27.272727272727273%\"\u003e\n \u003cp\u003eFBG、pulse rate、Scr、HBP、BMI、SBP、Waist circumference、AST、Age、daily staple food\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"21.21212121212121%\"\u003e\n \u003cp\u003eModified random\u003c/p\u003e\n \u003cp\u003eforest\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"13.131313131313131%\"\u003e\n \u003cp\u003eLasso\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 39.824%;\" width=\"38.38383838383838%\"\u003e\n \u003cp\u003eHbA1c、Adverse reactions、Satisfaction with follow-up、Medical Compliance - Good、FBG、diet control、Medication Adherence - Not Taking Medication、Concomitant disease、Outpatient follow-up、kidney disease\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 27.6002%;\" valign=\"top\" width=\"27.272727272727273%\"\u003e\n \u003cp\u003eHbA1c、Age、Medication Adherence - Not Taking、Adverse reactions\u0026nbsp;、Symmetry palpation of dorsalis pedis、pulse rate、SBP、FBG、Current the length of each exercise、BMI\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"21.21212121212121%\"\u003e\n \u003cp\u003eModified random\u003c/p\u003e\n \u003cp\u003eforest\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"13.131313131313131%\"\u003e\n \u003cp\u003eBoruta\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 39.824%;\" width=\"38.38383838383838%\"\u003e\n \u003cp\u003eplatelets、Adverse reactions、FBG、BMI、blood urea nitrogen、HbA1c、leukocyte、Satisfaction with follow-up、high blood pressure、Scr\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 27.6002%;\" valign=\"top\" width=\"27.272727272727273%\"\u003e\n \u003cp\u003eAge、Hb、HBP、BUN、Scr、AST、HbA1c、PLT、FBG、BMI\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003ch3\u003eModel performance\u003c/h3\u003e\n\u003cp\u003eThe machine learning results are shown in Table 3. The inputting methods used by all best models were all modified random forest inputting: The optimal feature screening method was Not screening and Boruta screening The optimal machine learning methods were ensemble learning and XGBoost. The AUC value of the best model for FBG was model 1 (AUC=0.8190), the worse one was model 5 (AUC=0.8082). The AUC value of the best model for HbA1c was model 1 (AUC=0.9704), the worse one was model 5 (AUC=0.9674). The AUC values of the ten best models were all greater than 0.75, indicating that the prediction model had good prediction performance, and had the possibility of certain clinical application. The ROC curves and P-R curves of the five best models are shown in Figures 3 and 4.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eTabel 3\u0026nbsp;Predictive model building results\u003c/p\u003e\n\u003ctable border=\"1\" cellpadding=\"0\" cellspacing=\"0\" width=\"100%\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" width=\"8.421052631578947%\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"10.526315789473685%\"\u003e\n \u003cp\u003eModel ID\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"7.368421052631579%\"\u003e\n \u003cp\u003eAUC\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"9.473684210526315%\"\u003e\n \u003cp\u003eAccuracy\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"9.473684210526315%\"\u003e\n \u003cp\u003ePrecision\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"8.421052631578947%\"\u003e\n \u003cp\u003eRecall\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"8.421052631578947%\"\u003e\n \u003cp\u003eF1Score\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"16.842105263157894%\"\u003e\n \u003cp\u003eImputing\u0026nbsp;methods\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"10.526315789473685%\"\u003e\n \u003cp\u003eScreening\u003c/p\u003e\n \u003cp\u003emethods\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"10.526315789473685%\"\u003e\n \u003cp\u003eModels\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd rowspan=\"5\" width=\"8.421052631578947%\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003cp\u003e\u003cstrong\u003eFBG\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"10.526315789473685%\"\u003e\n \u003cp\u003emodel 1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"7.368421052631579%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.819\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"9.473684210526315%\"\u003e\n \u003cp\u003e0.7439\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"9.473684210526315%\"\u003e\n \u003cp\u003e0.7733\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"8.421052631578947%\"\u003e\n \u003cp\u003e0.6901\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"8.421052631578947%\"\u003e\n \u003cp\u003e0.7293\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"16.842105263157894%\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003cp\u003eModified\u0026nbsp;random\u003c/p\u003e\n \u003cp\u003eforest inputting\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"10.526315789473685%\"\u003e\n \u003cp\u003eNot\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"10.526315789473685%\"\u003e\n \u003cp\u003eEnsemble learning\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"11.494252873563218%\"\u003e\n \u003cp\u003emodel 2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"8.045977011494253%\"\u003e\n \u003cp\u003e0.8163\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"10.344827586206897%\"\u003e\n \u003cp\u003e0.7423\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"10.344827586206897%\"\u003e\n \u003cp\u003e0.7674\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"9.195402298850574%\"\u003e\n \u003cp\u003e0.6955\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"9.195402298850574%\"\u003e\n \u003cp\u003e0.7297\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.39080459770115%\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003cp\u003eModified random\u003c/p\u003e\n \u003cp\u003eforest inputting\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.494252873563218%\"\u003e\n \u003cp\u003eNot\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.494252873563218%\"\u003e\n \u003cp\u003eXGBoost\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"11.494252873563218%\"\u003e\n \u003cp\u003emodel 3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"8.045977011494253%\"\u003e\n \u003cp\u003e0.8119\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"10.344827586206897%\"\u003e\n \u003cp\u003e0.7415\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"10.344827586206897%\"\u003e\n \u003cp\u003e0.7692\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"9.195402298850574%\"\u003e\n \u003cp\u003e0.69\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"9.195402298850574%\"\u003e\n \u003cp\u003e0.7275\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.39080459770115%\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003cp\u003eModified\u0026nbsp;random\u003c/p\u003e\n \u003cp\u003eforest\u0026nbsp;inputting\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.494252873563218%\"\u003e\n \u003cp\u003eBoruta\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.494252873563218%\"\u003e\n \u003cp\u003eEnsemble learning\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"11.494252873563218%\"\u003e\n \u003cp\u003emodel 4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"8.045977011494253%\"\u003e\n \u003cp\u003e0.8087\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"10.344827586206897%\"\u003e\n \u003cp\u003e0.7404\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"10.344827586206897%\"\u003e\n \u003cp\u003e0.769\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"9.195402298850574%\"\u003e\n \u003cp\u003e0.6872\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"9.195402298850574%\"\u003e\n \u003cp\u003e0.7258\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.39080459770115%\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003cp\u003eModified\u0026nbsp;random\u003c/p\u003e\n \u003cp\u003eforest inputting\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.494252873563218%\"\u003e\n \u003cp\u003eLasso\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.494252873563218%\"\u003e\n \u003cp\u003eEnsemble learning\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"11.494252873563218%\"\u003e\n \u003cp\u003emodel 5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"8.045977011494253%\"\u003e\n \u003cp\u003e0.8082\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"10.344827586206897%\"\u003e\n \u003cp\u003e0.7388\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"10.344827586206897%\"\u003e\n \u003cp\u003e0.7629\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"9.195402298850574%\"\u003e\n \u003cp\u003e0.6929\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"9.195402298850574%\"\u003e\n \u003cp\u003e0.7262\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.39080459770115%\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003cp\u003eModified\u0026nbsp;random\u003c/p\u003e\n \u003cp\u003eforest inputting\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.494252873563218%\"\u003e\n \u003cp\u003eBoruta\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.494252873563218%\"\u003e\n \u003cp\u003eXGBoost\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd rowspan=\"5\" width=\"8.421052631578947%\"\u003e\n \u003cp\u003e\u003cstrong\u003eHbA1c\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"10.526315789473685%\"\u003e\n \u003cp\u003emodel\u0026nbsp;1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"7.368421052631579%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.9704\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"9.473684210526315%\"\u003e\n \u003cp\u003e0.9217\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"9.473684210526315%\"\u003e\n \u003cp\u003e0.894\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"8.421052631578947%\"\u003e\n \u003cp\u003e0.9463\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"8.421052631578947%\"\u003e\n \u003cp\u003e0.9194\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"16.842105263157894%\"\u003e\n \u003cp\u003eModified\u0026nbsp;random\u003c/p\u003e\n \u003cp\u003eforest inputting\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"10.526315789473685%\"\u003e\n \u003cp\u003eBoruta\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"10.526315789473685%\"\u003e\n \u003cp\u003eEnsemble learning\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"11.494252873563218%\"\u003e\n \u003cp\u003emodel\u0026nbsp;2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"8.045977011494253%\"\u003e\n \u003cp\u003e0.9702\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"10.344827586206897%\"\u003e\n \u003cp\u003e0.924\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"10.344827586206897%\"\u003e\n \u003cp\u003e0.9135\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"9.195402298850574%\"\u003e\n \u003cp\u003e0.9268\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"9.195402298850574%\"\u003e\n \u003cp\u003e0.9201\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.39080459770115%\"\u003e\n \u003cp\u003eModified\u0026nbsp;random\u003c/p\u003e\n \u003cp\u003eforest inputting\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.494252873563218%\"\u003e\n \u003cp\u003eNot\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.494252873563218%\"\u003e\n \u003cp\u003eEnsemble learning\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"11.494252873563218%\"\u003e\n \u003cp\u003emodel\u0026nbsp;3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"8.045977011494253%\"\u003e\n \u003cp\u003e0.9697\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"10.344827586206897%\"\u003e\n \u003cp\u003e0.924\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"10.344827586206897%\"\u003e\n \u003cp\u003e0.9095\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"9.195402298850574%\"\u003e\n \u003cp\u003e0.9317\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"9.195402298850574%\"\u003e\n \u003cp\u003e0.9205\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.39080459770115%\"\u003e\n \u003cp\u003eModified\u0026nbsp;random\u003c/p\u003e\n \u003cp\u003eforest inputting\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.494252873563218%\"\u003e\n \u003cp\u003eLasso\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.494252873563218%\"\u003e\n \u003cp\u003eEnsemble learning\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"11.494252873563218%\"\u003e\n \u003cp\u003emodel\u0026nbsp;4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"8.045977011494253%\"\u003e\n \u003cp\u003e0.9688\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"10.344827586206897%\"\u003e\n \u003cp\u003e0.9263\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"10.344827586206897%\"\u003e\n \u003cp\u003e0.9179\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"9.195402298850574%\"\u003e\n \u003cp\u003e0.9268\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"9.195402298850574%\"\u003e\n \u003cp\u003e0.9223\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.39080459770115%\"\u003e\n \u003cp\u003eModified\u0026nbsp;random\u003c/p\u003e\n \u003cp\u003eforest inputting\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.494252873563218%\"\u003e\n \u003cp\u003eLasso\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.494252873563218%\"\u003e\n \u003cp\u003eXGBoost\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"11.494252873563218%\"\u003e\n \u003cp\u003emodel\u0026nbsp;5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"8.045977011494253%\"\u003e\n \u003cp\u003e0.9674\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"10.344827586206897%\"\u003e\n \u003cp\u003e0.9171\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"10.344827586206897%\"\u003e\n \u003cp\u003e0.9043\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"9.195402298850574%\"\u003e\n \u003cp\u003e0.922\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"9.195402298850574%\"\u003e\n \u003cp\u003e0.913\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.39080459770115%\"\u003e\n \u003cp\u003eModified\u0026nbsp;random\u003c/p\u003e\n \u003cp\u003eforest inputting\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.494252873563218%\"\u003e\n \u003cp\u003eNot\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.494252873563218%\"\u003e\n \u003cp\u003eXGBoost\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e\u0026nbsp;\u003c/p\u003e\n\u003ch3\u003eThe effect of different data processing methods on the results\u003c/h3\u003e\n\u003cp\u003eThe results of the ten-fold cross-validation analysis of the data imputing method for FBG are shown in Supplemental Table 4. The modified random forest imputing had the greater impact on the model effect, the AUC values for the FBG and HbA1c models were 0.749\u0026plusmn;0.044(95%CI=0.745-0.753) and 0.901\u0026plusmn;0.078\u0026nbsp;(95%CI=0.894-0.907). The results of the ten-fold cross-validation analysis of the feature screening of the validation set showed that (Supplemental Table 5), Lasso screening had the greatest impact on the model effect, the AUC values for the FBG, and HbA1c models were 0.728\u0026plusmn;0.038(95%CI=0.725-0.731) and 0.776\u0026plusmn;0.130(95%CI=0.766-0.785).\u003c/p\u003e\n\u003cp\u003eThe analysis results of the data imputing method in the bootstrapping sampling showed that the modified random forest imputing had a greater impact on the model effect（Supplemental Table 6）, the AUC values for the FBG, and HbA1c models were 0.754\u0026plusmn;0.048(95%CI=0.754-0.755) and 0.902\u0026plusmn;0.083(95%CI=0.902-0.903). The results of feature screening analysis in bootstrapping sampling showed that Boruta screening had the greatest impact on the FBG model effect, with an AUC value of 0.732\u0026plusmn;0.040(95%CI=0.731-0.732). For HbA1c, the greatest impact analysis screening method in bootstrapping sampling was Lasso screening, with an AUC value of 0.772\u0026plusmn;0.131(95%CI=0.772-0.773) ( Supplemental Table 7 ).According to the results of the test set and the validation set, the prediction ability of the model using the modified random forest imputing method was better. The ten prediction models established in the best model all adopted three feature screening methods. Overall, the results of this study can provide a good methodological reference for the establishment of subsequent T2DM prediction models.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eThe effect of different algorithms on the results\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eHypothesis testing was used to examine the impact of different algorithms on the model\u0026apos;s predictive performance. The ten-fold cross-validation results for FBG and HbA1c prediction model are shown in Table 4. The results showed that XGBoost had the greatest impact on the model effect, the AUC values for the FBG and HbA1c models were 0.761\u0026plusmn;0.029 (95%CI=0.756-0.766) and 0.802\u0026plusmn;0.159 (95%CI=0.773-0.831). Passive Aggressive had the least impact on the FBG model effect, with an AUC value of 0.610 \u0026plusmn;0.052 (95%CI=0.600-0.619) (P\u0026lt;0.0001), and Multinomial Naive Bayes had the least impact on the HbA1c model effect, with an AUC value of 0.662\u0026plusmn;0.076(95%CI=0.648-0.676). The results of bootstrapping sampling analysis (Supplemental Table 8) showed that among the 16 machine learning models used in this section, ensemble learning had the greatest impact on the model effect, the AUC values of FBG, and HbA1c are respectively 0.767\u0026plusmn;0.029 and 0.843\u0026plusmn;0.118.\u003c/p\u003e\n\u003cp\u003eTable 4 Machine learning algorithm ten-fold cross-validation analysis results\u003c/p\u003e\n\u003ctable border=\"1\" cellpadding=\"0\" cellspacing=\"0\" width=\"100%\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd rowspan=\"2\" style=\"width: 8.7749%;\" width=\"17.52577319587629%\"\u003e\n \u003cp\u003e\u003cstrong\u003eFBG\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" style=\"width: 6.2481%;\" width=\"16.49484536082474%\"\u003e\n \u003cp\u003e\u003cstrong\u003eAUC\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" style=\"width: 8.683%;\" width=\"16.49484536082474%\"\u003e\n \u003cp\u003e\u003cstrong\u003eAccuracy\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" style=\"width: 6.2481%;\" width=\"16.49484536082474%\"\u003e\n \u003cp\u003e\u003cstrong\u003ePrecision\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" style=\"width: 6.2481%;\" width=\"15.463917525773196%\"\u003e\n \u003cp\u003e\u003cstrong\u003eRecall\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" style=\"width: 6.8178%;\" width=\"17.52577319587629%\"\u003e\n \u003cp\u003e\u003cstrong\u003eF1Score\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"9.090909090909092%\"\u003e\n \u003cp\u003e\u003cstrong\u003eMean\u0026plusmn;SD\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"10.38961038961039%\"\u003e\n \u003cp\u003e\u003cstrong\u003e95%CI\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.2481%;\" width=\"10.38961038961039%\"\u003e\n \u003cp\u003e\u003cstrong\u003eMean\u0026plusmn;SD\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"10.38961038961039%\"\u003e\n \u003cp\u003e\u003cstrong\u003e95%CI\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"9.090909090909092%\"\u003e\n \u003cp\u003e\u003cstrong\u003eMean\u0026plusmn;SD\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"10.38961038961039%\"\u003e\n \u003cp\u003e\u003cstrong\u003e95%CI\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"9.090909090909092%\"\u003e\n \u003cp\u003e\u003cstrong\u003eMean\u0026plusmn;SD\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"9.090909090909092%\"\u003e\n \u003cp\u003e\u003cstrong\u003e95%CI\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"10.38961038961039%\"\u003e\n \u003cp\u003e\u003cstrong\u003eMean\u0026plusmn;SD\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.9587%;\" width=\"11.688311688311689%\"\u003e\n \u003cp\u003e\u003cstrong\u003e95%CI\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 8.7749%;\" width=\"18.085106382978722%\"\u003e\n \u003cp\u003e\u003cstrong\u003eAdaBoost\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"7.446808510638298%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.741\u0026plusmn;0.015\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"8.51063829787234%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.738-0.744\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.2481%;\" width=\"8.51063829787234%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.682\u0026plusmn;0.012\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"8.51063829787234%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.680-0.684\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"7.446808510638298%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.679\u0026plusmn;0.019\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"8.51063829787234%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.675-0.682\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"7.446808510638298%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.677\u0026plusmn;0.020\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"7.446808510638298%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.674-0.681\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"8.51063829787234%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.678\u0026plusmn;0.016\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.9587%;\" width=\"9.574468085106384%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.675-0.681\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 8.7749%;\" width=\"18.085106382978722%\"\u003e\n \u003cp\u003e\u003cstrong\u003eBagging\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"7.446808510638298%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.747\u0026plusmn;0.033\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"8.51063829787234%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.741-0.753\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.2481%;\" width=\"8.51063829787234%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.690\u0026plusmn;0.026\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"8.51063829787234%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.685-0.695\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"7.446808510638298%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.694\u0026plusmn;0.038\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"8.51063829787234%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.687-0.701\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"7.446808510638298%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.668\u0026plusmn;0.026\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"7.446808510638298%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.664-0.673\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"8.51063829787234%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.681\u0026plusmn;0.029\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.9587%;\" width=\"9.574468085106384%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.675-0.686\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 8.7749%;\" width=\"18.085106382978722%\"\u003e\n \u003cp\u003e\u003cstrong\u003eBernoulli_Naive_Bayes\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"7.446808510638298%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.722\u0026plusmn;0.010\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"8.51063829787234%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.721-0.724\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.2481%;\" width=\"8.51063829787234%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.673\u0026plusmn;0.008\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"8.51063829787234%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.672-0.675\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"7.446808510638298%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.663\u0026plusmn;0.009\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"8.51063829787234%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.661-0.665\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"7.446808510638298%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.688\u0026plusmn;0.022\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"7.446808510638298%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.684-0.692\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"8.51063829787234%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.675\u0026plusmn;0.015\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.9587%;\" width=\"9.574468085106384%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.673-0.678\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 8.7749%;\" width=\"18.085106382978722%\"\u003e\n \u003cp\u003e\u003cstrong\u003eDecision_Tree\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"7.446808510638298%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.738\u0026plusmn;0.019\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"8.51063829787234%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.735-0.742\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.2481%;\" width=\"8.51063829787234%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.685\u0026plusmn;0.015\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"8.51063829787234%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.683-0.688\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"7.446808510638298%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.689\u0026plusmn;0.037\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"8.51063829787234%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.682-0.695\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"7.446808510638298%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.671\u0026plusmn;0.038\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"7.446808510638298%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.664-0.678\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"8.51063829787234%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.678\u0026plusmn;0.015\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.9587%;\" width=\"9.574468085106384%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.675-0.681\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 8.7749%;\" width=\"18.085106382978722%\"\u003e\n \u003cp\u003e\u003cstrong\u003eExtra_Tree\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"7.446808510638298%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.724\u0026plusmn;0.017\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"8.51063829787234%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.721-0.727\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.2481%;\" width=\"8.51063829787234%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.677\u0026plusmn;0.013\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"8.51063829787234%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.674-0.679\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"7.446808510638298%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.677\u0026plusmn;0.025\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"8.51063829787234%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.673-0.682\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"7.446808510638298%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.663\u0026plusmn;0.028\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"7.446808510638298%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.658-0.668\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"8.51063829787234%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.669\u0026plusmn;0.016\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.9587%;\" width=\"9.574468085106384%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.667-0.672\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 8.7749%;\" width=\"18.085106382978722%\"\u003e\n \u003cp\u003e\u003cstrong\u003eGaussian_Naive_Bayes\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"7.446808510638298%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.696\u0026plusmn;0.014\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"8.51063829787234%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.693-0.698\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.2481%;\" width=\"8.51063829787234%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.647\u0026plusmn;0.012\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"8.51063829787234%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.645-0.649\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"7.446808510638298%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.630\u0026plusmn;0.010\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"8.51063829787234%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.628-0.631\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"7.446808510638298%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.694\u0026plusmn;0.030\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"7.446808510638298%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.689-0.700\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"8.51063829787234%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.660\u0026plusmn;0.017\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.9587%;\" width=\"9.574468085106384%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.657-0.663\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 8.7749%;\" width=\"18.085106382978722%\"\u003e\n \u003cp\u003e\u003cstrong\u003eGradient_Boosting\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"7.446808510638298%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.750\u0026plusmn;0.018\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"8.51063829787234%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.747-0.753\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.2481%;\" width=\"8.51063829787234%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.690\u0026plusmn;0.015\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"8.51063829787234%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.687-0.692\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"7.446808510638298%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.695\u0026plusmn;0.029\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"8.51063829787234%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.689-0.700\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"7.446808510638298%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.666\u0026plusmn;0.017\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"7.446808510638298%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.663-0.669\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"8.51063829787234%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.680\u0026plusmn;0.015\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.9587%;\" width=\"9.574468085106384%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.677-0.682\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 8.7749%;\" width=\"18.085106382978722%\"\u003e\n \u003cp\u003e\u003cstrong\u003eLDA\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"7.446808510638298%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.737\u0026plusmn;0.012\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"8.51063829787234%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.734-0.739\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.2481%;\" width=\"8.51063829787234%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.678\u0026plusmn;0.007\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"8.51063829787234%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.676-0.679\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"7.446808510638298%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.668\u0026plusmn;0.007\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"8.51063829787234%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.666-0.669\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"7.446808510638298%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.692\u0026plusmn;0.028\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"7.446808510638298%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.687-0.697\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"8.51063829787234%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.679\u0026plusmn;0.016\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.9587%;\" width=\"9.574468085106384%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.676-0.682\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 8.7749%;\" width=\"18.085106382978722%\"\u003e\n \u003cp\u003e\u003cstrong\u003eLogistic_Regression\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"7.446808510638298%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.737\u0026plusmn;0.012\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"8.51063829787234%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.735-0.740\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.2481%;\" width=\"8.51063829787234%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.679\u0026plusmn;0.008\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"8.51063829787234%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.677-0.680\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"7.446808510638298%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.670\u0026plusmn;0.008\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"8.51063829787234%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.669-0.672\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"7.446808510638298%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.688\u0026plusmn;0.029\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"7.446808510638298%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.683-0.693\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"8.51063829787234%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.679\u0026plusmn;0.017\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.9587%;\" width=\"9.574468085106384%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.676-0.682\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 8.7749%;\" width=\"18.085106382978722%\"\u003e\n \u003cp\u003e\u003cstrong\u003eMultinomial_Naive_Bayes\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"7.446808510638298%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.707\u0026plusmn;0.009\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"8.51063829787234%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.705-0.708\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.2481%;\" width=\"8.51063829787234%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.666\u0026plusmn;0.006\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"8.51063829787234%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.665-0.667\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"7.446808510638298%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.657\u0026plusmn;0.006\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"8.51063829787234%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.656-0.658\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"7.446808510638298%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.678\u0026plusmn;0.035\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"7.446808510638298%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.672-0.684\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"8.51063829787234%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.667\u0026plusmn;0.017\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.9587%;\" width=\"9.574468085106384%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.664-0.670\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 8.7749%;\" width=\"18.085106382978722%\"\u003e\n \u003cp\u003e\u003cstrong\u003ePassive_Aggressive\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"7.446808510638298%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.610\u0026plusmn;0.052\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"8.51063829787234%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.600-0.619\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.2481%;\" width=\"8.51063829787234%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.581\u0026plusmn;0.040\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"8.51063829787234%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.574-0.588\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"7.446808510638298%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.575\u0026plusmn;0.044\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"8.51063829787234%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.567-0.583\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"7.446808510638298%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.579\u0026plusmn;0.076\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"7.446808510638298%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.565-0.593\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"8.51063829787234%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.575\u0026plusmn;0.054\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.9587%;\" width=\"9.574468085106384%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.565-0.585\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 8.7749%;\" width=\"18.085106382978722%\"\u003e\n \u003cp\u003e\u003cstrong\u003eQDA\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"7.446808510638298%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.723\u0026plusmn;0.010\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"8.51063829787234%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.721-0.725\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.2481%;\" width=\"8.51063829787234%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.670\u0026plusmn;0.008\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"8.51063829787234%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.669-0.672\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"7.446808510638298%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.660\u0026plusmn;0.008\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"8.51063829787234%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.659-0.662\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"7.446808510638298%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.686\u0026plusmn;0.023\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"7.446808510638298%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.682-0.690\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"8.51063829787234%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.673\u0026plusmn;0.015\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.9587%;\" width=\"9.574468085106384%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.670-0.675\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 8.7749%;\" width=\"18.085106382978722%\"\u003e\n \u003cp\u003e\u003cstrong\u003eRandom_Forest\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"7.446808510638298%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.756\u0026plusmn;0.027\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"8.51063829787234%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.751-0.761\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.2481%;\" width=\"8.51063829787234%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.696\u0026plusmn;0.022\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"8.51063829787234%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.692-0.700\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"7.446808510638298%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.702\u0026plusmn;0.033\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"8.51063829787234%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.696-0.708\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"7.446808510638298%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.669\u0026plusmn;0.025\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"7.446808510638298%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.664-0.674\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"8.51063829787234%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.685\u0026plusmn;0.025\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.9587%;\" width=\"9.574468085106384%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.680-0.690\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 8.7749%;\" width=\"18.085106382978722%\"\u003e\n \u003cp\u003e\u003cstrong\u003eSGD\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"7.446808510638298%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.736\u0026plusmn;0.012\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"8.51063829787234%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.734-0.739\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.2481%;\" width=\"8.51063829787234%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.670\u0026plusmn;0.005\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"8.51063829787234%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.669-0.670\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"7.446808510638298%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.652\u0026plusmn;0.004\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"8.51063829787234%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.651-0.652\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"7.446808510638298%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.712\u0026plusmn;0.034\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"7.446808510638298%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.706-0.718\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"8.51063829787234%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.680\u0026plusmn;0.016\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.9587%;\" width=\"9.574468085106384%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.677-0.683\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 8.7749%;\" width=\"18.085106382978722%\"\u003e\n \u003cp\u003e\u003cstrong\u003eXGBoost\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"7.446808510638298%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.761\u0026plusmn;0.029\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"8.51063829787234%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.756-0.766\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.2481%;\" width=\"8.51063829787234%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.698\u0026plusmn;0.023\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"8.51063829787234%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.694-0.702\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"7.446808510638298%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.702\u0026plusmn;0.035\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"8.51063829787234%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.696-0.708\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"7.446808510638298%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.679\u0026plusmn;0.021\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"7.446808510638298%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.675-0.683\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"8.51063829787234%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.690\u0026plusmn;0.024\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.9587%;\" width=\"9.574468085106384%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.686-0.694\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 8.7749%;\" width=\"17.52577319587629%\"\u003e\n \u003cp\u003e\u003cstrong\u003eP value\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" style=\"width: 6.2481%;\" width=\"16.49484536082474%\"\u003e\n \u003cp\u003e\u003cstrong\u003eP\u0026lt;0.0001\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" style=\"width: 8.683%;\" width=\"16.49484536082474%\"\u003e\n \u003cp\u003e\u003cstrong\u003eP\u0026lt;0.0001\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" style=\"width: 6.2481%;\" width=\"16.49484536082474%\"\u003e\n \u003cp\u003e\u003cstrong\u003eP\u0026lt;0.0001\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" style=\"width: 6.2481%;\" width=\"15.463917525773196%\"\u003e\n \u003cp\u003e\u003cstrong\u003eP\u0026lt;0.0001\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" style=\"width: 6.8178%;\" width=\"17.52577319587629%\"\u003e\n \u003cp\u003e\u003cstrong\u003eP\u0026lt;0.0001\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd rowspan=\"2\" style=\"width: 8.7749%;\" width=\"17.346938775510203%\"\u003e\n \u003cp\u003e\u003cstrong\u003eHbA1c\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" style=\"width: 8.683%;\" width=\"16.3265306122449%\"\u003e\n \u003cp\u003e\u003cstrong\u003eAUC\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" style=\"width: 6.2481%;\" width=\"15.306122448979592%\"\u003e\n \u003cp\u003e\u003cstrong\u003eAccuracy\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" style=\"width: 6.2481%;\" width=\"16.3265306122449%\"\u003e\n \u003cp\u003e\u003cstrong\u003ePrecision\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" style=\"width: 10.1072%;\" width=\"17.346938775510203%\"\u003e\n \u003cp\u003e\u003cstrong\u003eRecall\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" style=\"width: 6.3859%;\" width=\"17.346938775510203%\"\u003e\n \u003cp\u003e\u003cstrong\u003eF1Score\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 6.2481%;\" width=\"10.256410256410257%\"\u003e\n \u003cp\u003e\u003cstrong\u003eMean\u0026plusmn;SD\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"8.974358974358974%\"\u003e\n \u003cp\u003e\u003cstrong\u003e95%CI\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"8.974358974358974%\"\u003e\n \u003cp\u003e\u003cstrong\u003eMean\u0026plusmn;SD\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"8.974358974358974%\"\u003e\n \u003cp\u003e\u003cstrong\u003e95%CI\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"10.256410256410257%\"\u003e\n \u003cp\u003e\u003cstrong\u003eMean\u0026plusmn;SD\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"10.256410256410257%\"\u003e\n \u003cp\u003e\u003cstrong\u003e95%CI\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"10.256410256410257%\"\u003e\n \u003cp\u003e\u003cstrong\u003eMean\u0026plusmn;SD\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.2481%;\" width=\"10.256410256410257%\"\u003e\n \u003cp\u003e\u003cstrong\u003e95%CI\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"11.538461538461538%\"\u003e\n \u003cp\u003e\u003cstrong\u003eMean\u0026plusmn;SD\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.5268%;\" width=\"10.256410256410257%\"\u003e\n \u003cp\u003e\u003cstrong\u003e95%CI\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 8.7749%;\" width=\"17.894736842105264%\"\u003e\n \u003cp\u003e\u003cstrong\u003eAdaBoost\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.2481%;\" width=\"8.421052631578947%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.782\u0026plusmn;0.141\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"7.368421052631579%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.756-0.807\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"7.368421052631579%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.737\u0026plusmn;0.120\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"7.368421052631579%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.716-0.759\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"8.421052631578947%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.762\u0026plusmn;0.130\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"8.421052631578947%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.738-0.786\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"8.421052631578947%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.737\u0026plusmn;0.150\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.2481%;\" width=\"8.421052631578947%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.710-0.764\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"9.473684210526315%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.739\u0026plusmn;0.111\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.5268%;\" width=\"8.421052631578947%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.719-0.759\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 8.7749%;\" width=\"17.894736842105264%\"\u003e\n \u003cp\u003e\u003cstrong\u003eBagging\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.2481%;\" width=\"8.421052631578947%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.796\u0026plusmn;0.161\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"7.368421052631579%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.767-0.825\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"7.368421052631579%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.747\u0026plusmn;0.143\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"7.368421052631579%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.721-0.772\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"8.421052631578947%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.763\u0026plusmn;0.131\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"8.421052631578947%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.739-0.786\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"8.421052631578947%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.741\u0026plusmn;0.145\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.2481%;\" width=\"8.421052631578947%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.715-0.767\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"9.473684210526315%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.749\u0026plusmn;0.133\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.5268%;\" width=\"8.421052631578947%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.725-0.773\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 8.7749%;\" width=\"17.894736842105264%\"\u003e\n \u003cp\u003e\u003cstrong\u003eBernoulli_Naive_Bayes\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.2481%;\" width=\"8.421052631578947%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.727\u0026plusmn;0.103\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"7.368421052631579%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.708-0.745\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"7.368421052631579%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.682\u0026plusmn;0.101\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"7.368421052631579%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.664-0.701\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"8.421052631578947%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.683\u0026plusmn;0.093\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"8.421052631578947%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.666-0.700\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"8.421052631578947%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.692\u0026plusmn;0.123\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.2481%;\" width=\"8.421052631578947%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.670-0.715\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"9.473684210526315%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.685\u0026plusmn;0.100\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.5268%;\" width=\"8.421052631578947%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.667-0.703\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 8.7749%;\" width=\"17.894736842105264%\"\u003e\n \u003cp\u003e\u003cstrong\u003eDecision_Tree\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.2481%;\" width=\"8.421052631578947%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.773\u0026plusmn;0.142\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"7.368421052631579%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.747-0.799\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"7.368421052631579%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.742\u0026plusmn;0.125\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"7.368421052631579%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.719-0.764\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"8.421052631578947%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.750\u0026plusmn;0.133\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"8.421052631578947%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.726-0.774\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"8.421052631578947%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.773\u0026plusmn;0.101\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.2481%;\" width=\"8.421052631578947%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.755-0.791\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"9.473684210526315%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.756\u0026plusmn;0.101\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.5268%;\" width=\"8.421052631578947%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.738-0.775\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 8.7749%;\" width=\"17.894736842105264%\"\u003e\n \u003cp\u003e\u003cstrong\u003eExtra_Tree\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.2481%;\" width=\"8.421052631578947%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.740\u0026plusmn;0.124\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"7.368421052631579%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.718-0.763\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"7.368421052631579%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.710\u0026plusmn;0.104\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"7.368421052631579%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.691-0.729\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"8.421052631578947%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.718\u0026plusmn;0.116\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"8.421052631578947%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.697-0.739\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"8.421052631578947%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.740\u0026plusmn;0.140\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.2481%;\" width=\"8.421052631578947%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.715-0.765\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"9.473684210526315%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.719\u0026plusmn;0.095\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.5268%;\" width=\"8.421052631578947%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.702-0.736\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 8.7749%;\" width=\"17.894736842105264%\"\u003e\n \u003cp\u003e\u003cstrong\u003eGaussian_Naive_Bayes\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.2481%;\" width=\"8.421052631578947%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.699\u0026plusmn;0.114\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"7.368421052631579%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.679-0.720\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"7.368421052631579%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.606\u0026plusmn;0.114\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"7.368421052631579%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.585-0.626\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"8.421052631578947%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.602\u0026plusmn;0.108\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"8.421052631578947%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.582-0.622\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"8.421052631578947%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.808\u0026plusmn;0.176\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.2481%;\" width=\"8.421052631578947%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.776-0.840\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"9.473684210526315%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.671\u0026plusmn;0.090\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.5268%;\" width=\"8.421052631578947%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.655-0.687\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 8.7749%;\" width=\"17.894736842105264%\"\u003e\n \u003cp\u003e\u003cstrong\u003eGradient_Boosting\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.2481%;\" width=\"8.421052631578947%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.801\u0026plusmn;0.151\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"7.368421052631579%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.774-0.829\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"7.368421052631579%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.747\u0026plusmn;0.135\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"7.368421052631579%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.723-0.772\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"8.421052631578947%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.765\u0026plusmn;0.131\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"8.421052631578947%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.742-0.789\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"8.421052631578947%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.754\u0026plusmn;0.130\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.2481%;\" width=\"8.421052631578947%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.731-0.778\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"9.473684210526315%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.755\u0026plusmn;0.117\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.5268%;\" width=\"8.421052631578947%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.734-0.776\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 8.7749%;\" width=\"17.894736842105264%\"\u003e\n \u003cp\u003e\u003cstrong\u003eLDA\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.2481%;\" width=\"8.421052631578947%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.766\u0026plusmn;0.132\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"7.368421052631579%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.742-0.790\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"7.368421052631579%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.716\u0026plusmn;0.118\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"7.368421052631579%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.695-0.737\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"8.421052631578947%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.715\u0026plusmn;0.103\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"8.421052631578947%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.696-0.734\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"8.421052631578947%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.743\u0026plusmn;0.126\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.2481%;\" width=\"8.421052631578947%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.720-0.765\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"9.473684210526315%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.726\u0026plusmn;0.106\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.5268%;\" width=\"8.421052631578947%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.707-0.745\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 8.7749%;\" width=\"17.894736842105264%\"\u003e\n \u003cp\u003e\u003cstrong\u003eLogistic_Regression\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.2481%;\" width=\"8.421052631578947%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.779\u0026plusmn;0.144\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"7.368421052631579%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.753-0.805\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"7.368421052631579%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.743\u0026plusmn;0.129\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"7.368421052631579%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.720-0.767\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"8.421052631578947%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.749\u0026plusmn;0.120\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"8.421052631578947%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.727-0.771\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"8.421052631578947%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.761\u0026plusmn;0.138\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.2481%;\" width=\"8.421052631578947%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.736-0.786\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"9.473684210526315%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.751\u0026plusmn;0.118\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.5268%;\" width=\"8.421052631578947%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.730-0.773\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 8.7749%;\" width=\"17.894736842105264%\"\u003e\n \u003cp\u003e\u003cstrong\u003eMultinomial_Naive_Bayes\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.2481%;\" width=\"8.421052631578947%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.662\u0026plusmn;0.076\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"7.368421052631579%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.648-0.676\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"7.368421052631579%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.603\u0026plusmn;0.054\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"7.368421052631579%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.593-0.613\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"8.421052631578947%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.602\u0026plusmn;0.051\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"8.421052631578947%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.592-0.611\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"8.421052631578947%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.613\u0026plusmn;0.161\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.2481%;\" width=\"8.421052631578947%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.584-0.642\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"9.473684210526315%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.599\u0026plusmn;0.086\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.5268%;\" width=\"8.421052631578947%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.583-0.614\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 8.7749%;\" width=\"17.894736842105264%\"\u003e\n \u003cp\u003e\u003cstrong\u003ePassive_Aggressive\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.2481%;\" width=\"8.421052631578947%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.694\u0026plusmn;0.148\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"7.368421052631579%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.668-0.721\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"7.368421052631579%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.649\u0026plusmn;0.131\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"7.368421052631579%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.625-0.672\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"8.421052631578947%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.646\u0026plusmn;0.148\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"8.421052631578947%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.620-0.673\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"8.421052631578947%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.648\u0026plusmn;0.186\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.2481%;\" width=\"8.421052631578947%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.615-0.682\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"9.473684210526315%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.641\u0026plusmn;0.156\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.5268%;\" width=\"8.421052631578947%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.612-0.669\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 8.7749%;\" width=\"17.894736842105264%\"\u003e\n \u003cp\u003e\u003cstrong\u003eQDA\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.2481%;\" width=\"8.421052631578947%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.730\u0026plusmn;0.111\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"7.368421052631579%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.710-0.750\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"7.368421052631579%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.680\u0026plusmn;0.100\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"7.368421052631579%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.662-0.698\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"8.421052631578947%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.674\u0026plusmn;0.088\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"8.421052631578947%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.658-0.690\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"8.421052631578947%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.719\u0026plusmn;0.109\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.2481%;\" width=\"8.421052631578947%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.699-0.738\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"9.473684210526315%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.694\u0026plusmn;0.090\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.5268%;\" width=\"8.421052631578947%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.677-0.710\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 8.7749%;\" width=\"17.894736842105264%\"\u003e\n \u003cp\u003e\u003cstrong\u003eRandom_Forest\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.2481%;\" width=\"8.421052631578947%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.791\u0026plusmn;0.152\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"7.368421052631579%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.763-0.818\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"7.368421052631579%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.740\u0026plusmn;0.135\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"7.368421052631579%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.715-0.764\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"8.421052631578947%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.762\u0026plusmn;0.131\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"8.421052631578947%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.738-0.786\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"8.421052631578947%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.731\u0026plusmn;0.129\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.2481%;\" width=\"8.421052631578947%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.708-0.754\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"9.473684210526315%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.743\u0026plusmn;0.121\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.5268%;\" width=\"8.421052631578947%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.721-0.765\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 8.7749%;\" width=\"17.894736842105264%\"\u003e\n \u003cp\u003e\u003cstrong\u003eSGD\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.2481%;\" width=\"8.421052631578947%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.785\u0026plusmn;0.140\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"7.368421052631579%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.760-0.810\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"7.368421052631579%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.740\u0026plusmn;0.132\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"7.368421052631579%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.716-0.764\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"8.421052631578947%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.744\u0026plusmn;0.125\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"8.421052631578947%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.721-0.767\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"8.421052631578947%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.767\u0026plusmn;0.127\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.2481%;\" width=\"8.421052631578947%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.745-0.790\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"9.473684210526315%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.752\u0026plusmn;0.116\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.5268%;\" width=\"8.421052631578947%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.731-0.773\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 8.7749%;\" width=\"17.894736842105264%\"\u003e\n \u003cp\u003e\u003cstrong\u003eXGBoost\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.2481%;\" width=\"8.421052631578947%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.802\u0026plusmn;0.159\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"7.368421052631579%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.773-0.831\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"7.368421052631579%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.753\u0026plusmn;0.146\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"7.368421052631579%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.727-0.779\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"8.421052631578947%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.768\u0026plusmn;0.138\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.389%;\" width=\"8.421052631578947%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.743-0.793\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"8.421052631578947%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.754\u0026plusmn;0.143\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 6.2481%;\" width=\"8.421052631578947%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.728-0.780\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 3.8591%;\" width=\"9.473684210526315%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.758\u0026plusmn;0.134\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 2.5268%;\" width=\"8.421052631578947%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.734-0.782\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 8.7749%;\" width=\"17.346938775510203%\"\u003e\n \u003cp\u003e\u003cstrong\u003eP value\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" style=\"width: 8.683%;\" width=\"16.3265306122449%\"\u003e\n \u003cp\u003e\u003cstrong\u003eP\u0026lt;0.0001\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" style=\"width: 6.2481%;\" width=\"15.306122448979592%\"\u003e\n \u003cp\u003e\u003cstrong\u003eP\u0026lt;0.0001\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" style=\"width: 6.2481%;\" width=\"16.3265306122449%\"\u003e\n \u003cp\u003e\u003cstrong\u003eP\u0026lt;0.0001\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" style=\"width: 10.1072%;\" width=\"17.346938775510203%\"\u003e\n \u003cp\u003e\u003cstrong\u003eP\u0026lt;0.0001\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" style=\"width: 6.3859%;\" width=\"17.346938775510203%\"\u003e\n \u003cp\u003e\u003cstrong\u003eP\u0026lt;0.0001\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e"},{"header":"Discussion","content":"\u003cp\u003eOur research is based on machine learning methods and aims to build artificial intelligence prediction models for FBG and\u0026nbsp;HbAlc\u0026nbsp;in patients with T2DM. By analyzing the influencing factors of related blood glucose control indicators in T2DM patients, we established two prediction models for FBG and\u0026nbsp;HbAlc, and \u003cu\u003ethey\u003c/u\u003e can assist clinical treatment and T2DM patient management, to allow early adjustment of the treatment plan and improve the treatment rate and control rate of T2DM. Our analysis results show that: 1) FBG, HbA1c, medication compliance, and dietary habits have a greater impact on both prediction models, 2) The multi-parameter predictive risk models incorporate variables from different domains, including baseline demographics, complications, and laboratory tests, and can accurately predict three-month FBG values and\u0026nbsp;HbAlc values, 3) In the machine learning-driven algorithm, the optimal models both adopt the ensemble learning algorithm. The influence of different algorithms and data processing methods on the results shows that the algorithms that have the greatest impact on the model effect are ensemble learning and XGBoost, and the best imputing method is the modified random forest imputing. Ensemble learning is an advanced machine learning strategy that can improve classification performance and generalization by combining multiple models\u003csup\u003e[18]\u003c/sup\u003e. Nemat H et al. utilized deep learning and ensemble learning to predict blood glucose levels, and compared the performance of the proposed ensemble model with the non-ensemble model, and the results showed that the developed ensemble model outperformed the non-ensemble baseline model\u003csup\u003e[19]\u003c/sup\u003e. In our research, we combine the remaining model indicators that have been trained, and the integrated model indicators have great advantages.\u003c/p\u003e\n\u003cp\u003eFrom the feature screening results, this study is consistent with other studies, FBG and\u0026nbsp;HbAlc\u0026nbsp;are both the most important predictors.\u0026nbsp;Del Parigi A\u0026nbsp;et al.\u003csup\u003e[20]\u003c/sup\u003e used several machine learning algorithms to find predictors of glycemic control in diabetes and found that HbA1c and FPG were the strongest predictors of achieving glycemic control. This is consistent with our findings. For this result, we explain that current FBG and HbA1c values have important effects on future FBG and HbA1c values, respectively. In our study, FBG and HbAlc were mutually important predictors, indicating an important correlation between glycated hemoglobin and fasting blood glucose. The reason maybe is that once the glucose in human blood combines with hemoglobin to form glycosylated hemoglobin, it will age with the aging of red blood cells, which is the product of an irreversible glycation reaction. The contact time between blood glucose and hemoglobin and the content of blood glucose can determine the level of HbA1c, so the content of HbA1c is positively correlated with the blood glucose content of diabetic patients, which may have the ability to predict each other. Studies have shown that postprandial blood glucose and fasting blood glucose are closely related to glycosylated hemoglobin, and for poorly controlled diabetic patients, the greater the value of HbA1c, the greater the contribution of fasting blood glucose value\u003csup\u003e[21]\u003c/sup\u003e. Wang J et al\u003csup\u003e[22]\u003c/sup\u003e .established a blood glucose prediction model. After feature screening, the top six indicators were: age, fasting glutamate transaminase (ALT), blood urea nitrogen (BUN), total protein (TP), uric acid (UA), and BMI. BMI is also in the top ten important features in this study, and the rest of the indicators did not enter the top ten in importance. However, our study also found that ALT and BUN will have a certain degree of influence on blood glucose.\u0026nbsp;Wang YS\u0026nbsp;et al established a T2DM prediction model in western Xinjiang, China, and used Lasso screening for feature screening\u003csup\u003e[23]\u003c/sup\u003e. The study showed that age, family history of T2DM, waist circumference, TC, TG, BMI, HDL-C, and previous history of hypertension had a significant impact on FBG. These factors are included in the feature selection results of our study. Chien KL et al\u003csup\u003e[24]\u003c/sup\u003e used multiple logistic regression to predict HbA1c and found that both waist circumference and BMI were associated with abnormal glycated hemoglobin levels. Age, family history of diabetes, systolic blood pressure, and biochemical markers including C-reactive protein and triglycerides were significantly associated with higher glycated hemoglobin levels. In our study, waist circumference and BMI had important effects on HbA1c, as did age and hypertension, but the study did not take into account enough variables, and our accuracy is higher. In addition, our study, based on a large sample of physical examination and follow-up data, found that patients\u0026apos; medication compliance, follow-up conditions, and living habits (including dietary habits and smoking) had a greater impact on blood glucose control. The results better clarify the importance of primary prevention of T2DM, which is to focus on changing environmental factors and lifestyles, reducing calorie intake, maintaining a low-salt, low-sugar, high-fiber diet, quitting smoking, limiting alcohol, and getting daily moderate exercise. At the same time, the results of this study also show the importance of follow-up for secondary and tertiary prevention of T2DM. Pourat N et al\u003csup\u003e[25]\u003c/sup\u003e conducted an observational study and found that timely linking behavioral health patients to outpatient follow-up after hospitalization is an effective care transition strategy that may reduce readmission rates. Tong L et al\u003csup\u003e[26]\u003c/sup\u003e also concluded that follow-up was associated with a reduced risk of readmission. Patients benefited the most from outpatient follow-up because face-to-face conversations allowed more information (both therapeutic and emotional) to be exchanged with patients and better individualized care for patients. In addition, adverse reaction monitoring during the follow-up process can timely detect the risk of hypoglycemia in patients, which brings greater benefits to patients.\u003c/p\u003e\n\u003cp\u003eMost studies tend to use machine learning algorithms such as decision trees, random forests, SVM, logistic regression, and neural networks to build T2DM prediction models, with AUC values ranging from 0.7 to 0.9\u003csup\u003e[27.28]\u003c/sup\u003e. Wang J et al adopted three commonly used machine learning algorithms (RF, SVM, and BP-ANN) combined with the elastic network (EN) to simulate and predict blood glucose status in China. The AUCs of RF, SVM and BP were 0.75, 0.72 and 0.72, respectively\u003csup\u003e[29]\u003c/sup\u003e. In a study of T2DM prediction models in Australia by Zhang L et al\u003csup\u003e[30]\u003c/sup\u003e, the model built using XGBoost had the best prediction ability, with a 3-year prediction model AUC value of 0.78 and a 10-year AUC value of 0.75. Xue M et al\u003csup\u003e[31]\u003c/sup\u003e established a T2DM prediction model using algorithms such as decision trees, random forests, AdaBoost with decision trees (AdaBoost), and extreme gradient boosting decision trees (XGBoost), and XGBoost had the best performance (AUC = 0.968). Usually, the AUC value is above 0.8, showing a good classification effect\u003csup\u003e[32]\u003c/sup\u003e. The AUC values of the five optimal FBG models obtained in our study are all greater than 0.8 and the AUC values of the five optimal HbA1c models obtained in our study are all greater than 0.9. Based on incorporating 100,000 pieces of data, 85 variables, and 16 machine learning algorithms for research, we obtained the \u0026nbsp; FBG prediction model with the best AUC value of 0.819 and the HbA1c prediction model with the best AUC value of 0.970, indicating that thes\u003cu\u003ee\u003c/u\u003e prediction models have better performance and better clinical prediction ability. The establishment of thes\u003cu\u003ee two\u003c/u\u003e models can imput the current gap in the prediction model of individualized treatment of T2DM patients, provide new ideas and methods for T2DM treatment, and provide T2DM patients with efficient and accurate individualized treatment plans to solve the real health problems of patients. In addition, this study explores the T2DM prediction model based on real-world medical data mining, uses multiple classifiers for comparative research, and selects the optimal model to ensure the optimization of the model, which effectively makes up for the current shortcomings of using a single classifier. Therefore, this study comprehensively and completely demonstrated the process of predicting the outcome of T2DM drug treatment with the help of data mining technology under the background of real-world research and provided a good methodological reference for the management of other chronic diseases.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eStrengths\u003c/strong\u003e\u003c/p\u003e\n\u003col\u003e\n \u003cli\u003eThe data in this study came from\u0026nbsp;the Public Health Service System and the Medical Record Homepage Management System of the Health Information Center of Sichuan Province. The data quality is reliable enough to meet the needs of modeling.\u003c/li\u003e\n \u003cli\u003eIn the process of data cleaning, this study is not limited to a single data preprocessing method but uses a variety of imputing methods, feature screening methods, and various data preprocessing methods and applies them to the data cleaning of each predictive indicator. The process avoids the possible impact of a certain data preprocessing method on the modeling effect.\u003c/li\u003e\n \u003cli\u003eThe modeling method has been improved in this study. Different from the previous use of one or several algorithms to build predictive models, our study used more than ten machine learning algorithms for modeling and selected the optimal five models. The model results are more reliable and the prediction performance is better.\u003c/li\u003e\n \u003cli\u003eThe information included in the study is more comprehensive, including the basic information about patients, disease-related factors, treatment factors, metabolic index factors, and lifestyle-related factors.\u003c/li\u003e\n\u003c/ol\u003e\n\u003cp\u003e\u003cstrong\u003eLimitations\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e1) This study only uses medical data from Sichuan Province, China for modeling. Differences in lifestyle and ethnicity in different regions may lead to a limited scope of application of the model.\u003c/p\u003e\n\u003cp\u003e2) Although the data set used for validation in this study is independent of the data set used for model development, the two are derived from practice records in the same database, and no more rigorous prospective external validation has been performed.\u003c/p\u003e\n\u003cp\u003e3) In this study, the classification of some variables may be wrong because the system automatically recognizes variables with more than 10 categories as continuous variables, but this operation will not affect the prediction effect of the final prediction model.\u003c/p\u003e\n\u003cp\u003e4) In this study, the AUC values of the two optimal prediction models differed by 0.1, possibly due to too little HbA1c data for modeling. In future studies, if the amount of data used to build the model can be increased, the difference in values may be reduced.\u003c/p\u003e\n\u003cp\u003e5) Some predictive factors, such as the course of diabetes, are not recorded in detail in the original database, so these variables are not included in the modeling, which may have an impact on the prediction results. In the following research, we will use more extensive and detailed data for modeling to obtain a more accurate prediction model.\u003c/p\u003e"},{"header":"Conclusion","content":"\u003cp\u003eIn this study, the three-month FBG prediction model and three-month HbA1c prediction model for T2DM patients were constructed using population data from Sichuan Province, China. The patient's FBG and HbA1c are both the most important predictors of two kinds of prediction models. This research can provide a methodological reference for other prediction models. The AUC values of the five best FBG prediction models finally established are all greater than 0.8, and the AUC values of the five best HbA1c prediction models finally established are all greater than 0.9, which could accurately predict the FBG and HbA1c for clinical applications.\u003c/p\u003e"},{"header":"Abbreviations","content":"\u003cp\u003eALT Alanine aminotransferase\u003c/p\u003e\n\u003cp\u003eAST\u0026nbsp; \u0026nbsp; \u0026nbsp;Aspartate aminotransferase\u003c/p\u003e\n\u003cp\u003eAUC Area under the curve\u003c/p\u003e\n\u003cp\u003eBMI Body mass index\u003c/p\u003e\n\u003cp\u003eBP-ANN\u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;Back-propagation artificial neural network\u003c/p\u003e\n\u003cp\u003eBUN Blood urea nitrogen\u003c/p\u003e\n\u003cp\u003eCI Confidence intervals\u003c/p\u003e\n\u003cp\u003eCPM Cycles per minute\u003c/p\u003e\n\u003cp\u003eDBP Diastolic blood pressure\u003c/p\u003e\n\u003cp\u003eEN\u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;Elastic network\u003c/p\u003e\n\u003cp\u003eFBG\u0026nbsp; \u0026nbsp; \u0026nbsp; Fasting blood glucose\u003c/p\u003e\n\u003cp\u003eFig Figure\u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eHb\u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;\u0026nbsp;Hemoglobin\u003c/p\u003e\n\u003cp\u003eHbA1c Glycosylated hemoglobin\u003c/p\u003e\n\u003cp\u003eHBP High blood pressure\u003c/p\u003e\n\u003cp\u003eHDL-C\u0026nbsp;High density lipoprotein cholesterol\u003c/p\u003e\n\u003cp\u003eLDA Linear discriminant analysis\u003c/p\u003e\n\u003cp\u003eP-R curve Precision-recall curve\u003c/p\u003e\n\u003cp\u003eQDA Quadratic discriminant analysis\u003c/p\u003e\n\u003cp\u003eRF Random forest\u003c/p\u003e\n\u003cp\u003eROC Receiver operating characteristic curve\u003c/p\u003e\n\u003cp\u003eSBP\u0026nbsp; \u0026nbsp; \u0026nbsp;\u0026nbsp;Systolic blood pressure\u003c/p\u003e\n\u003cp\u003eScr\u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;Serum creatinine\u003c/p\u003e\n\u003cp\u003eSD\u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;\u0026nbsp;Standard deviation\u003c/p\u003e\n\u003cp\u003eSGD Stochastic Gradient Descent\u003c/p\u003e\n\u003cp\u003eSVM Support vector machine\u003c/p\u003e\n\u003cp\u003eTC Total cholesterol\u003c/p\u003e\n\u003cp\u003eTG\u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;Triglyceride\u003c/p\u003e\n\u003cp\u003eTP\u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;\u0026nbsp;Total protein\u003c/p\u003e\n\u003cp\u003eUA \u0026nbsp; \u0026nbsp; \u0026nbsp; Uric acid\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eEthics approval and consent to participate\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis study was approved by the Medical Ethics Committee of the Sichuan Academy of Medical Sciences (Sichuan Provincial People\u0026apos;s Hospital). Details are in the supplementary material. A statement to confirm that all methods were carried out in accordance with relevant guidelines and regulations (declaration of helsinki). We collected patient data retrospectively, so informed consent could not be obtained from all participants. However, we used a unique ID to identify patient connection information, and all research operations carried out would not be traced to the individual patient, and the patient\u0026apos;s sensitive personal information (such as name, phone number, address, work unit, responsible doctor, etc.) would be deleted. The study has passed the ethical review, and patient privacy will not be disclosed.\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\u003eAvailability of data and materials\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe data that support the findings of this study are available from Sichuan Provincial Health Commission, but restrictions apply to the availability of these data, which were used under license for the current study, and so are not publicly available. Data are however available from the authors upon reasonable request and with permission of Sichuan Provincial Health Commission.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCompeting interests\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors declare that they have no competing interests.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eEWL was funded by the Sichuan Provincial Department of Science and Technology (Grant No. 2021YFS0197). MJ is supported by the Personalized Drug Therapy Key Laboratory of Sichuan Provincial People\u0026rsquo;s Hospital. XWW was supported by the National Natural Science Foundation of China (grant number: 72004020).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthors\u0026apos; contributions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eEWL and XT contributed to the idea and design of the study. MJ and BQZ conducted the literature search. YML and QH conducted the data analysis and created the figures. YML wrote the manuscript with support from WMG, MJ, XT, and XWW. YX, JQH, XS, RST and EWL provided critical feedback on the analysis and its interpretation and commented on the drafted manuscript. WYL, RST, XWW and EWL provided final approval of the manuscript for publication. XT is responsible for the integrity of the work as a whole.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAcknowledgements\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNot applicable\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthors\u0026apos; information (optional)\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNot applicable\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eAmerican Diabetes Association. Introduction: Standards of Medical Care in Diabetes-2022. Diabetes Care. 2022;45(Suppl 1):S1-S2.\u003c/li\u003e\n\u003cli\u003eSun H, Saeedi P, Karuranga S, et al. IDF Diabetes Atlas: Global, regional and country-level diabetes prevalence estimates for 2021 and projections for 2045. Diabetes Res Clin Pract. 2022 Jan;183:109119.\u003c/li\u003e\n\u003cli\u003eOgurtsova K, Guariguata L, Barengo NC, et al. IDF diabetes Atlas: Global estimates of undiagnosed diabetes in adults for 2021. Diabetes Res Clin Pract. 2022;183:109118. \u003c/li\u003e\n\u003cli\u003eZheng Y, Ley SH, Hu FB. Global aetiology and epidemiology of type 2 diabetes mellitus and its complications. Nat Rev Endocrinol. 2018 Feb;14(2):88-98.\u003c/li\u003e\n\u003cli\u003eHarding JL, Pavkov ME, Magliano DJ,et al. Global trends in diabetes complications: a review of current evidence. Diabetologia. 2019 Jan;62(1):3-16.\u003c/li\u003e\n\u003cli\u003eSafiri S, Karamzad N, Kaufman JS, et al. Prevalence, Deaths and Disability-Adjusted-Life-Years (DALYs) Due to Type 2 Diabetes and Its Attributable Risk Factors in 204 Countries and Territories, 1990-2019: Results From the Global Burden of Disease Study 2019. Front Endocrinol (Lausanne). 2022 Feb 25;13:838027. \u003c/li\u003e\n\u003cli\u003eXiao W, Jing L, Xu Y, et al. Different Data Mining Approaches Based Medical Text Data. J Healthc Eng. 2021 Dec 6;2021:1285167.\u003c/li\u003e\n\u003cli\u003eMakridakis S, Spiliotis E, Assimakopoulos V. Statistical and Machine Learning forecasting methods: Concerns and ways forward. PLoS One. 2018 Mar 27;13(3):e0194889.\u003c/li\u003e\n\u003cli\u003eLiao WC, Tu YK, Wu MS, et al. Blood glucose concentration and risk of pancreatic cancer: systematic review and dose-response meta-analysis. BMJ. 2015 Jan 2;350:g7371. \u003c/li\u003e\n\u003cli\u003eNagalakshmi CS, Santhosh NU, Krishnamurthy N, et al. Role of Altered Venous Blood Lactate and HbA1c in Women with Gestational Diabetes Mellitus. J Clin Diagn Res. 2016 Dec;10(12):BC18-BC20. \u003c/li\u003e\n\u003cli\u003eJia W, Weng J, Zhu D, et al; Chinese Diabetes Society. Standards of medical care for type 2 diabetes in China 2019. Diabetes Metab Res Rev. 2019 Sep;35(6):e3158. \u003c/li\u003e\n\u003cli\u003eFregoso-Aparicio L, Noguez J, Montesinos L, et al. Machine learning and deep learning predictive models for type 2 diabetes: a systematic review. Diabetol Metab Syndr. 2021 Dec 20;13(1):148. \u003c/li\u003e\n\u003cli\u003eZhu T, Li K, Herrero P, Georgiou P. Deep Learning for Diabetes: A Systematic Review. IEEE J Biomed Health Inform. 2021 Jul;25(7):2744-2757.\u003c/li\u003e\n\u003cli\u003eWHO. World Health Organization. ICD-10 version:2010 [Internet], 2010. Available from https://icd.who.int/browse10/2019/en. Accessed 16 March 2021.\u003c/li\u003e\n\u003cli\u003e Greenwood CJ, Youssef GJ, Letcher P, et al. A comparison of penalised regression methods for informing the selection of predictive markers. PLoS One. 2020;15(11):e0242730. Published 2020 Nov 20.\u003c/li\u003e\n\u003cli\u003e Zhao Z, Yang W, Zhai Y, et al. Identify DNA-Binding Proteins Through the Extreme Gradient Boosting Algorithm. Front Genet. 2022;12:821996. Published 2022 Jan 28.\u003c/li\u003e\n\u003cli\u003e Yang H., Luo Y., Ren X., et al. Risk Prediction of Diabetes: Big Data Mining with Fusion of Multifarious Physical Examination Indicators. Inf. 2021; 75:140\u0026ndash;149. \u003c/li\u003e\n\u003cli\u003eJin LP, Dong J. Ensemble Deep Learning for Biomedical Time Series Classification. Comput Intell Neurosci. 2016;2016:6212684. doi: 10.1155/2016/6212684. Epub 2016 Sep 20.\u003c/li\u003e\n\u003cli\u003eNemat H, Khadem H, Eissa MR, et al. Blood Glucose Level Prediction: Advanced Deep-Ensemble Learning Approach. IEEE J Biomed Health Inform. 2022 Jan 25;PP.\u003c/li\u003e\n\u003cli\u003eDel Parigi A, Tang W, Liu D, et al. Machine Learning to Identify Predictors of Glycemic Control in Type 2 Diabetes: An Analysis of Target HbA1c Reduction Using Empagliflozin/Linagliptin Data. Pharmaceut Med. 2019 Jun;33(3):209-217.\u003c/li\u003e\n\u003cli\u003eMakris K, Spanou L, Rambaouni-Antoneli A, et al. Relationship between mean blood glucose and glycated haemoglobin in Type 2 diabetic patients. Diabet Med. 2008 Feb;25(2):174-8.\u003c/li\u003e\n\u003cli\u003eWang J, Wang F, Liu Y, et al. Multiple Linear Regression and Artificial Neural Network to Predict Blood Glucose in Overweight Patients. Exp Clin Endocrinol Diabetes. 2016 Jan;124(1):34-8. \u003c/li\u003e\n\u003cli\u003eWang YS, Zhang YS, Wang K, et al. Nomogram Model for Screening the Risk of Type II Diabetes in Western Xinjiang, China.[J] .Diabetes Metab Syndr Obes, 2021, 14: 3541-3553.\u003c/li\u003e\n\u003cli\u003eChien KL, Lin HJ, Lee BC, et al. Prediction model for high glycated hemoglobin concentration among ethnic Chinese in Taiwan. Cardiovasc Diabetol. 2010 Sep 27;9:59.\u003c/li\u003e\n\u003cli\u003ePourat N, Chen X, Wu SH, Davis AC. Timely Outpatient Follow-up Is Associated with Fewer Hospital Readmissions among Patients with Behavioral Health Conditions. J Am Board Fam Med. 2019 May-Jun;32(3):353-361.\u003c/li\u003e\n\u003cli\u003eTong L, Arnold T, Yang J, et al. The association between outpatient follow-up visits and all-cause non-elective 30-day readmissions: A retrospective observational cohort study. PLoS One. 2018 Jul 17;13(7):e0200691.\u003c/li\u003e\n\u003cli\u003eDe Silva K, Enticott J, Barton C, et al. Use and performance of machine learning models for type 2 diabetes prediction in clinical and community care settings: Protocol for a systematic review and meta-analysis of predictive modeling studies. Digit Health. 2021 Sep 28;7:20552076211047390.\u003c/li\u003e\n\u003cli\u003eKim H, Lim DH, Kim Y. Classification and Prediction on the Effects of Nutritional Intake on Overweight/Obesity, Dyslipidemia, Hypertension and Type 2 Diabetes Mellitus Using Deep Learning Model: 4-7th Korea National Health and Nutrition Examination Survey. Int J Environ Res Public Health. 2021 May 24;18(11):5597.\u003c/li\u003e\n\u003cli\u003eWang J, Wang MY, Wang H, et al. Status of glycosylated hemoglobin and prediction of glycemic control among patients with insulin-treated type 2 diabetes in North China: a multicenter observational study. Chin Med J (Engl). 2020 Jan 5;133(1):17-24.\u003c/li\u003e\n\u003cli\u003eZhang L, Shang X, Sreedharan S, et al. Predicting the Development of Type 2 Diabetes in a Large Australian Cohort Using Machine-Learning Techniques: Longitudinal Survey Study. JMIR Med Inform. 2020 Jul 28;8(7):e16850.\u003c/li\u003e\n\u003cli\u003eXue M, Su Y, Li C, et al. Identification of Potential Type II Diabetes in a Large-Scale Chinese Population Using a Systematic Machine Learning Framework. J Diabetes Res. 2020 Sep 24;2020:6873891.\u003c/li\u003e\n\u003cli\u003eMandrekar JN. Receiver operating characteristic curve in diagnostic test assessment. J Thorac Oncol. 2010 Sep;5(9):1315-6.\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":true,"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":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"fasting blood glucose, glycated hemoglobin, type 2 diabetes mellitus, prediction model, machine learning","lastPublishedDoi":"10.21203/rs.3.rs-1868105/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-1868105/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003ch2\u003eBackground\u003c/h2\u003e \u003cp\u003eType 2 diabetes is the type with the largest proportion of people with diabetes.With the progression of the disease, patients with type 2 diabetes mellitus will have different degrees of complications, which will seriously reduce the quality of life of the patients and bring a heavy economic burden to the patient's families. Therefore, establishing a predictive model for glycemic control in patients with type 2 diabetes mellitus is of great help in optimizing the treatment of type 2 diabetes mellitus and delaying disease progression.\u003c/p\u003e\u003ch2\u003eDesign and Methods:\u003c/h2\u003e \u003cp\u003eA retrospective study was conducted on type 2 diabetes mellitus real-world medical data from 4 cities in Sichuan Province, China from January 2015 to December 2020, including basic patient information, medication status, laboratory results, dietary habits, exercise status, and the actual follow-up of the patient after treatment. After data preprocessing, data inputting, data sampling, and feature screening, 16 kinds of machine learning methods were used to construct fasting blood glucose prediction models and glycated hemoglobin prediction models for type 2 diabetes mellitus patients, and 5 prediction models with the best prediction performance were screened respectively.\u003c/p\u003e\u003ch2\u003eResults\u003c/h2\u003e \u003cp\u003eA total of 375,723 cases of type 2 diabetes mellitus patients were collected, 10,000 cases were included to establish the fasting blood glucose model, and 2,169 cases were established to establish the HbA1c model. The best prediction model both of fasting blood glucose and HbA1c finally obtained are realized by ensemble learning and modified random forest inputting, the AUC value are 0.819 and 0.970, respectively. The most important indicators of the fasting blood glucose and glycated hemoglobin prediction model were fasting blood glucose and glycated hemoglobin. Medication compliance, follow-up outcome, dietary habits, BMI, and waist circumference also had a greater impact onfasting blood glucose levels. But on the glycated hemoglobin level, laboratory indicators such as platelets, Serum creatinine, Aspartate Transaminase, Hemoglobin, etc. had more impact.\u003c/p\u003e\u003ch2\u003eConclusion\u003c/h2\u003e \u003cp\u003eThe prediction accuracy of the models of the two blood glucose control indicators is high and has certain clinical applicability. Glycated hemoglobin and fasting blood glucose are mutually important predictors, and there is a close relationship between them.\u003c/p\u003e","manuscriptTitle":"Predicting three-month fasting blood glucose and glycated hemoglobin of patients with type 2 diabetes based on multiple machine learning algorithms","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2022-09-15 12:47:48","doi":"10.21203/rs.3.rs-1868105/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"51e27d76-20b2-49d7-b816-afcc4d0220d7","owner":[],"postedDate":"September 15th, 2022","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2022-10-04T07:44:19+00:00","versionOfRecord":[],"versionCreatedAt":"2022-09-15 12:47:48","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-1868105","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-1868105","identity":"rs-1868105","version":["v1"]},"buildId":"ApUGefWb6u5IBVtyqm6d5","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}