Prediction of fasting plasma glucose status using anthropometric measures for diagnosing type 2 diabetes

RFFE - Random Forest Fuzzy Entropy for the classification of Diabetes Mellitus

Diabetes is a category of metabolic disease commonly known as a chronic illness. It causes the body to generate less insulin and raises blood sugar levels, leading to various issues and disrupting the functioning of organs, including the retinal, kidney and nerves. To prevent this, people with chronic illnesses require lifetime access to treatment. As a result, early diabetes detection is essential and might save many lives. Diagnosis of people at high risk of developing diabetes is utilized for preventing the disease in various aspects. This article presents a chronic illness prediction prototype based on a person's risk feature data to provide an early prediction for diabetes with Fuzzy Entropy random vectors that regulate the development of each tree in the Random Forest. The proposed prototype consists of data imputation, data sampling, feature selection, and various techniques to predict the disease, such as Fuzzy Entropy, Synthetic Minority Oversampling Technique (SMOTE), Convolutional Neural Network (CNN) with Stochastic Gradient Descent with Momentum (SGDM), Support Vector Machines (SVM), Classification and Regression Tree (CART), K-Nearest Neighbor (KNN), and Naïve Bayes (NB). This study uses the existing Pima Indian Diabetes (PID) dataset for diabetic disease prediction. The predictions' true/false positive/negative rate is investigated using the confusion matrix and the receiver operating characteristic area under the curve (ROCAUC). Findings on a PID dataset are compared with machine learning algorithms revealing that the proposed Random Forest Fuzzy Entropy (RFFE) is a valuable approach for diabetes prediction, with an accuracy of 98 percent.

Recent applications of machine learning and deep learning models in the prediction, diagnosis, and management of diabetes: a comprehensive review

Diabetes as a metabolic illness can be characterized by increased amounts of blood glucose. This abnormal increase can lead to critical detriment to the other organs such as the kidneys, eyes, heart, nerves, and blood vessels. Therefore, its prediction, prognosis, and management are essential to prevent harmful effects and also recommend more useful treatments. For these goals, machine learning algorithms have found considerable attention and have been developed successfully. This review surveys the recently proposed machine learning (ML) and deep learning (DL) models for the objectives mentioned earlier. The reported results disclose that the ML and DL algorithms are promising approaches for controlling blood glucose and diabetes. However, they should be improved and employed in large datasets to affirm their applicability.

Development and validation of prediction models for hypertension risks: A cross-sectional study based on 4,287,407 participants

Objective

To develop an optimal screening model to identify the individuals with a high risk of hypertension in China by comparing tree-based machine learning models, such as classification and regression tree, random forest, adaboost with a decision tree, extreme gradient boosting decision tree, and other machine learning models like an artificial neural network, naive Bayes, and traditional logistic regression models.

Methods

A total of 4,287,407 adults participating in the national physical examination were included in the study. Features were selected using the least absolute shrinkage and selection operator regression. The Borderline synthetic minority over-sampling technique was used for data balance. Non-laboratory and semi-laboratory analyses were carried out in combination with the selected features. The tree-based machine learning models, other machine learning models, and traditional logistic regression models were constructed to identify individuals with hypertension, respectively. Top features selected using the best algorithm and the corresponding variable importance score were visualized.

Results

A total of 24 variables were finally included for analyses after the least absolute shrinkage and selection operator regression model. The sample size of hypertensive patients in the training set was expanded from 689,025 to 2,312,160 using the borderline synthetic minority over-sampling technique algorithm. The extreme gradient boosting decision tree algorithm showed the best results (area under the receiver operating characteristic curve of non-laboratory: 0.893 and area under the receiver operating characteristic curve of semi-laboratory: 0.894). This study found that age, systolic blood pressure, waist circumference, diastolic blood pressure, albumin, drinking frequency, electrocardiogram, ethnicity (uyghur, hui, and other), body mass index, sex (female), exercise frequency, diabetes mellitus, and total bilirubin are important factors reflecting hypertension. Besides, some algorithms included in the semi-laboratory analyses showed less improvement in the predictive performance compared to the non-laboratory analyses.

Conclusion

Using multiple methods, a more significant prediction model can be built, which discovers risk factors and provides new insights into the prediction and prevention of hypertension.

A Machine Learning Based Framework to Identify and Classify Non-alcoholic Fatty Liver Disease in a Large-Scale Population

Non-alcoholic fatty liver disease (NAFLD) is a common serious health problem worldwide, which lacks efficient medical treatment. We aimed to develop and validate the machine learning (ML) models which could be used to the accurate screening of large number of people. This paper included 304,145 adults who have joined in the national physical examination and used their questionnaire and physical measurement parameters as model's candidate covariates. Absolute shrinkage and selection operator (LASSO) was used to feature selection from candidate covariates, then four ML algorithms were used to build the screening model for NAFLD, used a classifier with the best performance to output the importance score of the covariate in NAFLD. Among the four ML algorithms, XGBoost owned the best performance (accuracy = 0.880, precision = 0.801, recall = 0.894, F-1 = 0.882, and AUC = 0.951), and the importance ranking of covariates is accordingly BMI, age, waist circumference, gender, type 2 diabetes, gallbladder disease, smoking, hypertension, dietary status, physical activity, oil-loving and salt-loving. ML classifiers could help medical agencies achieve the early identification and classification of NAFLD, which is particularly useful for areas with poor economy, and the covariates' importance degree will be helpful to the prevention and treatment of NAFLD.

Comparison of anthropometric and body composition indices in the identification of metabolic risk factors

Whether anthropometric or body composition indices are better indicators of metabolic risk remains unclear. The objectives of this study were to compare the association of metabolic risk factors with anthropometric and body composition indices and to identify the better indicators for risk factors in a large-scale Korean population. In this cross-sectional study, the associations of body mass index (BMI), waist circumference (WC), and waist-to-height ratio (WHtR) as anthropometric indices and trunk fat mass (TFM), percent trunk fat mass (%TFM), whole-body total fat mass (WBTFM), and percent whole-body total fat mass (%WBTFM) as body composition indices with metabolic risk factors were compared by complex-samples multiple logistic regression models based on complex-sample survey data. In men, WHtR, BMI, and TFM were similarly associated with hypertension. Diabetes, hyperlipidemia, and hypo-high-density lipoprotein (HDL) cholesterolemia tended to be more strongly associated with WHtR and WC than body composition indices. Hypertriglyceridemia and hypercholesterolemia were more strongly associated with WHtR and %TFM than other indices. In women, hypertension tended to be more strongly associated with WHtR than other indices. TFM, %TFM, and WHtR were similarly associated with hyperlipidemia. Diabetes and hypo-HDL cholesterolemia were more strongly associated with WHtR and WC than body composition indices. Hypertriglyceridemia and hypercholesterolemia were more strongly associated with WHtR and %TFM than other indices. Among six metabolic risk factors, the validity and utility of the anthropometric indices in identifying risk factors tended to be similar to or better than those of the body composition indices, except for hypertension and hypercholesterolemia in men and hyperlipidemia and hypercholesterolemia in women.

Anthropometry did not approach adequate predictive accuracies for detecting elevated fasting plasma glucose or hemoglobin A1c levels among Chinese children aged 7-9 years

AIMS/INTRODUCTION

Elevated concentrations of fasting plasma glucose (FPG) or hemoglobin A1c (HbA1c) are well-established independent risk factors for progression to diabetes, cardiovascular comorbidities and mortality. Most previous studies on the relationships of anthropometric measures with hyperglycemia were carried out among adults and adolescents, but few data are available for the performance predication of the predictors for diagnosing elevated FPG or HbA1c among young children.

MATERIALS AND METHODS

Involving 5,556 students of aged 7-9 years, a school-based cross-sectional survey was carried out between March and June 2019 in Shenzhen, China. Receiver operating characteristic curve analysis was utilized.

RESULTS

The median was 4.6 (interquartile range [IQR] 4.3-4.8) mmol/L for FPG and 5.3% (IQR 5.1-5.5%) for HbA1c levels for all participants. For detecting elevated FPG, weight (0.651, IQR 0.583-0.719) and waist circumference (0.650, IQR 0.584-0.717) showed the highest area under the curve and 95% confidence interval, followed by body mass index and the z-score of body mass index (both 0.635, IQR 0.567-0.703); other anthropometric measures showed poorer diagnostic efficiencies or no ability. For detecting elevated HbA1c, lower efficiencies for the Conicity Index (0.651, IQR 0.583-0.719), waist-to-height ratio, waist-to-hip ratio and waist-to-chest ratio were shown. The correlations of FPG and HbA1c levels with anthropometric indices were weak (Spearman's r ≤ 0.179).

CONCLUSIONS

None of the evaluated anthropometric indicators approached an adequate predictive accuracy for the detection of elevated FPG or HbA1c levels in Shenzhen children aged 7-9 years. The current study did not recommend anthropometry screening for prediabetes in young children.

A Noninvasive Prediction Model for Hepatitis B Virus Disease in Patients with HIV: Based on the Population of Jiangsu, China

Objective

To establish a machine learning model for identifying patients coinfected with hepatitis B virus (HBV) and human immunodeficiency virus (HIV) through two sexual transmission routes in Jiangsu, China.

Methods

A total of 14197 HIV cases transmitted by homosexual and heterosexual routes were recruited. After data processing, 12469 cases (HIV and HBV, 1033; HIV, 11436) were left for further analysis, including 7849 cases with homosexual transmission and 4620 cases with heterosexual transmission. Univariate logistic regression was used to select variables with significant P value and odds ratio for multivariable analysis. In homosexual transmission and heterosexual transmission groups, 10 and 6 variables were selected, respectively. For identifying HIV individuals coinfected with HBV, a machine learning model was constructed with four algorithms, including Decision Tree, Random Forest, AdaBoost with decision tree (AdaBoost), and extreme gradient boosting decision tree (XGBoost). The detective value of each variable was calculated using the optimal machine learning algorithm.

Results

AdaBoost algorithm showed the highest efficiency in both transmission groups (homosexual transmission group: accuracy = 0.928, precision = 0.915, recall = 0.944, F − 1 = 0.930, and AUC = 0.96; heterosexual transmission group: accuracy = 0.892, precision = 0.881, recall = 0.905, F − 1 = 0.893, and AUC = 0.98). Calculated by AdaBoost algorithm, the detective value of PLA was the highest in homosexual transmission group, followed by CR, AST, HB, ALT, TBIL, leucocyte, age, marital status, and treatment condition; in the heterosexual transmission group, the detective value of PLA was the highest (consistent with the condition in the homosexual group), followed by ALT, AST, TBIL, leucocyte, and symptom severity.

Conclusions

The univariate logistics regression combined with the AdaBoost algorithm could accurately screen the risk factors of HBV in HIV coinfection without invasive testing. Further studies are needed to evaluate the utility and feasibility of this model in various settings.

Different Curve Shapes of Fasting Glucose and Various Obesity-Related Indices by Diabetes and Sex

Fasting plasma glucose (FPG) and obesity-related indices are prognostic factors for adverse outcomes in both subjects with and without diabetes. A few studies have investigated sex differences in obesity indices related to the risk of diabetes, however no studies have compared the relationship between FPG and obesity-related indices by diabetes and sex. Therefore, in this study, we compared the curve shapes of FPG and various obesity-related indices by diabetes, and further explored sex differences in these associations. Data were derived from the Taiwan Biobank database, which included 5000 registered individuals. We used an adjusted generalized linear regression model and calculated the difference of least square means (Lsmean; standard error, SE) for males and females with and without diabetes. Associations between obesity-related indices and fasting glucose level by diabetes and sex groups were estimated, and the ORTHOREG procedure was used to construct B-splines. The post-fitting for linear models procedure was used to determine the range at which the trends separated significantly. The diabetes/sex/FPG interaction term was significant for all obesity-related indices, including body mass index, waist circumference, hip circumference, waist-to-hip ratio, waist-to-height ratio, lipid accumulation product, body roundness index, conicity index, body adiposity index and abdominal volume index. B-spline comparisons between males and females did not reach significance. However, FPG affected the trend towards obesity-related indices. As the fasting glucose level increased, the values of obesity-related indices varied more obviously in the participants without diabetes than in those with diabetes mellitus. The current study revealed that there was a different relationship between FPG and obesity-related indices by diabetes and sex. FPG affected the trend towards obesity-related indices more obviously in participants without diabetes than in those with diabetes. Further studies with a longitudinal design would provide a better understanding of the underlying mechanisms for the relationships.

Importance of general adiposity, visceral adiposity and vital signs in predicting blood biomarkers using machine learning

INTRODUCTION

Blood biomarkers are measured for their ability to characterise physiological and disease states. Much is known about linear relations between blood biomarker concentrations and individual vital signs or adiposity indexes (eg, BMI). Comparatively little is known about non-linear relations with these easily accessible features, particularly when they are modelled in combination and can potentially interact with one another.

METHODS

In this study, we used advanced machine learning algorithms to create non-linear computational models for predicting blood biomarkers (cells, lipids, metabolic factors) from age, general adiposity (BMI), visceral adiposity (Waist-to-Height Ratio, a Body Shape Index) and vital signs (systolic blood pressure, diastolic blood pressure, pulse). We determined the predictive power of the overall feature set. We further calculated feature importance in our models to identify the features with the strongest relations with each blood biomarker. Data were collected in 2018 and 2019 and analysed in 2020.

RESULTS

Our findings characterise previously unknown relations between these predictors and blood biomarkers; in many instances the importance of certain features or feature classes (general adiposity, visceral adiposity or vital signs) differed from their expected contribution based on simplistic linear modelling techniques.

CONCLUSIONS

This work could lead to the formation of new hypotheses for explaining complex biological systems and informs the creation of predictive models for potential clinical applications.

A nomogram model for screening the risk of diabetes in a large-scale Chinese population: an observational study from 345,718 participants

Our study is major to establish and validate a simple type||diabetes mellitus (T2DM) screening model for identifying high-risk individuals among Chinese adults. A total of 643,439 subjects who participated in the national health examination had been enrolled in this cross-sectional study. After excluding subjects with missing data or previous medical history, 345,718 adults was included in the final analysis. We used the least absolute shrinkage and selection operator models to optimize feature selection, and used multivariable logistic regression analysis to build a predicting model. The results showed that the major risk factors of T2DM were age, gender, no drinking or drinking/time > 25 g, no exercise, smoking, waist-to-height ratio, heart rate, systolic blood pressure, fatty liver and gallbladder disease. The area under ROC was 0.811 for development group and 0.814 for validation group, and the p values of the two calibration curves were 0.053 and 0.438, the improvement of net reclassification and integrated discrimination are significant in our model. Our results give a clue that the screening models we conducted may be useful for identifying Chinses adults at high risk for diabetes. Further studies are needed to evaluate the utility and feasibility of this model in various settings.

Identification of metabolic syndrome using phenotypes consisting of triglyceride levels with anthropometric indices in Korean adults

BACKGROUND

The metabolic syndrome (MetS) has shown strong associations with the hypertriglyceridemic waist (HW) phenotype. The best anthropometric indicator of MetS remains controversial. Furthermore, no studies have investigated alternative indices that could replace waist circumference in the HW phenotype. The objectives of this study were to find the best indicator of MetS among anthropometric indices and to examine the predictive power of phenotypes consisting of triglyceride levels with anthropometric indices.

METHODS

A total of 12,025 subjects participated in this retrospective cross-sectional study. All subjects were recruited between November 2016 and August 2007 from hospitals in 28 urban and rural regions in South Korea. The data analyzed in this study were obtained from the Korean Health and Genome Epidemiology Study database and the Korea Institute of Oriental Medicine.

RESULTS

The proportion of patients with MetS ranged from 9 to 57% according to age and gender groups. Waist circumference (WC) was best indicator of MetS in men of all age groups. However, in women aged 40-49 years, the waist-to-height ratio (WHtR) was the best indicator of MetS. Rib circumference and chest circumference were the strongest indicators in women aged 50-59 years and 70-79 years, respectively. The combination of WC and triglyceride (TG) was the best indicator of MetS in men and women overall. However, interestingly, the best indicator was TG + WHtR in women aged 40-49 years and TG + forehead-to-waist ratio in women aged 70-79 years.

CONCLUSIONS

The best indicator of MetS in terms of individual anthropometric indices and the various phenotypes combining a single anthropometric index with TG may differ subtly according to age group in women, but not in men. Our findings provide insight into a simple and inexpensive method that could be used to identify MetS in initial health screening efforts in epidemiology and public health.

Identification of Potential Type II Diabetes in a Large-Scale Chinese Population Using a Systematic Machine Learning Framework

BACKGROUND

An estimated 425 million people globally have diabetes, accounting for 12% of the world's health expenditures, and the number continues to grow, placing a huge burden on the healthcare system, especially in those remote, underserved areas.

METHODS

A total of 584,168 adult subjects who have participated in the national physical examination were enrolled in this study. The risk factors for type II diabetes mellitus (T2DM) were identified by p values and odds ratio, using logistic regression (LR) based on variables of physical measurement and a questionnaire. Combined with the risk factors selected by LR, we used a decision tree, a random forest, AdaBoost with a decision tree (AdaBoost), and an extreme gradient boosting decision tree (XGBoost) to identify individuals with T2DM, compared the performance of the four machine learning classifiers, and used the best-performing classifier to output the degree of variables' importance scores of T2DM.

RESULTS

The results indicated that XGBoost had the best performance (accuracy = 0.906, precision = 0.910, recall = 0.902, F-1 = 0.906, and AUC = 0.968). The degree of variables' importance scores in XGBoost showed that BMI was the most significant feature, followed by age, waist circumference, systolic pressure, ethnicity, smoking amount, fatty liver, hypertension, physical activity, drinking status, dietary ratio (meat to vegetables), drink amount, smoking status, and diet habit (oil loving).

CONCLUSIONS

We proposed a classifier based on LR-XGBoost which used fourteen variables of patients which are easily obtained and noninvasive as predictor variables to identify potential incidents of T2DM. The classifier can accurately screen the risk of diabetes in the early phrase, and the degree of variables' importance scores gives a clue to prevent diabetes occurrence.

Artificial Intelligence Applications in Type 2 Diabetes Mellitus Care: Focus on Machine Learning Methods

Objectives

The incidence of type 2 diabetes mellitus has increased significantly in recent years. With the development of artificial intelligence applications in healthcare, they are used for diagnosis, therapeutic decision making, and outcome prediction, especially in type 2 diabetes mellitus. This study aimed to identify the artificial intelligence (AI) applications for type 2 diabetes mellitus care.

Methods

This is a review conducted in 2018. We searched the PubMed, Web of Science, and Embase scientific databases, based on a combination of related mesh terms. The article selection process was based on Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA). Finally, 31 articles were selected after inclusion and exclusion criteria were applied. Data gathering was done by using a data extraction form. Data were summarized and reported based on the study objectives.

Results

The main applications of AI for type 2 diabetes mellitus care were screening and diagnosis in different stages. Among all of the reviewed AI methods, machine learning methods with 71% (n = 22) were the most commonly applied techniques. Many applications were in multi method forms (23%). Among the machine learning algorithms applications, support vector machine (21%) and naive Bayesian (19%) were the most commonly used methods. The most important variables that were used in the selected studies were body mass index, fasting blood sugar, blood pressure, HbA1c, triglycerides, low-density lipoprotein, high-density lipoprotein, and demographic variables.

Conclusions

It is recommended to select optimal algorithms by testing various techniques. Support vector machine and naive Bayesian might achieve better performance than other applications due to the type of variables and targets in diabetes-related outcomes classification.

Nomogram prediction for the 3-year risk of type 2 diabetes in healthy mainland China residents

Aims

To develop a precise personalized type 2 diabetes mellitus (T2DM) prediction model by cost-effective and readily available parameters in a Central China population.

Methods

A 3-year cohort study was performed on 5557 nondiabetic individuals who underwent annual physical examination as the training cohort, and a subsequent validation cohort of 1870 individuals was conducted using the same procedures. Multiple logistic regression analysis was performed, and a simple nomogram was constructed via the stepwise method. Receiver operating characteristic (ROC) curve and decision curve analyses were performed by 500 bootstrap resamplings to assess the determination and clinical value of the nomogram, respectively. We also estimated the optimal cutoff values of each risk factor for T2DM prediction.

Results

The 3-year cumulative incidence of T2DM was 10.71%. We developed simple nomograms that predict the risk of T2DM for females and males by using the parameters of age, BMI, fasting blood glucose (FBG), low-density lipoprotein cholesterol (LDLc), high-density lipoprotein cholesterol (HDLc), and triglycerides (TG). In the training cohort, the area under the ROC curve (AUC) showed statistical accuracy (AUC = 0.863 for female, AUC = 0.751 for male), and similar results were shown in the subsequent validation cohort (AUC = 0.847 for female, AUC = 0.755 for male). Decision curve analysis demonstrated the clinical value of this nomogram. To optimally predict the risk of T2DM, the cutoff values of age, BMI, FBG, systolic blood pressure, diastolic blood pressure, total cholesterol, LDLc, HDLc, and TG were 47.5 and 46.5 years, 22.9 and 23.7 kg/m², 5.1 and 5.4 mmol/L, 118 and 123 mmHg, 71 and 85 mmHg, 5.06 and 4.94 mmol/L, 2.63 and 2.54 mmol/L, 1.53 and 1.34 mmol/L, and 1.07 and 1.65 mmol/L for females and males, respectively.

Conclusion

Our nomogram can be used as a simple, plausible, affordable, and widely implementable tool to predict a personalized risk of T2DM for Central Chinese residents. The successful identification of at-risk individuals and intervention at an early stage can provide advanced strategies from a predictive, preventive, and personalized medicine perspective.

Electronic supplementary material

The online version of this article (10.1007/s13167-019-00181-2) contains supplementary material, which is available to authorized users.

Association of type 2 diabetes with anthropometrics, bone mineral density, and body composition in a large-scale screening study of Korean adults

OBJECTIVES

Type 2 diabetes mellitus (T2DM) is a common, chronic disease that is closely associated with anthropometric indices related to obesity. However, no study published to date has simultaneously examined the associations of T2DM with anthropometrics, bone mineral density (BMD), and body composition variables. The present study aimed to evaluate the associations of T2DM with anthropometrics, BMD and body composition variables and to identify the best indicator of T2DM in Korean adults.

METHODS

The data used in this study were obtained from the Korea National Health and Nutrition Examination Survey conducted from 2008 to 2011. A total of 7,835 participants aged from 40 to 90 years were included in this study. A binary logistic regression analysis was performed to examine the significance of differences between the groups with and without T2DM, and the areas under the receiver operating characteristic (AUCs) curves were calculated to compare the predictive power of all variables.

RESULTS

In men, waist-to-height ratio (WHtR) displayed the strongest association with T2DM (adjusted odds ratio (OR) = 1.838 [1.513-2.233], adjusted p<0.001), and waist circumference (WC) and WHtR were the best indicators (WC: AUC = 0.662 [0.639-0.685], WHtR: AUC = 0.680 [0.658-0.703]) of T2DM among all the variables. In women, left leg (LL) and right leg (RL) fat displayed strong negative associations with T2DM (LL fat: adjusted OR = 0.367 [0.321-0.419], adjusted p<0.001, RL fat: adjusted OR = 0.375 [0.329-0.428], adjusted p<0.001), and WC and WHtR were excellent indicators (WC: AUC = 0.730 [0.709-0.750], WHtR: AUC = 0.747 [0.728-0.766]) of T2DM among all the variables. In particular, the WHtR in men and LL and RL fat in women exhibited the strongest associations with T2DM, and the predictive power of the WC and WHtR was stronger than BMD, fat, and muscle mass variables in both men and women. Additionally, the predictive power of the WC and WHtR was stronger in women than in men.

DISCUSSION

Of the anthropometric indices, BMD, and body fat and muscle variables, the best indicators of T2DM were WC and WHtR in both Korean men and women. The results of the present investigation will provide basic information for clinical studies of patients with T2DM and evidence for the prevention and management of T2DM.

A comparison of trunk circumference and width indices for hypertension and type 2 diabetes in a large-scale screening: a retrospective cross-sectional study

Anthropometric indices determine important risk factors for many chronic diseases. However, to date, no study has simultaneously analyzed the capabilities of trunk circumference and width indices to identify hypertension and type 2 diabetes in a large-scale screening study. The objectives of this study were to examine the associations of hypertension and - diabetes with circumference and width indices measured at the five identical positions (axillary, chest, rib, waist, and pelvic) and to compare the capabilities of circumference and width indices to identify the two diseases. Data were obtained from the Korean Health and Genome Epidemiology Study database. The associations and abilities of the circumference indices to identify diabetes were greater than those for hypertension. Overall, trunk circumference indices displayed stronger associations with and greater abilities to identify hypertension and diabetes than did trunk width indices at the five positions. In the comparative analysis between index pairs of circumference and width in patients with diabetes, significant differences were shown at all five positions and in the adjusted analysis of axillary, chest, rib, and pelvic positions. Therefore, width indices should not be used as an alternative indicator of type 2 diabetes in either men or women, except when measured at the waist.

Diabetes and Impaired Fasting Glucose Prediction Using Anthropometric Indices in Adults from Maracaibo City, Venezuela

To determine the predictive power of various anthropometric indices for the identification of dysglycemic states in Maracaibo, Venezuela. A cross-sectional study with randomized, multi-staged sampling was realized in 2230 adult subjects of both genders who had their body mass index (BMI), waist circumference (WC) and waist-height ratio (WHR) determined. Diagnoses of type 2 diabetes mellitus (DM2) and impaired fasting glucose (IFG) were made following ADA 2015 criteria. ROC curves were used to evaluate the predictive power of each anthropometric parameter. Area under the curve (AUC) values were compared through Delong's test. Of the total 2230 individuals (52.6 % females), 8.4 % were found to have DM2, and 19.5 % had IFG. Anthropometric parameters displayed greater predictive power regarding newly diagnosed diabetics, where WHR was the most important predictor in both females (AUC = 0.808; CI 95 % 0.715-0.900. Sensitivity: 82.8 %; specificity: 76.2 %) and males (AUC = 0.809; CI 95 % 0.736-0.882. Sensitivity: 78.6 %; specificity: 68.1 %), although all three parameters appeared to have comparable predictive power in this subset. In previously diagnosed diabetic subjects, WHR was superior to both WC and BMI in females, and WHR and WC were both superior to BMI in males. Lower predictive values were found for IFG in both genders. Accumulation of various altered anthropometric measurements was associated with increased odds ratios for both newly and previously diagnosed DM2. The predictive power of anthropometric measurements was greater for DM2 than IFG. We suggest assessment of as many available parameters as possible in the clinical setting.

Association of peptic ulcer disease with obesity, nutritional components, and blood parameters in the Korean population

Objectives

Peptic ulcer disease (PUD) is a common disorder, but whether an association exists between PUD and anthropometric indicators remains controversial. Furthermore, no studies on the association of PUD with anthropometric indices, blood parameters, and nutritional components have been reported. The aim of this study was to assess associations of anthropometrics, blood parameters, nutritional components, and lifestyle factors with PUD in the Korean population.

Methods

Data were collected from a nationally representative sample of the South Korean population using the Korea National Health and Nutrition Examination Survey. Logistic regression was used to examine associations of anthropometrics, blood parameters and nutritional components among patients with PUD.

Results

Age was the factor most strongly associated with PUD in women (p = <0.0001, odds ratio (OR) = 0.770 [0.683–0.869]) and men (p = <0.0001, OR = 0.715 [0.616–0.831]). In both crude and adjusted analyses, PUD was highly associated with weight (adjusted p = 0.0008, adjusted OR = 1.251 [95%CI: 1.098–1.426]), hip circumference (adjusted p = 0.005, adjusted OR = 1.198 [1.056–1.360]), and body mass index (adjusted p = 0.0001, adjusted OR = 1.303 [1.139–1.490]) in women and hip circumference (adjusted p = 0.0199, adjusted OR = 1.217 [1.031–1.435]) in men. PUD was significantly associated with intake of fiber (adjusted p = 0.0386, adjusted OR = 1.157 [1.008–1.328], vitamin B2 (adjusted p = 0.0477, adjusted OR = 1.155 [1.001–1.333]), sodium (adjusted p = 0.0154, adjusted OR = 1.191 [1.034–1.372]), calcium (adjusted p = 0.0079, adjusted OR = 1.243 [1.059–1.459]), and ash (adjusted p = 0.0468, adjusted OR = 1.152 [1.002–1.325] in women but not in men. None of the assessed blood parameters were associated with PUD in women, and only triglyceride level was associated with PUD in men (adjusted p = 0.0169, adjusted OR = 1.227 [1.037–1.451]).

Discussion

We found that obesity was associated with PUD in the Korean population; additionally, the association between nutritional components and PUD was greater in women than in men.

Comparison of machine-learning algorithms to build a predictive model for detecting undiagnosed diabetes - ELSA-Brasil: accuracy study

CONTEXT AND OBJECTIVE:

Type 2 diabetes is a chronic disease associated with a wide range of serious health complications that have a major impact on overall health. The aims here were to develop and validate predictive models for detecting undiagnosed diabetes using data from the Longitudinal Study of Adult Health (ELSA-Brasil) and to compare the performance of different machine-learning algorithms in this task.

DESIGN AND SETTING:

Comparison of machine-learning algorithms to develop predictive models using data from ELSA-Brasil.

METHODS:

After selecting a subset of 27 candidate variables from the literature, models were built and validated in four sequential steps: (i) parameter tuning with tenfold cross-validation, repeated three times; (ii) automatic variable selection using forward selection, a wrapper strategy with four different machine-learning algorithms and tenfold cross-validation (repeated three times), to evaluate each subset of variables; (iii) error estimation of model parameters with tenfold cross-validation, repeated ten times; and (iv) generalization testing on an independent dataset. The models were created with the following machine-learning algorithms: logistic regression, artificial neural network, naïve Bayes, K-nearest neighbor and random forest.

RESULTS:

The best models were created using artificial neural networks and logistic regression. -These achieved mean areas under the curve of, respectively, 75.24% and 74.98% in the error estimation step and 74.17% and 74.41% in the generalization testing step.

CONCLUSION:

Most of the predictive models produced similar results, and demonstrated the feasibility of identifying individuals with highest probability of having undiagnosed diabetes, through easily-obtained clinical data.

Machine Learning and Data Mining Methods in Diabetes Research

The remarkable advances in biotechnology and health sciences have led to a significant production of data, such as high throughput genetic data and clinical information, generated from large Electronic Health Records (EHRs). To this end, application of machine learning and data mining methods in biosciences is presently, more than ever before, vital and indispensable in efforts to transform intelligently all available information into valuable knowledge. Diabetes mellitus (DM) is defined as a group of metabolic disorders exerting significant pressure on human health worldwide. Extensive research in all aspects of diabetes (diagnosis, etiopathophysiology, therapy, etc.) has led to the generation of huge amounts of data. The aim of the present study is to conduct a systematic review of the applications of machine learning, data mining techniques and tools in the field of diabetes research with respect to a) Prediction and Diagnosis, b) Diabetic Complications, c) Genetic Background and Environment, and e) Health Care and Management with the first category appearing to be the most popular. A wide range of machine learning algorithms were employed. In general, 85% of those used were characterized by supervised learning approaches and 15% by unsupervised ones, and more specifically, association rules. Support vector machines (SVM) arise as the most successful and widely used algorithm. Concerning the type of data, clinical datasets were mainly used. The title applications in the selected articles project the usefulness of extracting valuable knowledge leading to new hypotheses targeting deeper understanding and further investigation in DM.

Identification of Hemoglobin Levels Based on Anthropometric Indices in Elderly Koreans

OBJECTIVES

Anemia is independently and strongly associated with an increased risk of mortality in older people and is also strongly associated with obesity. The objectives of the present study were to examine the associations between the hemoglobin level and various anthropometric indices, to predict low and normal hemoglobin levels using combined anthropometric indices, and to assess differences in the hemoglobin level and anthropometric indices between Korean men and women.

METHODS

A total of 7,156 individuals ranging in age from 53-90 years participated in this retrospective cross-sectional study. Binary logistic regression (LR) and naïve Bayes (NB) models were used to identify significant differences in the anthropometric indices between subjects with low and normal hemoglobin levels and to assess the predictive power of these indices for the hemoglobin level.

RESULTS

Among all of the variables, age displayed the strongest association with the hemoglobin level in both men (p < 0.0001, odds ratio [OR] = 0.487, area under the receiver operating characteristic curve based on the LR [LR-AUC] = 0.702, NB-AUC = 0.701) and women (p < 0.0001, OR = 0.636, LR-AUC = 0.625, NB-AUC = 0.624). Among the anthropometric indices, weight and body mass index (BMI) were the best predictors of the hemoglobin level. The predictive powers of all of the variables were higher in men than in women. The AUC values for the NB-Wrapper and LR-Wrapper predictive models generated using combined anthropometric indices were 0.734 and 0.723, respectively, for men and 0.649 and 0.652, respectively, for women. The use of combined anthropometric indices may improve the predictive power for the hemoglobin level.

DISCUSSION

Among the various anthropometric indices, with the exception of age, we did not identify any indices that were better predictors than weight and BMI for low and normal hemoglobin levels. In addition, none of the ratios between pairs of indices were good indicators of the hemoglobin level. Finally, the Korean men tended to have higher associations between the anthropometric indices and anemia than the women.

Clinical predictors differentiating non-diabetic renal diseases from diabetic nephropathy in a large population of type 2 diabetes patients

AIMS

Non-diabetic renal diseases (NDRDs) are associated with better renal outcomes than diabetic nephropathy (DN). This study was conducted to determine the common clinical markers predicting NDRDs in type 2 diabetes patients.

METHODS

Patients with type 2 diabetes mellitus who underwent a renal biopsy were screened. Eligible patients were categorized into two groups: DN group and NDRD group. Patient's clinical characteristics and laboratory data were collected. Logistic regression analysis was performed to identify risk factors for NDRD development, and the diagnostic performance of these variables was evaluated.

RESULTS

The study included 248 patients, 96 (38.71%) in the DN group and 152 (61.29%) in the NDRD group. Patients in the NDRD group had a shorter duration of DM and higher hemoglobin, estimated glomerular filtration rate, and urine osmotic pressure values as well as a higher incidence of glomerular hematuria than patients in the DN group. In the NDRD patients, the most common pathological type was membranous nephropathy (55, 36.18%). Absence of retinopathy (OR, 44.696, 95% CI, 15.91-125.566), glomerular hematuria (OR, 9.587, 95% CI, 2.027-45.333), and DM history ⩽5years (OR, 4.636, 95% CI, 1.721-12.486) were significant and independent risk factors for the development of NDRD (P<0.01). Absence of retinopathy achieved the overall highest diagnostic efficiency with a sensitivity of 92.11% and specificity of 82.29%. Glomerular hematuria had the highest specificity (93.75%).

CONCLUSION

Shorter duration of diabetes (⩽5years), absence of retinopathy, and presence of glomerular hematuria were independent indicators associated with NDRDs, indicating the need for renal biopsy.

Predictors of metabolic abnormalities in phenotypes that combined anthropometric indices and triglycerides

BACKGROUND

The hypertriglyceridemic waist (HW) phenotype has been shown to be strongly associated with metabolic abnormalities; however, to date, no study has reported the prediction of metabolic abnormalities using the HW phenotype along with waist circumference (WC) and the triglyceride (TG) level or various phenotypes consisting of an individual anthropometric index combined with the TG level. The objectives of this study were to assess the association of the HW phenotype with metabolic abnormalities in Korean women and to evaluate the predictive powers of various phenotypes with regard to metabolic abnormalities.

METHODS

Total cholesterol (TC), high- and low-density lipoprotein (HDL and LDL) cholesterol, and TG levels, systolic and diastolic blood pressures (SBP and DBP), and anthropometric indices were measured in 7661 women. The Naive Bayes algorithm and logistic regression were used to determine the predictive powers of the models using different phenotypes.

RESULTS

The HW phenotype demonstrated the strongest association with all metabolic components. The best phenotypic predictors were the forehead-to-rib circumference ratio + TG for the HDL level, age + TG for the LDL level, age + TG for SBP, and rib circumference + TG and neck circumference + TG for DBP. The associations between TG and TC or HDL were higher compared with those between WC and TC or HDL, whereas the associations between WC and SBP or DBP were higher compared with those between TG and SBP or DBP. Age was strongly associated with hypercholesterolemia, the HDL and LDL cholesterol levels, and SBP and had good predictive power, but not with respect to DBP.

CONCLUSIONS

We have determined that the HW phenotype is a useful indicator of metabolic abnormalities in Korean women; although HW had the strongest association with metabolic abnormalities, the best phenotype combination consisting of a single anthropometric index and the TG level may differ depending on the metabolic factors in question. Our findings provide insights into the detection of metabolic abnormalities in complementary and alternative medicine.

Association of hypertension with physical factors of wrist pulse waves using a computational approach: a pilot study

BACKGROUND

The objectives of this pilot study were to examine the association between hypertension and physical factors of wrist pulse waves to avoid subjective diagnoses in Traditional Chinese Medicine (TCM) and Traditional Korean Medicine (TKM). An additional objective was to assess the predictive power of individual and combined physical factors in order to identify the degree of agreement between diagnosis accuracies using physical factors and using a sphygmomanometer in the prediction of hypertension.

METHODS

In total, 393 women aged 46 to 73 years participated in this study. Logistic regression (LR) and a naïve Bayes algorithm (NB) were used to assess statistically significant differences and the predictive power of hypertension, and a wrapper-based machine learning method was used to evaluate the predictive power of combinations of physical factors.

RESULTS

In both wrists, L-PPI and R-PPI (maximum pulse amplitudes in the left Gwan and right Gwan) were the factors most strongly associated with hypertension after adjusting for age and body mass index (p = <0.001, odds ratio (OR) = 2.006 on the left and p = <0.001, OR = 2.504 on the right), and the best predictors (NB-AUC = 0.692, LR-AUC = 0.7 on the left and NB-AUC = 0.759, LR-AUC = 0.763 on the right). Analyses of both individual and combined physical factors revealed that the predictive power of the physical factors in the right wrist was higher than for the left wrist. The predictive powers of the combined physical factors were higher than those of the best single predictors in both the left and right wrists.

CONCLUSION

We suggested new physical factors related to the sum of the area on the particular region of pulse waves in both wrists. L-PPI and R-PPI among all variables used in this study were good indicators of hypertension. Our findings support the quantification and objectification of pulse patterns and disease in TCM and TKM for complementary and alternative medicine.

Identification of Type 2 Diabetes Risk Factors Using Phenotypes Consisting of Anthropometry and Triglycerides based on Machine Learning

The hypertriglyceridemic waist (HW) phenotype is strongly associated with type 2 diabetes; however, to date, no study has assessed the predictive power of phenotypes based on individual anthropometric measurements and triglyceride (TG) levels. The aims of the present study were to assess the association between the HW phenotype and type 2 diabetes in Korean adults and to evaluate the predictive power of various phenotypes consisting of combinations of individual anthropometric measurements and TG levels. Between November 2006 and August 2013, 11,937 subjects participated in this retrospective cross-sectional study. We measured fasting plasma glucose and TG levels and performed anthropometric measurements. We employed binary logistic regression (LR) to examine statistically significant differences between normal subjects and those with type 2 diabetes using HW and individual anthropometric measurements. For more reliable prediction results, two machine learning algorithms, naive Bayes (NB) and LR, were used to evaluate the predictive power of various phenotypes. All prediction experiments were performed using a tenfold cross validation method. Among all of the variables, the presence of HW was most strongly associated with type 2 diabetes (p < 0.001, adjusted odds ratio (OR) = 2.07 [95% CI, 1.72-2.49] in men; p < 0.001, adjusted OR = 2.09 [1.79-2.45] in women). When comparing waist circumference (WC) and TG levels as components of the HW phenotype, the association between WC and type 2 diabetes was greater than the association between TG and type 2 diabetes. The phenotypes tended to have higher predictive power in women than in men. Among the phenotypes, the best predictors of type 2 diabetes were waist-to-hip ratio + TG in men (AUC by NB = 0.653, AUC by LR = 0.661) and rib-to-hip ratio + TG in women (AUC by NB = 0.73, AUC by LR = 0.735). Although the presence of HW demonstrated the strongest association with type 2 diabetes, the predictive power of the combined measurements of the actual WC and TG values may not be the best manner of predicting type 2 diabetes. Our findings may provide clinical information concerning the development of clinical decision support systems for the initial screening of type 2 diabetes.

Indicators of hypertriglyceridemia from anthropometric measures based on data mining

BACKGROUND

The best indicator for the prediction of hypertriglyceridemia derived from anthropometric measures of body shape remains a matter of debate. The objectives are to determine the strongest predictor of hypertriglyceridemia from anthropometric measures and to investigate whether a combination of measures can improve the prediction accuracy compared with individual measures.

METHODS

A total of 5517 subjects aged 20-90 years participated in this study. The numbers of normal and hypertriglyceridemia subjects were 3022 and 653 females, respectively, and 1306 and 536 males, respectively. We evaluated 33 anthropometric measures for the prediction of hypertriglyceridemia using statistical analysis and data mining.

RESULTS

In the 20-90-year-old groups, age in women was the variable that exhibited the highest predictive power; however, this was not the case in men in all age groups. Of the anthropometric measures, the waist-to-height ratio (WHtR) was the best predictor of hypertriglyceridemia in women. In men, the rib-to-forehead circumference ratio (RFcR) was the strongest indicator. The use of a combination of measures provides better predictive power compared with individual measures in both women and men. However, in the subgroups of ages 20-50 and 51-90 years, the strongest indicators for hypertriglyceridemia were rib circumference in the 20-50-year-old group and WHtR in the 51-90-year-old group in women and RFcR in the 20-50-year-old group and BMI in the 51-90-year-old group in men.

CONCLUSIONS

Our results demonstrated that the best predictor of hypertriglyceridemia may differ according to gender and age.

Predicting visceral obesity based on facial characteristics

Background

Visceral obesity is associated with facial characteristics and chronic disease, but no studies on the best predictor of visceral obesity based on facial characteristics have been reported. The aims of the present study were to investigate the association of visceral obesity with facial characteristics, to determine the best predictor of normal waist and visceral obesity among these characteristics, and to compare the predictive power of individual and combined characteristics.

Methods

Cross-sectional data were obtained from 11347 adult Korean men and women ranging from 18 to 80 years old. We examined 15 facial characteristics to identify the strongest predictor of normal and viscerally obese subjects and assessed the predictive power of the combined characteristics.

Results

FD_94_194 (the distance between both inferior ear lobes) was the best indicator of the normal and viscerally obese subjects in the following groups: Men-18-50 (p ≤ 0.0001, OR = 4.610, AUC = 0.821), Men-50-80 (p ≤ 0.0001, OR = 2.624, AUC = 0.735), and Women-18-50 (p ≤ 0.0001, OR = 2.979, AUC = 0.76). In contrast, FD_43_143 (mandibular width) was the strongest predictor in Women-50-80 (p ≤ 0.0001, OR = 2.099, AUC = 0.679). In a comparison of the combined characteristics, the area under the receiver operating characteristic curve (AUC) and the kappa values of the 4 groups ranged from 0.826 to 0.702 and from 0.483 to 0.279, respectively. The model for Men-18-50 showed the strongest predictive values and the model for Women-51-80 had the lowest predictive value for both the individual and combined characteristics.

Conclusions

In both men and women, the predictive power of the young and middle-age groups was better than that of the elderly groups for predicting normal waist and viscerally obese subjects for both the individual and combined characteristics. The predictive power appeared to increase slightly with the combined characteristics.

A comparison of the predictive power of anthropometric indices for hypertension and hypotension risk

BACKGROUND AND AIMS

It is commonly accepted that body fat distribution is associated with hypertension, but the strongest anthropometric indicator of the risk of hypertension is still controversial. Furthermore, no studies on the association of hypotension with anthropometric indices have been reported. The objectives of the present study were to determine the best predictors of hypertension and hypotension among various anthropometric indices and to assess the use of combined indices as a method of improving the predictive power in adult Korean women and men.

METHODS

For 12789 subjects 21-85 years of age, we assessed 41 anthropometric indices using statistical analyses and data mining techniques to determine their ability to discriminate between hypertension and normotension as well as between hypotension and normotension. We evaluated the predictive power of combined indices using two machine learning algorithms and two variable subset selection techniques.

RESULTS

The best indicator for predicting hypertension was rib circumference in both women (p = <0.0001; OR = 1.813; AUC = 0.669) and men (p = <0.0001; OR = 1.601; AUC = 0.627); for hypotension, the strongest predictor was chest circumference in women (p = <0.0001; OR = 0.541; AUC = 0.657) and neck circumference in men (p = <0.0001; OR = 0.522; AUC = 0.672). In experiments using combined indices, the areas under the receiver operating characteristic curves (AUC) for the prediction of hypertension risk in women and men were 0.721 and 0.652, respectively, according to the logistic regression with wrapper-based variable selection; for hypotension, the corresponding values were 0.675 in women and 0.737 in men, according to the naïve Bayes with wrapper-based variable selection.

CONCLUSIONS

The best indicators of the risk of hypertension and the risk of hypotension may differ. The use of combined indices seems to slightly improve the predictive power for both hypertension and hypotension.