Effect of visceral fat on onset of metabolic syndrome

Effect of visceral fat on onset of metabolic syndrome | 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 Article Effect of visceral fat on onset of metabolic syndrome Hiroto Bushita, Naoki Ozato, Kenta Mori, Hiromitsu Kawada, Yoshihisa Katsuragi, and 6 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-3996594/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 Metabolic syndrome (MetS) increases the risk of cardiovascular and lifestyle-related diseases. Therefore, early detection is important to prevent MetS. This study analysed the effects of visceral fat on MetS using health examination. A MetS onset prediction algorithm was developed. Methods Health examination data were obtained from the Iwaki Health Promotion Project conducted in Aomori Prefecture in Japan, wherein labels indicated the development of MetS within the three years (213 onset and 1320 non-onset cases). The data were divided into training and test data (8:2 ratio), and 18 onset prediction models were developed to support the SHapley Additive exPlanations (SHAP) value. The onset labels and non-invasive input data were used as the output and input variables, respectively. We selected the model with the highest area under the curve (AUC) score when conducting five-fold cross validation, and the AUC of the test data was calculated. Feature impact was calculated based on SHAP. Results There were 169 and 1058 people in the metabolic and non-metabolic syndrome groups, respectively. The visceral fat area was significantly higher in the onset group than in the non-onset group (p < 0.00001). The cut-off value based on the receiver operating characteristic curve was 82 cm 2 , and the AUC was 0.86. Machine learning was employed on six items reported to contribute to the onset of MetS in addition to visceral fat to build an onset prediction algorithm. The cross-validation AUC = 0.90 and test AUC = 0.88 indicated a high-accuracy algorithm. The visceral fat was found to be the main factor, as confirmed by conventional feature importance in machine learning. Conclusions Visceral fat is crucial to determining the onset of MetS in the future. A high-accuracy onset prediction algorithm was developed based on non-invasive parameters, including visceral fat. Health sciences/Health care/Disease prevention/Preventive medicine Health sciences/Medical research/Epidemiology Figures Figure 1 Figure 2 Figure 3 Figure 4 Introduction Cardiometabolic diseases, including metabolic syndrome (MetS), have become the leading cause of death worldwide in the past 20 years [ 1 ]. MetS involves several risk factors such as abdominal obesity, hyperglycemia, hypertension, and dyslipidaemia and is becoming a serious problem worldwide [ 2 , 3 , 4 ]. In 2020, approximately 25.8 million children and 35.5 million adolescents worldwide were affected by MetS [ 5 ]. Therefore, the prevention and mitigation of MetS is an urgent issue. Various preventive approaches have been proposed (exercise guidance [ 6 ] and dietary guidance [ 7 , 8 , 9 ]). On the other hand, it has been reported that behavioural changes can be expected by knowing the risk of disease [ 10 ]. Therefore, the ability to predict the onset of metabolic syndrome years in advance is considered to be one of the most useful tools for preventing the development of MetS. Several algorithms for predicting the onset of MetS have been developed [ 11 , 12 , 13 , 14 , 15 , 16 , 17 , 18 , 19 , 20 , 21 , 22 ]. However, the definition of MetS differs in different country [ 23 ]. In Japan, visceral fat accumulation is essential for diagnosis, with a cut-off value of 100 cm 2 [ 24 , 25 ]. One reason for this is that Asians are more susceptible to MetS than Westerners, and even those with a low BMI accumulate visceral fat [ 26 , 27 ]. Thus, accumulating research knowledge in accordance with each country’s standards is important. The relationship between visceral fat and MetS has been the subject of several cross-sectional studies in Japan, China, South Korea, and the United States [ 28 , 29 , 30 , 31 , 32 , 33 ]. A 3.3-year longitudinal study in the United States [ 34 ] and a 17-month longitudinal study in South Korea [ 35 ] reported that visceral fat increases MetS onset risk as defined by the waist size. A 3-year longitudinal study in Japan reported that an increase in visceral fat is correlated with MetS risk factors [ 36 ]. However, no large-scale studies have been conducted in Japan measuring longitudinal changes in visceral fat area (VFA) in the same subjects over a period of 1–6 years. Moreover, the relationship between MetS onset risk and visceral fat remains relatively unexplored. We have been collecting visceral fat-related health data since 2015. This study used the data for the period 2015–2020 to clarify the relationship between visceral fat and MetS onset. An algorithm to predicts MetS onset was developed. Methods Dataset construction The Iwaki Health Promotion Project [ 37 ] was a health examination for residents aged 20 years and older conducted in Hirosaki City, Aomori Prefecture, Japan. A variety of data were collected in this project, including those on genes, physical characteristics, behavioural habits, and intestinal bacteria. Visceral fat data were collected from 2015 onwards, and this study used data for the period 2015–2020. The health examination data included visceral fat area (VFA), blood data, dietary data, and interview data. VFA was estimated by the abdominal bio-impedance method using EW-FA90 (Panasonic Corporation), an approved medical device in Japan (No. 22500BZX00522000) [ 38 ]. Measurement values from this device have been reported to have a strong correlation with those obtained using computed tomography (CT), which is the gold standard for measuring visceral fat [ 39 ]. Other basic blood parameters were obtained as previously reported [ 40 , 41 , 42 , 43 ]. We have previously used the data from the present study to conduct research on diet [ 40 , 43 ], exercise [ 41 ], and intestinal bacteria [ 42 ]. This study was approved by the Hirosaki University Graduate School of Medicine ethics committee (approval number: 2014-377-1, 2016-028-1, 2021-030, 2018-012, 2020-046-4, 2020-046-1) and conducted in accordance with the recommendations set out in the Declaration of Helsinki. All participants provided written informed consent. MetS onset criteria The following were used as MetS diagnosis criteria. The presence or absence of MetS was determined for each participant at each time point in the dataset [ 24 ]. Required criteria: “VFA ≥ 100 cm 2 (for both males and females)” One of the following conditions apply: (1) Triglycerides (TG) ≥ 150 mg/dL, high-density lipoprotein cholesterol (HDLC) < 50 mg/dL, or currently receiving medication for that disease; or all three (2) Systolic blood pressure (SBP) ≥ 130 mmHg, diastolic blood pressure (DBP) ≥ 85 mmHg, or taking antihypertensive medication; or all three (3) Blood sugar (BS) ≥ 110 mg/dL, receiving antidiabetic therapy, or both Finally, from the data compiled from six years of Iwaki Health Promotion Project (n = 5905), we excluded those participants who were not assigned a label within three years (n = 3870) and those who currently fell under the category of MetS (n = 502) to create the final dataset (n = 1533). Data pre-processing We used the DataRobot AI Platform (ver.9.1, DataRobot) to convert the numerical and categorical variables into the format that was most appropriate for each model, including standardisation, missing value imputation, one-hot encoding, rigid transformation, and ordinal encoding. Dataset creation and analysis method The dataset (n = 1533) was randomly divided into training data for five-fold cross validation (80%, n = 1227) and test data (20%, n = 306) using DataRobot. When dividing the data, we used “group” partition and set “individual ID” as the group ID. This avoids leakage, as data from the same lot is not spread across training, validation, and test data. We computed the accuracy and stability of each model using five-fold cross validation to select the optimal model. We used the test data to estimate model prediction accuracy for unknown data not used in model creation. The data for cross validation were used to confirm the effect of visceral fat on MetS onset. Differences between each parameter of the MetS onset and non-onset groups were analysed using R software version 4.3.1, and Fisher’s exact and Mann–Whitney U (corrected for ties) tests were applied for the categorical and continuous variables, respectively. We evaluated the effect of visceral fat on the onset of MetS as follows. We used box plots, the exact Mann–Whitney U test, and logistic regression analysis adjustments for confounding factors (gender, age, number of cigarettes smoked, alcohol intake, and exercise intensity) to obtain receiver operating characteristic (ROC) curves using R and calculated the VFA cut-off value for MetS onset risk based on the Youden index [ 44 ] or the ROC curve closest to (0,1) [ 45 , 46 ]. Next, we built an MetS onset prediction model using the cross-validation data. First, we calculated the area under the curve (AUC) using VFA as the only parameter. We then considered seven additional parameters previously reported to be associated with MetS (gender, age, BMI, systolic blood pressure, diastolic blood pressure, number of drinking days per week, and number of cigarettes smoked) [ 11 ] for a total of eight input features in the prediction model. Machine learning was employed using DataRobot to estimate their effect on prediction accuracy. The model with the highest prediction accuracy based on the AUC score was chosen from the 18 models. The chosen model was used for MetS onset risk predictions on the test dataset and calculate the AUC score. The 18 models included general linear and classification models, including XGBoost, Elastic-Net, LightGBM, logistic regression, and residual neural networks. We used an AUC difference test (DeLong’s test [ 47 ]) for the AUC calculated from this analysis. For each model, we analysed the difference between the cross-validation and test AUCs. We visualised the contribution of each of the eight features to the MetS onset risk, using the SHapley Additive exPlanations (SHAP) [ 48 ] function of DataRobot. First, we used SHAP to visualise the effect of fluctuations in a given feature on the onset risk of each individual. Next, we calculated the absolute value of the degree of influence of each feature in each individual. The relative magnitude of the averaged value was compared and displayed as the feature impact. This facilitated ranking the features according to their contribution to MetS onset risk. Results Physical characteristics In the data for cross validation (n = 1227), the onset and non-onset groups comprised 169 and 1058 cases, respectively. In the test data (n = 306), the onset and non-onset groups comprised 44 and 262 cases, respectively. The MetS incidence in each dataset was 13.8% and 14.4%, respectively. Table 1 shows the subject background information at the baseline in the cross-validation data. In the onset group, 36.7% were females and 63.3% were males, and in the non-onset group, 67.1% were females and 32.9% were males. Onset and gender were significantly related (p < 0.001). The median age in the onset group was 57 years, which was significantly higher than that in the non-onset group (52 years). Compared with the non-onset group, the onset group had a significantly greater height, and a significantly larger BMI, VFA, and waist circumference (p < 0.001). In addition, Triglycerides (TG), high-density lipoprotein cholesterol (HDLC), low-density lipoprotein cholesterol (LDLC), systolic blood pressure (SBP), diastolic blood pressure (DBP), Blood sugar (BS), and haemoglobin A1c (HbA1c), which are factors involved in MetS, were significantly higher (p < 0.001) in the onset group. Energy intake (BDHQ), alcohol intake (BDHQ, after adjusting for energy), number of cigarettes smoked, and sleeping time were significantly longer (p < 0.001, p = 0.003, p = 0.008, p = 0.003) ) in the onset group. No significant differences were observed between the two groups with respect to number of drinking days and exercise intensity (p = 0.654, p = 0.399). Effect of visceral fat on MetS onset The VFA at baseline was compared between the MetS onset and non-onset groups. In the onset group, 45% were VFA < 100 cm 2 cases and 55% were VFA ≥ 100 cm 2 cases. In the non-onset group, 86.1% were VFA < 100 cm 2 cases and 13.9% were VFA ≥ 100 cm 2 cases. The VFA quartile values in the onset group were as follows: 25%, 92 cm 2 ; 50%, 105 cm 2 ; and 75%, 130 cm 2 . In the non-onset group, the values were as follows: 25%, 42 cm 2 ; 50%, 63 cm 2 ; and 75%, 85 cm 2 , with the VFA being significantly higher in the onset group (Fig. 2 A; p < 0.001). Logistic regression analysis confirmed that there was no change in these results even after adjusting for gender, age, number of cigarettes smoked, alcohol intake, and exercise intensity (p < 0.001). AUC for the ROC curve was 0.8597. The cut-off values based on the Youden index and the ROC curve closest to (0,1) were 82.5 cm 2 (Fig. 2 B) and 85.5 cm 2 , respectively. These results suggest that baseline VFA alone can predict the onset of MetS. It was confirmed that 45% of the non-onset group did not experience the onset of MetS despite their baseline VFA exceeding 100 cm 2 . Construction and validation of MetS onset prediction model The baseline VFA was determined to be an important factor for the prediction of MetS onset; thus, a model for predicting MetS onset was constructed using baseline visceral fat. First, 18 different machine learning models compatible with SHAP were constructed using only baseline VFA as the input parameter, and the prediction model with the highest prediction accuracy was selected. The prediction model constructed using Elastic-Net yielded a cross-validation AUC of 0.8591 and a test AUC of 0.8686 (Fig. 3 , Supplemental Table 1). Supplemental Fig. 1 shows the flowchart of model construction and presents the hyperparameters optimised using grid search. A previous study reported on a MetS onset prediction model that was constructed using BMI, number of cigarettes smoked, gender, age, DBP, SBP, and other factors. Previous epidemiological research has suggested that these are important factors [ 11 , 13 , 49 , 50 , 51 ]. Therefore, we built an onset prediction model based on visceral fat and seven parameters from the literature [ 11 ]. First, we constructed 18 different machine learning models using these eight input parameters, and selected the model with the highest prediction accuracy. Results confirm the accuracy of the model (cross-validation AUC = 0.8992, test AUC = 0.8845) without overfitting (Model 1). Next, we ranked the parameters based on their SHAP values. We then constructed 18 models with seven input parameters after excluding the lowest ranked parameter (number of drinking days). We selected the model with the highest prediction accuracy (Model 2). This process was repeated until only one parameter remained (Model 3). Results showed that Model 2, constructed the seven input features: visceral fat, BMI, number of cigarettes smoked, gender, age, DBP, and SBP, and trained with LightGBM, had the highest prediction accuracy (cross-validation AUC = 0.9004, test AUC = 0.8836), with no overfitting (Fig. 3 , Supplemental Table 1)). Supplemental Fig. 2 shows the flowchart for model construction and hyperparameter optimisation using grid search. The onset and non-onset cases were correctly determined with accuracies of 82% and 84%, respectively, when the Matthews correlation coefficient (MCC) [ 52 ] was maximised (minimising false positives and false negatives). The correct answer rate was 84%. Replacing VFA with waist circumference in the VFA-only model (Model 3) and the optimised model that included VFA (Model 2) significantly decreased the prediction accuracy (Supplemental Figs. 3 and 4). Therefore, VFA had a greater contribution to the prediction accuracy than waist circumference when predicting MetS onset. In the optimised MetS onset prediction model (Model 2), the SHAP value of each individual was calculated (Fig. 4 ) as the feature impact based on the SHAP value. An examination of the influence of each item (gender, age, VFA, BMI, DBP, SBP, and number of cigarettes smoked) on MetS onset showed that VFA was the largest contributor to the prediction of MetS onset. BMI was the second most influential factor, but its feature effect was approximately half that of VFA (Fig. 4 ). Discussion Diet and exercise management are important in preventing MetS [ 53 ]; however, predicting the onset is also an important approach. In Japan, visceral fat level is an essential parameter in the diagnosis of MetS [ 24 ]. The gold standard measurement method for visceral fat is CT. However, previous studies on MetS prediction have been conducted without obtaining data on visceral fat [ 11 , 12 , 13 , 14 , 15 , 16 , 17 , 18 , 19 , 20 , 21 , 22 ] because of the time-consuming and invasive process. We built a device that measures visceral fat non-invasively and constructed human dataset that includes VFA between 2015–2020. We investigated the relationship between MetS onset and visceral fat, and developed a MetS onset prediction algorithm. The dataset comprised 169 and 1058 cases in the MetS onset and non-onset groups, respectively. The percentage of cases affected by visceral fat onset within three years was 13.8%. Analysis of the baseline VFA and MetS onset using a box plot showed that the VFA quartiles in the onset group were 25%: 92 cm 2 ; and 75%, 130 cm 2 , and the VFA quartiles in the non-onset group were 25%: 42 cm 2 ; and 75%, 85 cm 2 ; showing that the data overlap was quite small (Fig. 2 A). These results indicated a large difference in the initial visceral fat amount between the onset and non-onset cases. The ROC AUC of 0.8597 suggested that the baseline visceral fat was a strong factor influencing the prediction of MetS onset. The MetS cut-off value was found to be 82.5 cm 2 . Many cross-sectional studies have reported that visceral fat is associated with MetS onset [ 54 , 55 ]. Regarding cut-off values of VFA for MetS, a cross-sectional study of type II diabetes patients aged 18–75 years in China reported values of 84.7 cm 2 in males and 81.1 cm 2 in females [ 31 ]. A cross-sectional study of Chinese patients aged 35–75 years reported cut-off values of 79.2 cm 2 for both males and females [ 29 ]. A longitudinal study reported values of 84 cm 2 and 58 cm 2 in Korean males and females, respectively [ 35 ]. In the present study, the MetS cut-off value was 82.5 cm 2 for both males and females (Fig. 2 B), and the above previous studies suggest that our study was valid. MetS is recognised worldwide as a useful indicator for predicting cardiovascular risk [ 56 ], and the present study confirmed that controlling visceral fat was important in preventing MetS in Japanese people. We built an algorithm for predicting MetS onset using visceral fat. There are multiple algorithms available worldwide for predicting MetS onset [ 11 , 12 , 13 , 14 , 15 , 16 , 17 , 18 , 19 , 20 , 21 , 22 ]. One challenge in predicting MetS onset is that the diagnostic definition of MetS differs in each country. In Japan, despite the fact that visceral fat is an essential parameter, no algorithm for predicting disease onset using visceral fat values has been constructed thus far. Therefore, in this study, we developed an onset prediction algorithm using visceral fat measurements as the input parameter. Examination results confirmed the successful construction of an onset prediction algorithm using six parameters in addition to visceral fat (Model 2) (AUC = 0.9004 in cross validation data). On the test data, the model yielded an AUC of 0.8836 (Fig. 3 ). The feature impact analysis showed visceral fat to be a dominant contributor (Fig. 4 ). There have previously been many onset prediction algorithms that use blood items [ 12 , 13 , 14 , 15 , 16 , 17 , 18 , 19 , 20 , 21 , 22 ]. The visceral fat meter developed in this study is a non-invasive device. The algorithm developed in this study demonstrated that MetS onset can be predicted based on non-invasively measured parameters such as visceral fat, BMI, number of cigarettes smoked, gender, age, and DBP. Although this device has previously been used for visceral fat measurements, the findings of this study may result in the expansion of the functionality of the device. The MetS algorithm uses measurements of only non-invasive parameters and has high medical interpretability; thus, it is expected to facilitate easy and convincing understanding of onset risk. Moreover, it can be used in a variety of applications. Algorithm problems often involve regional differences; however, because the developed algorithm uses parameters with high medical interpretability, there may be fewer validation tests required. Medical expenses for obese people with visceral fat are higher than that for obese people without visceral fat [ 57 ], and the algorithm developed in this study can guide people to reduce their visceral fat before MetS onset. In this study, the prediction model with six parameters associated with MetS (BMI, number of cigarettes smoked, gender, age, DBP, and SBP) in addition to VFA (Model 2) exhibited a high prediction accuracy (cross-validation AUC = 0.9004, test AUC = 0.8836). The prediction accuracy significantly decreased when replacing VFA with waist circumference in Model 2 (Supplemental Fig. 4). Therefore, it was inferred that VFA was a more important factor than waist circumference when predicting MetS onset. Previous research shows only a weak correlation between VFA and waist circumference (males: r = 0.68, females: r = 0.65), with considerable individual-level variance between the VFA and waist circumference readings. For example, in the Japanese population, males with a waist circumference of 85.0–86.0 cm have a VFA of 67–137 cm 2 [ 25 ]. We obtained six years of data on visceral fat, and built a MetS onset prediction model to determine whether onset would occur within three years of the baseline time. Previous research did not measure visceral fat, and the models generally include blood data, which can be an obstacle in daily monitoring. Therefore, the present study added medically important evidence after accurately examining a prediction model with an eye toward social implementation. The present study has several limitations. The dataset used to build the prediction model was limited to participants of a specific race and region; thus, the model’s performance must be checked by including participants from different races and regions. The MetS onset risk prediction model is a just a guidance tool, and when implemented, people must be asked to change their lifestyles and later confirm whether there is an actual decrease in the incidence of MetS. Conclusions We built a six-year medical dataset that included visceral fat measurements. Visceral fat was found to be an important factor for determining the onset of MetS in the future. We developed a high-accuracy onset prediction algorithm using non-invasive parameters, including visceral fat. Declarations Competing interests This study was supported by JST Grant Number JPMJCE1302, JPMJCA2201, JPMJPF2210, and Kao Co. (Tokyo, Japan). Authors HB, NOzato, KMori, HK, YK, and NOsaki were employed by Kao Corporation (Tokyo, Japan). All other authors declare no potential competing interests. The results of the study are presented clearly, honestly, and without fabrication, falsification, or inappropriate data manipulation. Acknowledgements The authors thank all participants in the Iwaki Health Promotion Project and the entire staff of the project. The authors are thankful to Takuji Yasukawa, Takuro Iwane, and Yoshikuni Sugimura for valuable assistance, constant support, and advice in the preparation of this manuscript. The authors are grateful to Daisuke Kasuga, Eiichiro Uchino, and Yoko Sugiura for building the research environment. Author Contributions HB designed the study, acquired, analysed and interpreted the data. NOzato contributed to data acquisition. All authors reviewed and edited the manuscript and approved its final version. Data availability Data cannot be shared publicly because of the ethical concerns. Data are available from the Hirosaki University COI Institutional Data Access / Ethics Committee (contact via e-mail: [email protected] ) for researchers who meet the criteria for access to the data. Researchers need to be approved by research ethics review board at the organization of their affiliation. References Geneva WHO. Global health estimates 2020: Deaths by cause, age, sex, by country and by region, 2000–2019. 2020 Saklayen MG. The Global Epidemic of the Metabolic Syndrome. Curr Hypertens Rep. 2018;20(2):12. Published 2018 Feb 26. doi: 10.1007/s11906-018-0812-z Shin S, Jee H. Prevalence of metabolic syndrome in the Gulf Cooperation Council countries: meta-analysis of cross-sectional studies. J Exerc Rehabil. 2020;16(1):27–35. Published 2020 Feb 26. doi: 10.12965/jer.1938758.379 Prasun P. Mitochondrial dysfunction in metabolic syndrome. Biochim Biophys Acta Mol Basis Dis. 2020;1866(10):165838. doi: 10.1016/j.bbadis.2020.165838 Noubiap JJ, Nansseu JR, Lontchi-Yimagou E, et al. Global, regional, and country estimates of metabolic syndrome burden in children and adolescents in 2020: a systematic review and modelling analysis. Lancet Child Adolesc Health. 2022;6(3):158–170. doi: 10.1016/S2352-4642(21)00374-6 Cho ER, Shin A, Kim J, Jee SH, Sung J. Leisure-time physical activity is associated with a reduced risk for metabolic syndrome. Ann Epidemiol. 2009;19(11):784–792. doi: 10.1016/j.annepidem.2009.06.010 Min C, Noh H, Kang YS, et al. Skipping breakfast is associated with diet quality and metabolic syndrome risk factors of adults. Nutr Res Pract. 2011;5(5):455–463. doi: 10.4162/nrp.2011.5.5.455 Shin A, Lim SY, Sung J, Shin HR, Kim J. Dietary intake, eating habits, and metabolic syndrome in Korean men. J Am Diet Assoc. 2009;109(4):633–640. doi: 10.1016/j.jada.2008.12.015 Bianchi C, Penno G, Daniele G, Benzi L, Del Prato S, Miccoli R. Optimizing management of metabolic syndrome to reduce risk: focus on life-style. Intern Emerg Med. 2008;3(2):87–98. doi: 10.1007/s11739-008-0122-6 Janz NK, Becker MH. The Health Belief Model: a decade later. Health Educ Q. 1984;11(1):1–47. doi: 10.1177/109019818401100101 Salim AA, Kawasoe S, Kubozono T, et al. Development of predictive equation and score for 5-year metabolic syndrome incidence in Japanese adults. PLoS One. 2023;18(4):e0284139. Published 2023 Apr 7. doi: 10.1371/journal.pone.0284139 Zahedi AS, Daneshpour MS, Akbarzadeh M, Hedayati M, Azizi F, Zarkesh M. Association of baseline and changes in adiponectin, homocysteine, high-sensitivity C-reactive protein, interleukin-6, and interleukin-10 levels and metabolic syndrome incidence: Tehran lipid and glucose study. Heliyon. 2023;9(9):e19911. Published 2023 Sep 6. doi: 10.1016/j.heliyon.2023.e19911 Zou TT, Zhou YJ, Zhou XD, et al. MetS Risk Score: A Clear Scoring Model to Predict a 3-Year Risk for Metabolic Syndrome. Horm Metab Res. 2018;50(9):683–689. doi: 10.1055/a-0677-2720 Karimi-Alavijeh F, Jalili S, Sadeghi M. Predicting metabolic syndrome using decision tree and support vector machine methods. ARYA Atheroscler. 2016;12(3):146–152. Lee S, Lee H, Choi JR, Koh SB. Development and Validation of Prediction Model for Risk Reduction of Metabolic Syndrome by Body Weight Control: A Prospective Population-based Study. Sci Rep. 2020;10(1):10006. Published 2020 Jun 19. doi: 10.1038/s41598-020-67238-5 Szabo de Edelenyi F, Goumidi L, Bertrais S, et al. Prediction of the metabolic syndrome status based on dietary and genetic parameters, using Random Forest. Genes Nutr. 2008;3(3–4):173–176. doi: 10.1007/s12263-008-0097-y Lee S, Lee SK, Kim JY, Cho N, Shin C. Sasang constitutional types for the risk prediction of metabolic syndrome: a 14-year longitudinal prospective cohort study. BMC Complement Altern Med. 2017;17(1):438. Published 2017 Sep 2. doi: 10.1186/s12906-017-1936-4 Li G, Esangbedo IC, Xu L, et al. Childhood retinol-binding protein 4 (RBP4) levels predicting the 10-year risk of insulin resistance and metabolic syndrome: the BCAMS study. Cardiovasc Diabetol. 2018;17(1):69. Published 2018 May 14. doi: 10.1186/s12933-018-0707-y Yang H, Yu B, OUYang P, et al. Machine learning-aided risk prediction for metabolic syndrome based on 3 years study. Sci Rep. 2022;12(1):2248. Published 2022 Feb 10. doi: 10.1038/s41598-022-06235-2 Daniel Tavares L, Manoel A, Henrique Rizzi Donato T, et al. Prediction of metabolic syndrome: A machine learning approach to help primary prevention. Diabetes Res Clin Pract. 2022;191:110047. doi: 10.1016/j.diabres.2022.110047 Hirose H, Takayama T, Hozawa S, Hibi T, Saito I. Prediction of metabolic syndrome using artificial neural network system based on clinical data including insulin resistance index and serum adiponectin. Comput Biol Med. 2011;41(11):1051–1056. doi: 10.1016/j.compbiomed.2011.09.005 Liu W, Tang X, Cui T, Zhao H, Song G. Development and visualization of a risk prediction model for metabolic syndrome: a longitudinal cohort study based on health check-up data in China. Front Nutr. 2023;10:1286654. Published 2023 Nov 21. doi: 10.3389/fnut.2023.1286654 Alberti KG, Eckel RH, Grundy SM, et al. Harmonizing the metabolic syndrome: a joint interim statement of the International Diabetes Federation Task Force on Epidemiology and Prevention; National Heart, Lung, and Blood Institute; American Heart Association; World Heart Federation; International Atherosclerosis Society; and International Association for the Study of Obesity. Circulation. 2009;120(16):1640–1645. doi: 10.1161/CIRCULATIONAHA.109.192644 Matsuzawa Y. Metabolic syndrome–definition and diagnostic criteria in Japan. J Atheroscler Thromb. 2005;12(6):301. doi: 10.5551/jat.12.301 Examination Committee of Criteria for 'Obesity Disease' in Japan; Japan Society for the Study of Obesity. New criteria for 'obesity disease' in Japan. Circ J. 2002;66(11):987–992. doi: 10.1253/circj.66.987 Nyamdorj R, Pitkäniemi J, Tuomilehto J, et al. Ethnic comparison of the association of undiagnosed diabetes with obesity [published correction appears in Int J Obes (Lond). 2010;34(3):597] [published correction appears in Int J Obes (Lond). 2011;35(2):313-4]. Int J Obes (Lond). 2010;34(2):332–339. doi: 10.1038/ijo.2009.225 Nishizawa H, Shimomura I. Population Approaches Targeting Metabolic Syndrome Focusing on Japanese Trials. Nutrients. 2019;11(6):1430. Published 2019 Jun 25. doi: 10.3390/nu11061430 Oka R, Kobayashi J, Yagi K, et al. Reassessment of the cutoff values of waist circumference and visceral fat area for identifying Japanese subjects at risk for the metabolic syndrome. Diabetes Res Clin Pract. 2008;79(3):474–481. doi: 10.1016/j.diabres.2007.10.016 Bao Y, Lu J, Wang C, et al. Optimal waist circumference cutoffs for abdominal obesity in Chinese. Atherosclerosis. 2008;201(2):378–384. doi: 10.1016/j.atherosclerosis.2008.03.001 Kim JA, Choi CJ, Yum KS. Cut-off values of visceral fat area and waist circumference: diagnostic criteria for abdominal obesity in a Korean population. J Korean Med Sci. 2006;21(6):1048–1053. doi: 10.3346/jkms.2006.21.6.1048 Yang X, Lin Y, Xu GD, et al. Optimal Cut-Off Values of Visceral Fat Area for Predicting Metabolic Syndrome Among Type 2 Diabetes Patients in Ningbo, China. Diabetes Metab Syndr Obes. 2021;14:1375–1383. Published 2021 Mar 25. doi: 10.2147/DMSO.S304164 Kim SH, Chung JH, Song SW, Jung WS, Lee YA, Kim HN. Relationship between deep subcutaneous abdominal adipose tissue and metabolic syndrome: a case control study. Diabetol Metab Syndr. 2016;8:10. Published 2016 Feb 12. doi: 10.1186/s13098-016-0127-7 Lee S, Kuk JL, Kim Y, Arslanian SA. Measurement site of visceral adipose tissue and prediction of metabolic syndrome in youth. Pediatr Diabetes. 2011;12(3 Pt 2):250–257. doi: 10.1111/j.1399-5448.2010.00705.x Shah RV, Murthy VL, Abbasi SA, et al. Visceral adiposity and the risk of metabolic syndrome across body mass index: the MESA Study. JACC Cardiovasc Imaging. 2014;7(12):1221–1235. doi: 10.1016/j.jcmg.2014.07.017 Cho SA, Joo HJ, Cho JY, et al. Visceral Fat Area and Serum Adiponectin Level Predict the Development of Metabolic Syndrome in a Community-Based Asymptomatic Population. PLoS One. 2017;12(1):e0169289. Published 2017 Jan 3. doi: 10.1371/journal.pone.0169289 Matsushita Y, Nakagawa T, Yamamoto S, et al. Effect of longitudinal changes in visceral fat area on incidence of metabolic risk factors: the Hitachi health study. Obesity (Silver Spring). 2013;21(10):2126–2129. doi: 10.1002/oby.20347 Nakaji S, Ihara K, Sawada K, et al. Social innovation for life expectancy extension utilizing a platform-centered system used in the Iwaki health promotion project: A protocol paper. SAGE Open Med. 2021;9:20503121211002606. Published 2021 Mar 19. doi: 10.1177/20503121211002606 Yamaguchi T, Ozato N, Katashima M, et al. A Novel Method to Visualize the Dietary Macronutrient Composition of Smaller Visceral Fat Accumulation. Front Nutr. 2020;6:194. Published 2020 Jan 24. doi: 10.3389/fnut.2019.00194 Ryo M, Maeda K, Onda T, et al. A new simple method for the measurement of visceral fat accumulation by bioelectrical impedance. Diabetes Care. 2005;28(2):451–453. doi: 10.2337/diacare.28.2.451 Ozato N, Saito S, Yamaguchi T, et al. Association between Nutrients and Visceral Fat in Healthy Japanese Adults: A 2-Year Longitudinal Study Brief Title: Micronutrients Associated with Visceral Fat Accumulation. Nutrients. 2019;11(11):2698. Published 2019 Nov 7. doi: 10.3390/nu11112698 Kinoshita K, Ozato N, Yamaguchi T, et al. The effect of age on the association between daily gait speed and abdominal obesity in Japanese adults. Sci Rep. 2021;11(1):19975. Published 2021 Oct 7. doi: 10.1038/s41598-021-98679-1 Ozato N, Saito S, Yamaguchi T, et al. Blautia genus associated with visceral fat accumulation in adults 20–76 years of age. NPJ Biofilms Microbiomes. 2019;5(1):28. Published 2019 Oct 4. doi: 10.1038/s41522-019-0101-x Yamaguchi T, Ozato N, Katashima M, et al. A Novel Method to Visualize the Dietary Macronutrient Composition of Smaller Visceral Fat Accumulation. Front Nutr. 2020;6:194. Published 2020 Jan 24. doi: 10.3389/fnut.2019.00194 Youden WJ. Index for rating diagnostic tests. Cancer. 1950;3(1):32–35. doi: 10.1002/1097-0142(1950)3:13.0.co;2-3 Metz CE. Basic principles of ROC analysis. Semin Nucl Med. 1978;8(4):283–298. doi: 10.1016/s0001-2998(78)80014-2 Vermont J, Bosson JL, François P, et al. Strategies for graphical threshold determination. Comput Methods Programs Biomed. 1991;35(2):141–150. doi: 10.1016/0169-2607(91)90072-2 DeLong ER, DeLong DM, Clarke-Pearson DL. Comparing the areas under two or more correlated receiver operating characteristic curves: a nonparametric approach. Biometrics. 1988;44(3):837–845. Lundberg S, Lee S-I. A unified approach to interpreting model predictions. Arxiv. 2017, https://arxiv.org/abs/1705.07874 . Urashima M, Wada T, Fukumoto T, et al. Prevalence of metabolic syndrome in a 22,892 Japanese population and its associations with life style. Japan Medical Association Journal. 2005;48:441–450. Nakanishi N, Takatorige T, Suzuki K. Cigarette smoking and the risk of the metabolic syndrome in middle-aged Japanese male office workers. Ind Health. 2005;43(2):295–301. doi: 10.2486/indhealth.43.295 Hiuge-Shimizu A, Kishida K, Funahashi T, et al. Absolute value of visceral fat area measured on computed tomography scans and obesity-related cardiovascular risk factors in large-scale Japanese general population (the VACATION-J study). Ann Med. 2012;44(1):82–92. doi: 10.3109/07853890.2010.526138 Matthews BW. Comparison of the predicted and observed secondary structure of T4 phage lysozyme. Biochim Biophys Acta. 1975;405(2):442–451. doi: 10.1016/0005-2795(75)90109-9 World Health Organization. Obesity: preventing and managing the global epidemic. Report of a WHO consultation. World Health Organ Tech Rep Ser. 2000;894:1–253. Jeon HH, Lee YK, Kim DH, Pak H, Shin SY, Seo JH. Risk for metabolic syndrome in the population with visceral fat area measured by bioelectrical impedance analysis. Korean J Intern Med. 2021;36(1):97–105. doi: 10.3904/kjim.2018.427 Kim YA, Kwak SG, Cho YJ. Optimal cutoff values for visceral fat volume to predict metabolic syndrome in a Korean population. Medicine (Baltimore). 2021;100(36):e27114. doi: 10.1097/MD.0000000000027114 Fujita T. The metabolic syndrome in Japan. Nat Clin Pract Cardiovasc Med. 2008;5 Suppl 1:S15-S18. doi: 10.1038/ncpcardio0808 Sairenchi T, Iso H, Yamagishi K, et al. Impact and attribute of each obesity-related cardiovascular risk factor in combination with abdominal obesity on total health expenditures in adult Japanese National Health insurance beneficiaries: The Ibaraki Prefectural health study. J Epidemiol. 2017;27(8):354–359. doi: 10.1016/j.je.2016.08.009 Table Table 1 is available in the Supplementary Files section Additional Declarations (Not answered) Supplementary Files SupplementalMaterials.pptx SUPPLEMENTAL CAPTIONS Supplemental Figure 1. Flowchart for constructing a MetS onset prediction model with VFA as the input features, taken from DataRobot. Hyperparameters for Elastic-Net: PostProcessing: wordcloud : False; PreProcessing: frozen parameters lid : None; Forest: n_jobs : 1; PostProcessing: stack_margin : False; PostProcessing: stack_folds : 5; fit_intercept : True; Forest: random_state : 1234; PostProcessing: right_censoring : None; random_state : 1234; tol : 0.0001; PostProcessing: prime_alpha_index : None; PreProcessing: language : None; Stepwise: backwards : 0; ShapFit: shap_fit : True; max_iter : 100; PostProcessing: stack_keep_top_n : 0; fit_alpha_scaler : True; enet_alpha : 0.5; enet_lambda : auto; beta_transform : id; ShapFit: shap_center : True; loss : log; PostProcessing: predictions_to_boost : False; warm_start : False; PostProcessing: move_imputed : False; Forest: n_estimators : 1; Stepwise: test_fraction : 0.25; PostProcessing: left_censoring : None; PostProcessing: stage : None; PostProcessing: stack : True; PostProcessing: stack_sequential : False; Decay: Type : None; PreProcessing: balance_weights : False; sigma : 1e-06 Supplemental Figure 2. Flowchart for constructing a MetS onset prediction model with VFA and six MetS-related parameters as input features , taken from DataRobot. Hyperparameters for LightGBM: n_jobs : -1; num_leaves : 2,4,16; reg_alpha : 0.0; PreProcessing: frozen parameters lid : None; early_stopping_rounds : 200; boosting_type : gbdt; PostProcessing: stack_margin : False; PostProcessing: stack_folds : 5; subsample_for_bin : 50000; min_child_samples : 10; sigmoid : 1.0; reg_lambda : 0.0; min_split_gain : 0.0; PostProcessing: right_censoring : None; Forest: random_state : 1234; max_bin : 255; objective : binary; PreProcessing: language : None; max_depth : none; Stepwise: backwards : 0; ShapFit: shap_fit : True; subsample_freq : 1; PostProcessing: stack_keep_top_n : 0; learning_rate : 0.05; PostProcessing: prime_alpha_index : None; PostProcessing: wordcloud : False; n_estimators : 1000; ShapFit: shap_center : False; Forest: n_jobs : 1; PostProcessing: predictions_to_boost : False; link_transform : None; PostProcessing: move_imputed : False; colsample_bytree : 1.0; min_child_weight : 5; Forest: n_estimators : 1; Stepwise: test_fraction : 0.25; PostProcessing: left_censoring : None; PostProcessing: stage : None; subsample : 1.0; PostProcessing: stack : True; PostProcessing: stack_sequential : False; is_unbalance : False; Decay: Type : None; PreProcessing: balance_weights : False; verbosity : 0 Supplemental Figure 3. Prediction accuracy when VFA was replaced with waist circumference as the input feature. In both models, the learner and hyperparameters were optimised according to the input features. AUC differences between the two models were calculated using Delong’s test. Features: VFA only, VFA; WC only, WC Learner: VFA only, Elastic-Net; WC only, Residual Neural Network Supplemental Figure 4. Prediction accuracy when VFA + six parameters was changed to waist circumference + six parameters. In both models, the learner and hyperparameters were optimised according to the input features. AUC differences between the two models were calculated using Delong’s test. Features: VFA + six parameters, VFA, BMI, SBP, Gender, DBP, Age, Cigarettes; WC + six parameters, WC, Gender, BMI, Age, SBP, DBP, Cigarettes Learner: VFA + six parameters, LightGBM; WC + six parameters, Logistic regression Supplemental Table 1. Model accuracy with input variables changed. Model 1 was constructed using LightGBM and input parameters VFA, BMI, number of cigarettes smoked, gender, age, SBP, DBP, and drinking firstly, and display the feature impact based on SHAP value. After excluding the lowest ranked feature, Model 2 was constructed using LightGBM and input parameters VFA, BMI, number of cigarettes smoked, gender, age, SBP, and DBP. The process was repeated until only one parameter remained and Model 3 was constructed using Elastic-Net and input parameter VFA. Table.pptx Table 1. Physical characteristics at baseline for the MetS onset and non-onset groups. The p value indicates the difference between the MetS onset and non-onset groups. a, Mann–Whitney U test (corrected for ties) was used for continuous variables; b, Fisher’s exact test was used for categorical variables. 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-3996594","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":276483762,"identity":"fd6ebd23-efd3-4375-ae07-eee391cbc7f3","order_by":0,"name":"Hiroto Bushita","email":"","orcid":"https://orcid.org/0000-0002-1045-890X","institution":"Kao Corporation","correspondingAuthor":false,"prefix":"","firstName":"Hiroto","middleName":"","lastName":"Bushita","suffix":""},{"id":276483763,"identity":"21184e6e-bd09-4263-a0e2-5d903d195fb1","order_by":1,"name":"Naoki Ozato","email":"","orcid":"https://orcid.org/0000-0002-2038-4705","institution":"Kao Corporation","correspondingAuthor":false,"prefix":"","firstName":"Naoki","middleName":"","lastName":"Ozato","suffix":""},{"id":276483764,"identity":"89452517-9bdf-44c7-bc68-fbb8fd114797","order_by":2,"name":"Kenta Mori","email":"","orcid":"","institution":"Kao Corporation","correspondingAuthor":false,"prefix":"","firstName":"Kenta","middleName":"","lastName":"Mori","suffix":""},{"id":276483765,"identity":"6efdecfb-4c4a-426c-bcd8-578bf2abfc5d","order_by":3,"name":"Hiromitsu Kawada","email":"","orcid":"","institution":"Kao Corporation","correspondingAuthor":false,"prefix":"","firstName":"Hiromitsu","middleName":"","lastName":"Kawada","suffix":""},{"id":276483766,"identity":"047f8745-bf63-48e8-9678-7cf44d0ab4ff","order_by":4,"name":"Yoshihisa Katsuragi","email":"","orcid":"","institution":"Kao Corporation","correspondingAuthor":false,"prefix":"","firstName":"Yoshihisa","middleName":"","lastName":"Katsuragi","suffix":""},{"id":276483767,"identity":"8ca1e0d7-1e9c-4d51-a161-20dd4d5ddd8c","order_by":5,"name":"Noriko Osaki","email":"","orcid":"","institution":"Kao Corporation","correspondingAuthor":false,"prefix":"","firstName":"Noriko","middleName":"","lastName":"Osaki","suffix":""},{"id":276483768,"identity":"25225982-dd64-4274-9909-0ec20e7a260d","order_by":6,"name":"Tatsuya Mikami","email":"","orcid":"","institution":"Hirosaki University Graduate School of Medicine","correspondingAuthor":false,"prefix":"","firstName":"Tatsuya","middleName":"","lastName":"Mikami","suffix":""},{"id":276483769,"identity":"7a808e65-483e-4333-b7ec-dbabee52dd12","order_by":7,"name":"Ken Itoh","email":"","orcid":"","institution":"Hirosaki University Graduate School of Medicine","correspondingAuthor":false,"prefix":"","firstName":"Ken","middleName":"","lastName":"Itoh","suffix":""},{"id":276483770,"identity":"8a4a45f2-8250-46b7-b318-0ec0c6ed83ea","order_by":8,"name":"Koichi Murashita","email":"","orcid":"","institution":"Hirosaki University","correspondingAuthor":false,"prefix":"","firstName":"Koichi","middleName":"","lastName":"Murashita","suffix":""},{"id":276483771,"identity":"08b3fa97-753b-4bde-bca7-99db3039203f","order_by":9,"name":"Shigeyuki Nakaji","email":"","orcid":"https://orcid.org/0000-0001-8445-8606","institution":"Hirosaki University Graduate School of Medicine","correspondingAuthor":false,"prefix":"","firstName":"Shigeyuki","middleName":"","lastName":"Nakaji","suffix":""},{"id":276483761,"identity":"3796ca3a-f6eb-4275-ab7c-94b26e2e7ee7","order_by":10,"name":"Yoshinori Tamada","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAABAklEQVRIiWNgGAWjYPACZgZ+CcYGEMuADS6ERzWYkJxBshaDGxCuAUEXGRzgP/jpRo21vPHt5rYPH/7YGfMx8Bgw/KhhYDfHqYWZWTrnWLrhtjsHm2fObEs2YwNqYew5xsBs2YBTC4N0Dtthxm03EpuZeRuYbdjk3xgw8DYwMBscwG3L75x/h+03zwBq4flTbwO25S9+LWzSuW2HEzdIgLSwHQY7jBmfLZKHmc2sc/vSk2cAHcY4s+24MRsDW8FhmWMSOP3Cd7zx8e2cb9a2/TPSHzN8+FNtOL+BeePDNzU2ybhCTOEwNlGgkySSccWQPA7bGRjsCEfqKBgFo2AUjBAAAJ2kUVgmK1DXAAAAAElFTkSuQmCC","orcid":"https://orcid.org/0000-0003-3483-9712","institution":"Hirosaki University Graduate School of Medicine","correspondingAuthor":true,"prefix":"","firstName":"Yoshinori","middleName":"","lastName":"Tamada","suffix":""}],"badges":[],"createdAt":"2024-02-28 12:12:09","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-3996594/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-3996594/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":52030496,"identity":"0b3280d5-dc18-4f2a-8595-a2426f0b92d5","added_by":"auto","created_at":"2024-03-05 16:22:08","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":29150,"visible":true,"origin":"","legend":"\u003cp\u003eResearch steps employed in this study.\u003c/p\u003e","description":"","filename":"Slide1.png","url":"https://assets-eu.researchsquare.com/files/rs-3996594/v1/d35e6708eddc6a4739b7cc17.png"},{"id":52096412,"identity":"e8b4d773-0adb-4960-9514-ed3a289daf15","added_by":"auto","created_at":"2024-03-06 16:07:14","extension":"jpg","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":46403,"visible":true,"origin":"","legend":"\u003cp\u003eEffect of baseline visceral fat on MetS onset (A) VFA values at baseline in MetS onset group (n=169) and non-onset group (n=1058), The p value indicates the difference between the MetS onset and non-onset groups using the Mann–Whitney U test (corrected for ties); (B) ROC curve of baseline VFA that determines MetS onset risk. The cut-off value was calculated using the Youden index.\u003c/p\u003e","description":"","filename":"figure2.jpg","url":"https://assets-eu.researchsquare.com/files/rs-3996594/v1/e1b12c0e7bab65bc9e149c3f.jpg"},{"id":52032735,"identity":"ff05e4db-d669-4839-bd70-5f21a1a40b11","added_by":"auto","created_at":"2024-03-05 16:38:08","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":12815,"visible":true,"origin":"","legend":"\u003cp\u003eComparison of AUC during cross validation of two models for MetS onset prediction.\u003c/p\u003e\n\u003cp\u003eModel 3 was constructed using Elastic-Net and the input feature was VFA. Model 2 was constructed using LightGBM and the input features were VFA, BMI, number of cigarettes smoked, gender, age, SBP and DBP. The AUC difference was calculated using DeLong’s test.\u003c/p\u003e","description":"","filename":"Slide3.png","url":"https://assets-eu.researchsquare.com/files/rs-3996594/v1/d8dab0ff82282f2ab776908d.png"},{"id":52030497,"identity":"432a805f-a4ef-4f16-8c48-291e6fe78914","added_by":"auto","created_at":"2024-03-05 16:22:08","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":11186,"visible":true,"origin":"","legend":"\u003cp\u003eFeature importance based on SHAP value.\u003c/p\u003e","description":"","filename":"Slide4.png","url":"https://assets-eu.researchsquare.com/files/rs-3996594/v1/873849ed5f98ad72416a88c9.png"},{"id":52181823,"identity":"e07012d6-cbe1-48c6-a5d0-4e6a553bee1e","added_by":"auto","created_at":"2024-03-07 18:04:37","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":554417,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-3996594/v1/7390506e-5e9c-4f0e-a7ae-984dcf0c6e79.pdf"},{"id":52030500,"identity":"3476bc8e-ce93-4c14-a505-9a3bb8059c25","added_by":"auto","created_at":"2024-03-05 16:22:08","extension":"pptx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":117273,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eSUPPLEMENTAL CAPTIONS\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eSupplemental Figure 1. Flowchart for constructing a MetS onset prediction model with VFA as the input features, taken from DataRobot.\u003c/p\u003e\n\u003cp\u003eHyperparameters for Elastic-Net: PostProcessing: wordcloud : False; PreProcessing: frozen parameters lid : None; Forest: n_jobs : 1; PostProcessing: stack_margin : False; PostProcessing: stack_folds : 5; fit_intercept : True; Forest: random_state : 1234; PostProcessing: right_censoring : None; random_state : 1234; tol : 0.0001; PostProcessing: prime_alpha_index : None; PreProcessing: language : None; Stepwise: backwards : 0; ShapFit: shap_fit : True; max_iter : 100; PostProcessing: stack_keep_top_n : 0; fit_alpha_scaler : True; enet_alpha : 0.5; enet_lambda : auto; beta_transform : id; ShapFit: shap_center : True; loss : log; PostProcessing: predictions_to_boost : False; warm_start : False; PostProcessing: move_imputed : False; Forest: n_estimators : 1; Stepwise: test_fraction : 0.25; PostProcessing: left_censoring : None; PostProcessing: stage : None; PostProcessing: stack : True; PostProcessing: stack_sequential : False; Decay: Type : None; PreProcessing: balance_weights : False; sigma : 1e-06\u003c/p\u003e\n\u003cp\u003eSupplemental Figure 2. Flowchart for constructing a MetS onset prediction model with VFA and six MetS-related parameters as input features , taken from DataRobot.\u003c/p\u003e\n\u003cp\u003eHyperparameters for LightGBM: n_jobs : -1; num_leaves : 2,4,16; reg_alpha : 0.0; PreProcessing: frozen parameters lid : None; early_stopping_rounds : 200; boosting_type : gbdt; PostProcessing: stack_margin : False; PostProcessing: stack_folds : 5; subsample_for_bin : 50000; min_child_samples : 10; sigmoid : 1.0; reg_lambda : 0.0; min_split_gain : 0.0; PostProcessing: right_censoring : None; Forest: random_state : 1234; max_bin : 255; objective : binary; PreProcessing: language : None; max_depth : none; Stepwise: backwards : 0; ShapFit: shap_fit : True; subsample_freq : 1; PostProcessing: stack_keep_top_n : 0; learning_rate : 0.05; PostProcessing: prime_alpha_index : None; PostProcessing: wordcloud : False; n_estimators : 1000; ShapFit: shap_center : False; Forest: n_jobs : 1; PostProcessing: predictions_to_boost : False; link_transform : None; PostProcessing: move_imputed : False; colsample_bytree : 1.0; min_child_weight : 5; Forest: n_estimators : 1; Stepwise: test_fraction : 0.25; PostProcessing: left_censoring : None; PostProcessing: stage : None; subsample : 1.0; PostProcessing: stack : True; PostProcessing: stack_sequential : False; is_unbalance : False; Decay: Type : None; PreProcessing: balance_weights : False; verbosity : 0\u003c/p\u003e\n\u003cp\u003eSupplemental Figure 3. Prediction accuracy when VFA was replaced with waist circumference as the input feature.\u003c/p\u003e\n\u003cp\u003eIn both models, the learner and hyperparameters were optimised according to the input features. AUC differences between the two models were calculated using Delong’s test.\u003c/p\u003e\n\u003cp\u003eFeatures: VFA only, VFA; WC only, WC\u003c/p\u003e\n\u003cp\u003eLearner: VFA only, Elastic-Net; WC only, Residual Neural Network\u003c/p\u003e\n\u003cp\u003eSupplemental Figure 4. Prediction accuracy when VFA + six parameters was changed to waist circumference + six parameters.\u003c/p\u003e\n\u003cp\u003eIn both models, the learner and hyperparameters were optimised according to the input features. AUC differences between the two models were calculated using Delong’s test.\u003c/p\u003e\n\u003cp\u003eFeatures: VFA + six parameters, VFA, BMI, SBP, Gender, DBP, Age, Cigarettes; WC + six parameters, WC, Gender, BMI, Age, SBP, DBP, Cigarettes\u003c/p\u003e\n\u003cp\u003eLearner: VFA + six parameters, LightGBM; WC + six parameters, Logistic regression\u003c/p\u003e\n\u003cp\u003eSupplemental Table 1. Model accuracy with input variables changed.\u003c/p\u003e\n\u003cp\u003eModel 1 was constructed using LightGBM and input parameters VFA, BMI, number of cigarettes smoked, gender, age, SBP, DBP, and drinking firstly, and display the feature impact based on SHAP value. After excluding the lowest ranked feature, Model 2 was constructed using LightGBM and input parameters VFA, BMI, number of cigarettes smoked, gender, age, SBP, and DBP. The process was repeated until only one parameter remained and Model 3 was constructed using Elastic-Net and input parameter VFA.\u003c/p\u003e","description":"","filename":"SupplementalMaterials.pptx","url":"https://assets-eu.researchsquare.com/files/rs-3996594/v1/2b9a76bb44317e0897daf3c2.pptx"},{"id":52031639,"identity":"d93cbccb-d52c-4260-a1b0-0bcc327213d0","added_by":"auto","created_at":"2024-03-05 16:30:08","extension":"pptx","order_by":2,"title":"","display":"","copyAsset":false,"role":"supplement","size":47143,"visible":true,"origin":"","legend":"\u003cp\u003eTable 1. Physical characteristics at baseline for the MetS onset and non-onset groups.\u003c/p\u003e\n\u003cp\u003eThe p value indicates the difference between the MetS onset and non-onset groups.\u003c/p\u003e\n\u003cp\u003ea, Mann–Whitney U test (corrected for ties) was used for continuous variables; b, Fisher’s exact test was used for categorical variables.\u003c/p\u003e","description":"","filename":"Table.pptx","url":"https://assets-eu.researchsquare.com/files/rs-3996594/v1/38625fcedd26c6ef6959f689.pptx"}],"financialInterests":"(Not answered)","formattedTitle":"Effect of visceral fat on onset of metabolic syndrome","fulltext":[{"header":"Introduction","content":"\u003cp\u003eCardiometabolic diseases, including metabolic syndrome (MetS), have become the leading cause of death worldwide in the past 20 years [\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e]. MetS involves several risk factors such as abdominal obesity, hyperglycemia, hypertension, and dyslipidaemia and is becoming a serious problem worldwide [\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e, \u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e, \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e]. In 2020, approximately 25.8\u0026nbsp;million children and 35.5\u0026nbsp;million adolescents worldwide were affected by MetS [\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e]. Therefore, the prevention and mitigation of MetS is an urgent issue. Various preventive approaches have been proposed (exercise guidance [\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e] and dietary guidance [\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e, \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e, \u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e]). On the other hand, it has been reported that behavioural changes can be expected by knowing the risk of disease [\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e]. Therefore, the ability to predict the onset of metabolic syndrome years in advance is considered to be one of the most useful tools for preventing the development of MetS. Several algorithms for predicting the onset of MetS have been developed [\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e, \u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e, \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e, \u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e, \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e, \u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e, \u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e, \u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e, \u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e, \u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e, \u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e, \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e]. However, the definition of MetS differs in different country [\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e]. In Japan, visceral fat accumulation is essential for diagnosis, with a cut-off value of 100 cm\u003csup\u003e2\u003c/sup\u003e [\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e, \u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e]. One reason for this is that Asians are more susceptible to MetS than Westerners, and even those with a low BMI accumulate visceral fat [\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e, \u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e]. Thus, accumulating research knowledge in accordance with each country\u0026rsquo;s standards is important.\u003c/p\u003e \u003cp\u003eThe relationship between visceral fat and MetS has been the subject of several cross-sectional studies in Japan, China, South Korea, and the United States [\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e, \u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e, \u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e, \u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e, \u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e, \u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eA 3.3-year longitudinal study in the United States [\u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e] and a 17-month longitudinal study in South Korea [\u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e] reported that visceral fat increases MetS onset risk as defined by the waist size. A 3-year longitudinal study in Japan reported that an increase in visceral fat is correlated with MetS risk factors [\u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e]. However, no large-scale studies have been conducted in Japan measuring longitudinal changes in visceral fat area (VFA) in the same subjects over a period of 1\u0026ndash;6 years. Moreover, the relationship between MetS onset risk and visceral fat remains relatively unexplored. We have been collecting visceral fat-related health data since 2015. This study used the data for the period 2015\u0026ndash;2020 to clarify the relationship between visceral fat and MetS onset. An algorithm to predicts MetS onset was developed.\u003c/p\u003e"},{"header":"Methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003eDataset construction\u003c/h2\u003e \u003cp\u003eThe Iwaki Health Promotion Project [\u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e] was a health examination for residents aged 20 years and older conducted in Hirosaki City, Aomori Prefecture, Japan. A variety of data were collected in this project, including those on genes, physical characteristics, behavioural habits, and intestinal bacteria. Visceral fat data were collected from 2015 onwards, and this study used data for the period 2015\u0026ndash;2020. The health examination data included visceral fat area (VFA), blood data, dietary data, and interview data. VFA was estimated by the abdominal bio-impedance method using EW-FA90 (Panasonic Corporation), an approved medical device in Japan (No. 22500BZX00522000) [\u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e]. Measurement values from this device have been reported to have a strong correlation with those obtained using computed tomography (CT), which is the gold standard for measuring visceral fat [\u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e39\u003c/span\u003e]. Other basic blood parameters were obtained as previously reported [\u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e40\u003c/span\u003e, \u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e41\u003c/span\u003e, \u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e42\u003c/span\u003e, \u003cspan citationid=\"CR43\" class=\"CitationRef\"\u003e43\u003c/span\u003e]. We have previously used the data from the present study to conduct research on diet [\u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e40\u003c/span\u003e, \u003cspan citationid=\"CR43\" class=\"CitationRef\"\u003e43\u003c/span\u003e], exercise [\u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e41\u003c/span\u003e], and intestinal bacteria [\u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e42\u003c/span\u003e]. This study was approved by the Hirosaki University Graduate School of Medicine ethics committee (approval number: 2014-377-1, 2016-028-1, 2021-030, 2018-012, 2020-046-4, 2020-046-1) and conducted in accordance with the recommendations set out in the Declaration of Helsinki. All participants provided written informed consent.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec4\" class=\"Section2\"\u003e \u003ch2\u003eMetS onset criteria\u003c/h2\u003e \u003cp\u003eThe following were used as MetS diagnosis criteria. The presence or absence of MetS was determined for each participant at each time point in the dataset [\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eRequired criteria: \u0026ldquo;VFA\u0026thinsp;\u0026ge;\u0026thinsp;100 cm\u003csup\u003e2\u003c/sup\u003e (for both males and females)\u0026rdquo;\u003c/p\u003e \u003cp\u003eOne of the following conditions apply:\u003c/p\u003e \u003cp\u003e(1) Triglycerides (TG)\u0026thinsp;\u0026ge;\u0026thinsp;150 mg/dL, high-density lipoprotein cholesterol (HDLC)\u0026thinsp;\u0026lt;\u0026thinsp;50 mg/dL, or currently receiving medication for that disease; or all three\u003c/p\u003e \u003cp\u003e(2) Systolic blood pressure (SBP)\u0026thinsp;\u0026ge;\u0026thinsp;130 mmHg, diastolic blood pressure (DBP)\u0026thinsp;\u0026ge;\u0026thinsp;85 mmHg, or taking antihypertensive medication; or all three\u003c/p\u003e \u003cp\u003e(3) Blood sugar (BS)\u0026thinsp;\u0026ge;\u0026thinsp;110 mg/dL, receiving antidiabetic therapy, or both\u003c/p\u003e \u003cp\u003eFinally, from the data compiled from six years of Iwaki Health Promotion Project (n\u0026thinsp;=\u0026thinsp;5905), we excluded those participants who were not assigned a label within three years (n\u0026thinsp;=\u0026thinsp;3870) and those who currently fell under the category of MetS (n\u0026thinsp;=\u0026thinsp;502) to create the final dataset (n\u0026thinsp;=\u0026thinsp;1533).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec5\" class=\"Section2\"\u003e \u003ch2\u003eData pre-processing\u003c/h2\u003e \u003cp\u003eWe used the DataRobot AI Platform (ver.9.1, DataRobot) to convert the numerical and categorical variables into the format that was most appropriate for each model, including standardisation, missing value imputation, one-hot encoding, rigid transformation, and ordinal encoding.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec6\" class=\"Section2\"\u003e \u003ch2\u003eDataset creation and analysis method\u003c/h2\u003e \u003cp\u003eThe dataset (n\u0026thinsp;=\u0026thinsp;1533) was randomly divided into training data for five-fold cross validation (80%, n\u0026thinsp;=\u0026thinsp;1227) and test data (20%, n\u0026thinsp;=\u0026thinsp;306) using DataRobot. When dividing the data, we used \u0026ldquo;group\u0026rdquo; partition and set \u0026ldquo;individual ID\u0026rdquo; as the group ID. This avoids leakage, as data from the same lot is not spread across training, validation, and test data. We computed the accuracy and stability of each model using five-fold cross validation to select the optimal model. We used the test data to estimate model prediction accuracy for unknown data not used in model creation.\u003c/p\u003e \u003cp\u003eThe data for cross validation were used to confirm the effect of visceral fat on MetS onset. Differences between each parameter of the MetS onset and non-onset groups were analysed using R software version 4.3.1, and Fisher\u0026rsquo;s exact and Mann\u0026ndash;Whitney U (corrected for ties) tests were applied for the categorical and continuous variables, respectively.\u003c/p\u003e \u003cp\u003eWe evaluated the effect of visceral fat on the onset of MetS as follows. We used box plots, the exact Mann\u0026ndash;Whitney U test, and logistic regression analysis adjustments for confounding factors (gender, age, number of cigarettes smoked, alcohol intake, and exercise intensity) to obtain receiver operating characteristic (ROC) curves using R and calculated the VFA cut-off value for MetS onset risk based on the Youden index [\u003cspan citationid=\"CR44\" class=\"CitationRef\"\u003e44\u003c/span\u003e] or the ROC curve closest to (0,1) [\u003cspan citationid=\"CR45\" class=\"CitationRef\"\u003e45\u003c/span\u003e, \u003cspan citationid=\"CR46\" class=\"CitationRef\"\u003e46\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eNext, we built an MetS onset prediction model using the cross-validation data. First, we calculated the area under the curve (AUC) using VFA as the only parameter. We then considered seven additional parameters previously reported to be associated with MetS (gender, age, BMI, systolic blood pressure, diastolic blood pressure, number of drinking days per week, and number of cigarettes smoked) [\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e] for a total of eight input features in the prediction model. Machine learning was employed using DataRobot to estimate their effect on prediction accuracy. The model with the highest prediction accuracy based on the AUC score was chosen from the 18 models. The chosen model was used for MetS onset risk predictions on the test dataset and calculate the AUC score. The 18 models included general linear and classification models, including XGBoost, Elastic-Net, LightGBM, logistic regression, and residual neural networks. We used an AUC difference test (DeLong\u0026rsquo;s test [\u003cspan citationid=\"CR47\" class=\"CitationRef\"\u003e47\u003c/span\u003e]) for the AUC calculated from this analysis. For each model, we analysed the difference between the cross-validation and test AUCs. We visualised the contribution of each of the eight features to the MetS onset risk, using the SHapley Additive exPlanations (SHAP) [\u003cspan citationid=\"CR48\" class=\"CitationRef\"\u003e48\u003c/span\u003e] function of DataRobot. First, we used SHAP to visualise the effect of fluctuations in a given feature on the onset risk of each individual. Next, we calculated the absolute value of the degree of influence of each feature in each individual. The relative magnitude of the averaged value was compared and displayed as the feature impact. This facilitated ranking the features according to their contribution to MetS onset risk.\u003c/p\u003e \u003c/div\u003e"},{"header":"Results","content":"\u003cdiv id=\"Sec8\" class=\"Section2\"\u003e \u003ch2\u003ePhysical characteristics\u003c/h2\u003e \u003cp\u003eIn the data for cross validation (n\u0026thinsp;=\u0026thinsp;1227), the onset and non-onset groups comprised 169 and 1058 cases, respectively. In the test data (n\u0026thinsp;=\u0026thinsp;306), the onset and non-onset groups comprised 44 and 262 cases, respectively. The MetS incidence in each dataset was 13.8% and 14.4%, respectively. Table\u0026nbsp;1 shows the subject background information at the baseline in the cross-validation data.\u003c/p\u003e \u003cp\u003eIn the onset group, 36.7% were females and 63.3% were males, and in the non-onset group, 67.1% were females and 32.9% were males. Onset and gender were significantly related (p\u0026thinsp;\u0026lt;\u0026thinsp;0.001). The median age in the onset group was 57 years, which was significantly higher than that in the non-onset group (52 years). Compared with the non-onset group, the onset group had a significantly greater height, and a significantly larger BMI, VFA, and waist circumference (p\u0026thinsp;\u0026lt;\u0026thinsp;0.001). In addition, Triglycerides (TG), high-density lipoprotein cholesterol (HDLC), low-density lipoprotein cholesterol (LDLC), systolic blood pressure (SBP), diastolic blood pressure (DBP), Blood sugar (BS), and haemoglobin A1c (HbA1c), which are factors involved in MetS, were significantly higher (p\u0026thinsp;\u0026lt;\u0026thinsp;0.001) in the onset group. Energy intake (BDHQ), alcohol intake (BDHQ, after adjusting for energy), number of cigarettes smoked, and sleeping time were significantly longer (p\u0026thinsp;\u0026lt;\u0026thinsp;0.001, p\u0026thinsp;=\u0026thinsp;0.003, p\u0026thinsp;=\u0026thinsp;0.008, p\u0026thinsp;=\u0026thinsp;0.003) ) in the onset group. No significant differences were observed between the two groups with respect to number of drinking days and exercise intensity (p\u0026thinsp;=\u0026thinsp;0.654, p\u0026thinsp;=\u0026thinsp;0.399).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec9\" class=\"Section2\"\u003e \u003ch2\u003eEffect of visceral fat on MetS onset\u003c/h2\u003e \u003cp\u003eThe VFA at baseline was compared between the MetS onset and non-onset groups. In the onset group, 45% were VFA\u0026thinsp;\u0026lt;\u0026thinsp;100 cm\u003csup\u003e2\u003c/sup\u003e cases and 55% were VFA\u0026thinsp;\u0026ge;\u0026thinsp;100 cm\u003csup\u003e2\u003c/sup\u003e cases. In the non-onset group, 86.1% were VFA\u0026thinsp;\u0026lt;\u0026thinsp;100 cm\u003csup\u003e2\u003c/sup\u003e cases and 13.9% were VFA\u0026thinsp;\u0026ge;\u0026thinsp;100 cm\u003csup\u003e2\u003c/sup\u003e cases. The VFA quartile values in the onset group were as follows: 25%, 92 cm\u003csup\u003e2\u003c/sup\u003e; 50%, 105 cm\u003csup\u003e2\u003c/sup\u003e; and 75%, 130 cm\u003csup\u003e2\u003c/sup\u003e. In the non-onset group, the values were as follows: 25%, 42 cm\u003csup\u003e2\u003c/sup\u003e; 50%, 63 cm\u003csup\u003e2\u003c/sup\u003e; and 75%, 85 cm\u003csup\u003e2\u003c/sup\u003e, with the VFA being significantly higher in the onset group (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eA; p\u0026thinsp;\u0026lt;\u0026thinsp;0.001). Logistic regression analysis confirmed that there was no change in these results even after adjusting for gender, age, number of cigarettes smoked, alcohol intake, and exercise intensity (p\u0026thinsp;\u0026lt;\u0026thinsp;0.001). AUC for the ROC curve was 0.8597. The cut-off values based on the Youden index and the ROC curve closest to (0,1) were 82.5 cm\u003csup\u003e2\u003c/sup\u003e (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eB) and 85.5 cm\u003csup\u003e2\u003c/sup\u003e, respectively. These results suggest that baseline VFA alone can predict the onset of MetS. It was confirmed that 45% of the non-onset group did not experience the onset of MetS despite their baseline VFA exceeding 100 cm\u003csup\u003e2\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cdiv id=\"Sec10\" class=\"Section3\"\u003e \u003ch2\u003eConstruction and validation of MetS onset prediction model\u003c/h2\u003e \u003cp\u003eThe baseline VFA was determined to be an important factor for the prediction of MetS onset; thus, a model for predicting MetS onset was constructed using baseline visceral fat. First, 18 different machine learning models compatible with SHAP were constructed using only baseline VFA as the input parameter, and the prediction model with the highest prediction accuracy was selected. The prediction model constructed using Elastic-Net yielded a cross-validation AUC of 0.8591 and a test AUC of 0.8686 (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e, Supplemental Table\u0026nbsp;1). Supplemental Fig.\u0026nbsp;1 shows the flowchart of model construction and presents the hyperparameters optimised using grid search.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eA previous study reported on a MetS onset prediction model that was constructed using BMI, number of cigarettes smoked, gender, age, DBP, SBP, and other factors. Previous epidemiological research has suggested that these are important factors [\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e, \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e, \u003cspan citationid=\"CR49\" class=\"CitationRef\"\u003e49\u003c/span\u003e, \u003cspan citationid=\"CR50\" class=\"CitationRef\"\u003e50\u003c/span\u003e, \u003cspan citationid=\"CR51\" class=\"CitationRef\"\u003e51\u003c/span\u003e]. Therefore, we built an onset prediction model based on visceral fat and seven parameters from the literature [\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e]. First, we constructed 18 different machine learning models using these eight input parameters, and selected the model with the highest prediction accuracy. Results confirm the accuracy of the model (cross-validation AUC\u0026thinsp;=\u0026thinsp;0.8992, test AUC\u0026thinsp;=\u0026thinsp;0.8845) without overfitting (Model 1). Next, we ranked the parameters based on their SHAP values. We then constructed 18 models with seven input parameters after excluding the lowest ranked parameter (number of drinking days). We selected the model with the highest prediction accuracy (Model 2). This process was repeated until only one parameter remained (Model 3).\u003c/p\u003e \u003cp\u003eResults showed that Model 2, constructed the seven input features: visceral fat, BMI, number of cigarettes smoked, gender, age, DBP, and SBP, and trained with LightGBM, had the highest prediction accuracy (cross-validation AUC\u0026thinsp;=\u0026thinsp;0.9004, test AUC\u0026thinsp;=\u0026thinsp;0.8836), with no overfitting (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e, Supplemental Table\u0026nbsp;1)). Supplemental Fig.\u0026nbsp;2 shows the flowchart for model construction and hyperparameter optimisation using grid search. The onset and non-onset cases were correctly determined with accuracies of 82% and 84%, respectively, when the Matthews correlation coefficient (MCC) [\u003cspan citationid=\"CR52\" class=\"CitationRef\"\u003e52\u003c/span\u003e] was maximised (minimising false positives and false negatives). The correct answer rate was 84%.\u003c/p\u003e \u003cp\u003eReplacing VFA with waist circumference in the VFA-only model (Model 3) and the optimised model that included VFA (Model 2) significantly decreased the prediction accuracy (Supplemental Figs.\u0026nbsp;3 and 4). Therefore, VFA had a greater contribution to the prediction accuracy than waist circumference when predicting MetS onset.\u003c/p\u003e \u003cp\u003eIn the optimised MetS onset prediction model (Model 2), the SHAP value of each individual was calculated (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e) as the feature impact based on the SHAP value. An examination of the influence of each item (gender, age, VFA, BMI, DBP, SBP, and number of cigarettes smoked) on MetS onset showed that VFA was the largest contributor to the prediction of MetS onset. BMI was the second most influential factor, but its feature effect was approximately half that of VFA (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003c/div\u003e"},{"header":"Discussion","content":"\u003cp\u003eDiet and exercise management are important in preventing MetS [\u003cspan citationid=\"CR53\" class=\"CitationRef\"\u003e53\u003c/span\u003e]; however, predicting the onset is also an important approach. In Japan, visceral fat level is an essential parameter in the diagnosis of MetS [\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e]. The gold standard measurement method for visceral fat is CT. However, previous studies on MetS prediction have been conducted without obtaining data on visceral fat [\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e, \u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e, \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e, \u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e, \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e, \u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e, \u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e, \u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e, \u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e, \u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e, \u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e, \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e] because of the time-consuming and invasive process. We built a device that measures visceral fat non-invasively and constructed human dataset that includes VFA between 2015\u0026ndash;2020. We investigated the relationship between MetS onset and visceral fat, and developed a MetS onset prediction algorithm. The dataset comprised 169 and 1058 cases in the MetS onset and non-onset groups, respectively. The percentage of cases affected by visceral fat onset within three years was 13.8%. Analysis of the baseline VFA and MetS onset using a box plot showed that the VFA quartiles in the onset group were 25%: 92 cm\u003csup\u003e2\u003c/sup\u003e; and 75%, 130 cm\u003csup\u003e2\u003c/sup\u003e, and the VFA quartiles in the non-onset group were 25%: 42 cm\u003csup\u003e2\u003c/sup\u003e; and 75%, 85 cm\u003csup\u003e2\u003c/sup\u003e; showing that the data overlap was quite small (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eA). These results indicated a large difference in the initial visceral fat amount between the onset and non-onset cases. The ROC AUC of 0.8597 suggested that the baseline visceral fat was a strong factor influencing the prediction of MetS onset. The MetS cut-off value was found to be 82.5 cm\u003csup\u003e2\u003c/sup\u003e. Many cross-sectional studies have reported that visceral fat is associated with MetS onset [\u003cspan citationid=\"CR54\" class=\"CitationRef\"\u003e54\u003c/span\u003e, \u003cspan citationid=\"CR55\" class=\"CitationRef\"\u003e55\u003c/span\u003e]. Regarding cut-off values of VFA for MetS, a cross-sectional study of type II diabetes patients aged 18\u0026ndash;75 years in China reported values of 84.7 cm\u003csup\u003e2\u003c/sup\u003e in males and 81.1 cm\u003csup\u003e2\u003c/sup\u003e in females [\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e]. A cross-sectional study of Chinese patients aged 35\u0026ndash;75 years reported cut-off values of 79.2 cm\u003csup\u003e2\u003c/sup\u003e for both males and females [\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e]. A longitudinal study reported values of 84 cm\u003csup\u003e2\u003c/sup\u003e and 58 cm\u003csup\u003e2\u003c/sup\u003e in Korean males and females, respectively [\u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e]. In the present study, the MetS cut-off value was 82.5 cm\u003csup\u003e2\u003c/sup\u003e for both males and females (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eB), and the above previous studies suggest that our study was valid. MetS is recognised worldwide as a useful indicator for predicting cardiovascular risk [\u003cspan citationid=\"CR56\" class=\"CitationRef\"\u003e56\u003c/span\u003e], and the present study confirmed that controlling visceral fat was important in preventing MetS in Japanese people.\u003c/p\u003e \u003cp\u003eWe built an algorithm for predicting MetS onset using visceral fat. There are multiple algorithms available worldwide for predicting MetS onset [\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e, \u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e, \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e, \u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e, \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e, \u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e, \u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e, \u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e, \u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e, \u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e, \u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e, \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e]. One challenge in predicting MetS onset is that the diagnostic definition of MetS differs in each country. In Japan, despite the fact that visceral fat is an essential parameter, no algorithm for predicting disease onset using visceral fat values has been constructed thus far. Therefore, in this study, we developed an onset prediction algorithm using visceral fat measurements as the input parameter. Examination results confirmed the successful construction of an onset prediction algorithm using six parameters in addition to visceral fat (Model 2) (AUC\u0026thinsp;=\u0026thinsp;0.9004 in cross validation data). On the test data, the model yielded an AUC of 0.8836 (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e). The feature impact analysis showed visceral fat to be a dominant contributor (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e). There have previously been many onset prediction algorithms that use blood items [\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e, \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e, \u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e, \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e, \u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e, \u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e, \u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e, \u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e, \u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e, \u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e, \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e]. The visceral fat meter developed in this study is a non-invasive device. The algorithm developed in this study demonstrated that MetS onset can be predicted based on non-invasively measured parameters such as visceral fat, BMI, number of cigarettes smoked, gender, age, and DBP. Although this device has previously been used for visceral fat measurements, the findings of this study may result in the expansion of the functionality of the device. The MetS algorithm uses measurements of only non-invasive parameters and has high medical interpretability; thus, it is expected to facilitate easy and convincing understanding of onset risk. Moreover, it can be used in a variety of applications. Algorithm problems often involve regional differences; however, because the developed algorithm uses parameters with high medical interpretability, there may be fewer validation tests required. Medical expenses for obese people with visceral fat are higher than that for obese people without visceral fat [\u003cspan citationid=\"CR57\" class=\"CitationRef\"\u003e57\u003c/span\u003e], and the algorithm developed in this study can guide people to reduce their visceral fat before MetS onset.\u003c/p\u003e \u003cp\u003eIn this study, the prediction model with six parameters associated with MetS (BMI, number of cigarettes smoked, gender, age, DBP, and SBP) in addition to VFA (Model 2) exhibited a high prediction accuracy (cross-validation AUC\u0026thinsp;=\u0026thinsp;0.9004, test AUC\u0026thinsp;=\u0026thinsp;0.8836). The prediction accuracy significantly decreased when replacing VFA with waist circumference in Model 2 (Supplemental Fig.\u0026nbsp;4). Therefore, it was inferred that VFA was a more important factor than waist circumference when predicting MetS onset. Previous research shows only a weak correlation between VFA and waist circumference (males: r\u0026thinsp;=\u0026thinsp;0.68, females: r\u0026thinsp;=\u0026thinsp;0.65), with considerable individual-level variance between the VFA and waist circumference readings. For example, in the Japanese population, males with a waist circumference of 85.0\u0026ndash;86.0 cm have a VFA of 67\u0026ndash;137 cm\u003csup\u003e2\u003c/sup\u003e [\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eWe obtained six years of data on visceral fat, and built a MetS onset prediction model to determine whether onset would occur within three years of the baseline time. Previous research did not measure visceral fat, and the models generally include blood data, which can be an obstacle in daily monitoring. Therefore, the present study added medically important evidence after accurately examining a prediction model with an eye toward social implementation. The present study has several limitations. The dataset used to build the prediction model was limited to participants of a specific race and region; thus, the model\u0026rsquo;s performance must be checked by including participants from different races and regions. The MetS onset risk prediction model is a just a guidance tool, and when implemented, people must be asked to change their lifestyles and later confirm whether there is an actual decrease in the incidence of MetS.\u003c/p\u003e"},{"header":"Conclusions","content":"\u003cp\u003eWe built a six-year medical dataset that included visceral fat measurements. Visceral fat was found to be an important factor for determining the onset of MetS in the future. We developed a high-accuracy onset prediction algorithm using non-invasive parameters, including visceral fat.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eCompeting interests\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis study was supported by JST Grant Number JPMJCE1302, JPMJCA2201, JPMJPF2210, and Kao Co. (Tokyo, Japan). Authors HB, NOzato, KMori, HK, YK, and NOsaki were employed by Kao Corporation (Tokyo, Japan). All other authors declare no potential competing interests. The results of the study are presented clearly, honestly, and without fabrication, falsification, or inappropriate data manipulation.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAcknowledgements\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors thank all participants in the Iwaki Health Promotion Project and the entire staff of the project. The authors are thankful to Takuji Yasukawa, Takuro Iwane, and Yoshikuni Sugimura for valuable assistance, constant support, and advice in the preparation of this manuscript. The authors are grateful to Daisuke Kasuga, Eiichiro Uchino, and Yoko Sugiura for building the research environment.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthor Contributions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eHB designed the study, acquired, analysed and interpreted the data. NOzato contributed to data acquisition. All authors reviewed and edited the manuscript and approved its final version.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eData availability\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eData cannot be shared publicly because of the ethical concerns. Data are available from the Hirosaki University COI Institutional Data Access / Ethics Committee (contact via e-mail: [email protected]) for researchers who meet the criteria for access to the data. Researchers need to be approved by research ethics review board at the organization of their affiliation.\u003cstrong\u003e\u003cbr\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eGeneva WHO. Global health estimates 2020: Deaths by cause, age, sex, by country and by region, 2000\u0026ndash;2019. 2020\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSaklayen MG. The Global Epidemic of the Metabolic Syndrome. Curr Hypertens Rep. 2018;20(2):12. Published 2018 Feb 26. doi:\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1007/s11906-018-0812-z\u003c/span\u003e\u003cspan address=\"10.1007/s11906-018-0812-z\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eShin S, Jee H. Prevalence of metabolic syndrome in the Gulf Cooperation Council countries: meta-analysis of cross-sectional studies. J Exerc Rehabil. 2020;16(1):27\u0026ndash;35. Published 2020 Feb 26. doi:\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.12965/jer.1938758.379\u003c/span\u003e\u003cspan address=\"10.12965/jer.1938758.379\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePrasun P. Mitochondrial dysfunction in metabolic syndrome. Biochim Biophys Acta Mol Basis Dis. 2020;1866(10):165838. doi:\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.bbadis.2020.165838\u003c/span\u003e\u003cspan address=\"10.1016/j.bbadis.2020.165838\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNoubiap JJ, Nansseu JR, Lontchi-Yimagou E, et al. Global, regional, and country estimates of metabolic syndrome burden in children and adolescents in 2020: a systematic review and modelling analysis. Lancet Child Adolesc Health. 2022;6(3):158\u0026ndash;170. doi:\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/S2352-4642(21)00374-6\u003c/span\u003e\u003cspan address=\"10.1016/S2352-4642(21)00374-6\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCho ER, Shin A, Kim J, Jee SH, Sung J. Leisure-time physical activity is associated with a reduced risk for metabolic syndrome. Ann Epidemiol. 2009;19(11):784\u0026ndash;792. doi:\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.annepidem.2009.06.010\u003c/span\u003e\u003cspan address=\"10.1016/j.annepidem.2009.06.010\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMin C, Noh H, Kang YS, et al. Skipping breakfast is associated with diet quality and metabolic syndrome risk factors of adults. Nutr Res Pract. 2011;5(5):455\u0026ndash;463. doi:\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.4162/nrp.2011.5.5.455\u003c/span\u003e\u003cspan address=\"10.4162/nrp.2011.5.5.455\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eShin A, Lim SY, Sung J, Shin HR, Kim J. Dietary intake, eating habits, and metabolic syndrome in Korean men. J Am Diet Assoc. 2009;109(4):633\u0026ndash;640. doi:\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.jada.2008.12.015\u003c/span\u003e\u003cspan address=\"10.1016/j.jada.2008.12.015\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBianchi C, Penno G, Daniele G, Benzi L, Del Prato S, Miccoli R. Optimizing management of metabolic syndrome to reduce risk: focus on life-style. Intern Emerg Med. 2008;3(2):87\u0026ndash;98. doi:\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1007/s11739-008-0122-6\u003c/span\u003e\u003cspan address=\"10.1007/s11739-008-0122-6\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJanz NK, Becker MH. The Health Belief Model: a decade later. Health Educ Q. 1984;11(1):1\u0026ndash;47. doi:\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1177/109019818401100101\u003c/span\u003e\u003cspan address=\"10.1177/109019818401100101\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSalim AA, Kawasoe S, Kubozono T, et al. Development of predictive equation and score for 5-year metabolic syndrome incidence in Japanese adults. PLoS One. 2023;18(4):e0284139. Published 2023 Apr 7. doi:\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1371/journal.pone.0284139\u003c/span\u003e\u003cspan address=\"10.1371/journal.pone.0284139\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZahedi AS, Daneshpour MS, Akbarzadeh M, Hedayati M, Azizi F, Zarkesh M. Association of baseline and changes in adiponectin, homocysteine, high-sensitivity C-reactive protein, interleukin-6, and interleukin-10 levels and metabolic syndrome incidence: Tehran lipid and glucose study. Heliyon. 2023;9(9):e19911. Published 2023 Sep 6. doi:\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.heliyon.2023.e19911\u003c/span\u003e\u003cspan address=\"10.1016/j.heliyon.2023.e19911\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZou TT, Zhou YJ, Zhou XD, et al. MetS Risk Score: A Clear Scoring Model to Predict a 3-Year Risk for Metabolic Syndrome. Horm Metab Res. 2018;50(9):683\u0026ndash;689. doi:\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1055/a-0677-2720\u003c/span\u003e\u003cspan address=\"10.1055/a-0677-2720\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKarimi-Alavijeh F, Jalili S, Sadeghi M. Predicting metabolic syndrome using decision tree and support vector machine methods. ARYA Atheroscler. 2016;12(3):146\u0026ndash;152.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLee S, Lee H, Choi JR, Koh SB. Development and Validation of Prediction Model for Risk Reduction of Metabolic Syndrome by Body Weight Control: A Prospective Population-based Study. Sci Rep. 2020;10(1):10006. Published 2020 Jun 19. doi:\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1038/s41598-020-67238-5\u003c/span\u003e\u003cspan address=\"10.1038/s41598-020-67238-5\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSzabo de Edelenyi F, Goumidi L, Bertrais S, et al. Prediction of the metabolic syndrome status based on dietary and genetic parameters, using Random Forest. Genes Nutr. 2008;3(3\u0026ndash;4):173\u0026ndash;176. doi:\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1007/s12263-008-0097-y\u003c/span\u003e\u003cspan address=\"10.1007/s12263-008-0097-y\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLee S, Lee SK, Kim JY, Cho N, Shin C. Sasang constitutional types for the risk prediction of metabolic syndrome: a 14-year longitudinal prospective cohort study. BMC Complement Altern Med. 2017;17(1):438. Published 2017 Sep 2. doi:\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1186/s12906-017-1936-4\u003c/span\u003e\u003cspan address=\"10.1186/s12906-017-1936-4\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLi G, Esangbedo IC, Xu L, et al. Childhood retinol-binding protein 4 (RBP4) levels predicting the 10-year risk of insulin resistance and metabolic syndrome: the BCAMS study. Cardiovasc Diabetol. 2018;17(1):69. Published 2018 May 14. doi:\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1186/s12933-018-0707-y\u003c/span\u003e\u003cspan address=\"10.1186/s12933-018-0707-y\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYang H, Yu B, OUYang P, et al. Machine learning-aided risk prediction for metabolic syndrome based on 3 years study. Sci Rep. 2022;12(1):2248. Published 2022 Feb 10. doi:\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1038/s41598-022-06235-2\u003c/span\u003e\u003cspan address=\"10.1038/s41598-022-06235-2\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDaniel Tavares L, Manoel A, Henrique Rizzi Donato T, et al. Prediction of metabolic syndrome: A machine learning approach to help primary prevention. Diabetes Res Clin Pract. 2022;191:110047. doi:\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.diabres.2022.110047\u003c/span\u003e\u003cspan address=\"10.1016/j.diabres.2022.110047\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHirose H, Takayama T, Hozawa S, Hibi T, Saito I. Prediction of metabolic syndrome using artificial neural network system based on clinical data including insulin resistance index and serum adiponectin. Comput Biol Med. 2011;41(11):1051\u0026ndash;1056. doi:\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.compbiomed.2011.09.005\u003c/span\u003e\u003cspan address=\"10.1016/j.compbiomed.2011.09.005\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLiu W, Tang X, Cui T, Zhao H, Song G. Development and visualization of a risk prediction model for metabolic syndrome: a longitudinal cohort study based on health check-up data in China. Front Nutr. 2023;10:1286654. Published 2023 Nov 21. doi:\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3389/fnut.2023.1286654\u003c/span\u003e\u003cspan address=\"10.3389/fnut.2023.1286654\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAlberti KG, Eckel RH, Grundy SM, et al. Harmonizing the metabolic syndrome: a joint interim statement of the International Diabetes Federation Task Force on Epidemiology and Prevention; National Heart, Lung, and Blood Institute; American Heart Association; World Heart Federation; International Atherosclerosis Society; and International Association for the Study of Obesity. Circulation. 2009;120(16):1640\u0026ndash;1645. doi:\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1161/CIRCULATIONAHA.109.192644\u003c/span\u003e\u003cspan address=\"10.1161/CIRCULATIONAHA.109.192644\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMatsuzawa Y. Metabolic syndrome\u0026ndash;definition and diagnostic criteria in Japan. J Atheroscler Thromb. 2005;12(6):301. doi:\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.5551/jat.12.301\u003c/span\u003e\u003cspan address=\"10.5551/jat.12.301\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eExamination Committee of Criteria for 'Obesity Disease' in Japan; Japan Society for the Study of Obesity. New criteria for 'obesity disease' in Japan. Circ J. 2002;66(11):987\u0026ndash;992. doi:\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1253/circj.66.987\u003c/span\u003e\u003cspan address=\"10.1253/circj.66.987\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNyamdorj R, Pitk\u0026auml;niemi J, Tuomilehto J, et al. Ethnic comparison of the association of undiagnosed diabetes with obesity [published correction appears in Int J Obes (Lond). 2010;34(3):597] [published correction appears in Int J Obes (Lond). 2011;35(2):313-4]. Int J Obes (Lond). 2010;34(2):332\u0026ndash;339. doi:\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1038/ijo.2009.225\u003c/span\u003e\u003cspan address=\"10.1038/ijo.2009.225\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNishizawa H, Shimomura I. Population Approaches Targeting Metabolic Syndrome Focusing on Japanese Trials. Nutrients. 2019;11(6):1430. Published 2019 Jun 25. doi:\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3390/nu11061430\u003c/span\u003e\u003cspan address=\"10.3390/nu11061430\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eOka R, Kobayashi J, Yagi K, et al. Reassessment of the cutoff values of waist circumference and visceral fat area for identifying Japanese subjects at risk for the metabolic syndrome. Diabetes Res Clin Pract. 2008;79(3):474\u0026ndash;481. doi:\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.diabres.2007.10.016\u003c/span\u003e\u003cspan address=\"10.1016/j.diabres.2007.10.016\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBao Y, Lu J, Wang C, et al. Optimal waist circumference cutoffs for abdominal obesity in Chinese. Atherosclerosis. 2008;201(2):378\u0026ndash;384. doi:\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.atherosclerosis.2008.03.001\u003c/span\u003e\u003cspan address=\"10.1016/j.atherosclerosis.2008.03.001\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKim JA, Choi CJ, Yum KS. Cut-off values of visceral fat area and waist circumference: diagnostic criteria for abdominal obesity in a Korean population. J Korean Med Sci. 2006;21(6):1048\u0026ndash;1053. doi:\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3346/jkms.2006.21.6.1048\u003c/span\u003e\u003cspan address=\"10.3346/jkms.2006.21.6.1048\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYang X, Lin Y, Xu GD, et al. Optimal Cut-Off Values of Visceral Fat Area for Predicting Metabolic Syndrome Among Type 2 Diabetes Patients in Ningbo, China. Diabetes Metab Syndr Obes. 2021;14:1375\u0026ndash;1383. Published 2021 Mar 25. doi:\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.2147/DMSO.S304164\u003c/span\u003e\u003cspan address=\"10.2147/DMSO.S304164\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKim SH, Chung JH, Song SW, Jung WS, Lee YA, Kim HN. Relationship between deep subcutaneous abdominal adipose tissue and metabolic syndrome: a case control study. Diabetol Metab Syndr. 2016;8:10. Published 2016 Feb 12. doi:\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1186/s13098-016-0127-7\u003c/span\u003e\u003cspan address=\"10.1186/s13098-016-0127-7\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLee S, Kuk JL, Kim Y, Arslanian SA. Measurement site of visceral adipose tissue and prediction of metabolic syndrome in youth. Pediatr Diabetes. 2011;12(3 Pt 2):250\u0026ndash;257. doi:\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1111/j.1399-5448.2010.00705.x\u003c/span\u003e\u003cspan address=\"10.1111/j.1399-5448.2010.00705.x\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eShah RV, Murthy VL, Abbasi SA, et al. Visceral adiposity and the risk of metabolic syndrome across body mass index: the MESA Study. JACC Cardiovasc Imaging. 2014;7(12):1221\u0026ndash;1235. doi:\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.jcmg.2014.07.017\u003c/span\u003e\u003cspan address=\"10.1016/j.jcmg.2014.07.017\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCho SA, Joo HJ, Cho JY, et al. Visceral Fat Area and Serum Adiponectin Level Predict the Development of Metabolic Syndrome in a Community-Based Asymptomatic Population. PLoS One. 2017;12(1):e0169289. Published 2017 Jan 3. doi:\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1371/journal.pone.0169289\u003c/span\u003e\u003cspan address=\"10.1371/journal.pone.0169289\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMatsushita Y, Nakagawa T, Yamamoto S, et al. Effect of longitudinal changes in visceral fat area on incidence of metabolic risk factors: the Hitachi health study. Obesity (Silver Spring). 2013;21(10):2126\u0026ndash;2129. doi:\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1002/oby.20347\u003c/span\u003e\u003cspan address=\"10.1002/oby.20347\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNakaji S, Ihara K, Sawada K, et al. Social innovation for life expectancy extension utilizing a platform-centered system used in the Iwaki health promotion project: A protocol paper. SAGE Open Med. 2021;9:20503121211002606. Published 2021 Mar 19. doi:\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1177/20503121211002606\u003c/span\u003e\u003cspan address=\"10.1177/20503121211002606\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYamaguchi T, Ozato N, Katashima M, et al. A Novel Method to Visualize the Dietary Macronutrient Composition of Smaller Visceral Fat Accumulation. Front Nutr. 2020;6:194. Published 2020 Jan 24. doi:\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3389/fnut.2019.00194\u003c/span\u003e\u003cspan address=\"10.3389/fnut.2019.00194\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRyo M, Maeda K, Onda T, et al. A new simple method for the measurement of visceral fat accumulation by bioelectrical impedance. Diabetes Care. 2005;28(2):451\u0026ndash;453. doi:\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.2337/diacare.28.2.451\u003c/span\u003e\u003cspan address=\"10.2337/diacare.28.2.451\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eOzato N, Saito S, Yamaguchi T, et al. Association between Nutrients and Visceral Fat in Healthy Japanese Adults: A 2-Year Longitudinal Study Brief Title: Micronutrients Associated with Visceral Fat Accumulation. Nutrients. 2019;11(11):2698. Published 2019 Nov 7. doi:\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3390/nu11112698\u003c/span\u003e\u003cspan address=\"10.3390/nu11112698\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKinoshita K, Ozato N, Yamaguchi T, et al. The effect of age on the association between daily gait speed and abdominal obesity in Japanese adults. Sci Rep. 2021;11(1):19975. Published 2021 Oct 7. doi:\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1038/s41598-021-98679-1\u003c/span\u003e\u003cspan address=\"10.1038/s41598-021-98679-1\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eOzato N, Saito S, Yamaguchi T, et al. Blautia genus associated with visceral fat accumulation in adults 20\u0026ndash;76 years of age. NPJ Biofilms Microbiomes. 2019;5(1):28. Published 2019 Oct 4. doi:\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1038/s41522-019-0101-x\u003c/span\u003e\u003cspan address=\"10.1038/s41522-019-0101-x\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYamaguchi T, Ozato N, Katashima M, et al. A Novel Method to Visualize the Dietary Macronutrient Composition of Smaller Visceral Fat Accumulation. Front Nutr. 2020;6:194. Published 2020 Jan 24. doi:\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3389/fnut.2019.00194\u003c/span\u003e\u003cspan address=\"10.3389/fnut.2019.00194\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYouden WJ. Index for rating diagnostic tests. Cancer. 1950;3(1):32\u0026ndash;35. doi:\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1002/1097-0142(1950)3:1\u0026lt;32::aid-cncr2820030106\u0026gt;3.0.co;2-3\u003c/span\u003e\u003cspan address=\"10.1002/1097-0142(1950)3:1%3C32::aid-cncr2820030106%3E3.0.co;2-3\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMetz CE. Basic principles of ROC analysis. Semin Nucl Med. 1978;8(4):283\u0026ndash;298. doi:\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/s0001-2998(78)80014-2\u003c/span\u003e\u003cspan address=\"10.1016/s0001-2998(78)80014-2\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eVermont J, Bosson JL, Fran\u0026ccedil;ois P, et al. Strategies for graphical threshold determination. Comput Methods Programs Biomed. 1991;35(2):141\u0026ndash;150. doi:\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/0169-2607(91)90072-2\u003c/span\u003e\u003cspan address=\"10.1016/0169-2607(91)90072-2\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDeLong ER, DeLong DM, Clarke-Pearson DL. Comparing the areas under two or more correlated receiver operating characteristic curves: a nonparametric approach. Biometrics. 1988;44(3):837\u0026ndash;845.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLundberg S, Lee S-I. A unified approach to interpreting model predictions. Arxiv. 2017, \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://arxiv.org/abs/1705.07874\u003c/span\u003e\u003cspan address=\"https://arxiv.org/abs/1705.07874\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eUrashima M, Wada T, Fukumoto T, et al. Prevalence of metabolic syndrome in a 22,892 Japanese population and its associations with life style. Japan Medical Association Journal. 2005;48:441\u0026ndash;450.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNakanishi N, Takatorige T, Suzuki K. Cigarette smoking and the risk of the metabolic syndrome in middle-aged Japanese male office workers. Ind Health. 2005;43(2):295\u0026ndash;301. doi:\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.2486/indhealth.43.295\u003c/span\u003e\u003cspan address=\"10.2486/indhealth.43.295\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHiuge-Shimizu A, Kishida K, Funahashi T, et al. Absolute value of visceral fat area measured on computed tomography scans and obesity-related cardiovascular risk factors in large-scale Japanese general population (the VACATION-J study). Ann Med. 2012;44(1):82\u0026ndash;92. doi:\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3109/07853890.2010.526138\u003c/span\u003e\u003cspan address=\"10.3109/07853890.2010.526138\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMatthews BW. Comparison of the predicted and observed secondary structure of T4 phage lysozyme. Biochim Biophys Acta. 1975;405(2):442\u0026ndash;451. doi:\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/0005-2795(75)90109-9\u003c/span\u003e\u003cspan address=\"10.1016/0005-2795(75)90109-9\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWorld Health Organization. Obesity: preventing and managing the global epidemic. Report of a WHO consultation. World Health Organ Tech Rep Ser. 2000;894:1\u0026ndash;253.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJeon HH, Lee YK, Kim DH, Pak H, Shin SY, Seo JH. Risk for metabolic syndrome in the population with visceral fat area measured by bioelectrical impedance analysis. Korean J Intern Med. 2021;36(1):97\u0026ndash;105. doi:\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3904/kjim.2018.427\u003c/span\u003e\u003cspan address=\"10.3904/kjim.2018.427\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKim YA, Kwak SG, Cho YJ. Optimal cutoff values for visceral fat volume to predict metabolic syndrome in a Korean population. Medicine (Baltimore). 2021;100(36):e27114. doi:\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1097/MD.0000000000027114\u003c/span\u003e\u003cspan address=\"10.1097/MD.0000000000027114\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFujita T. The metabolic syndrome in Japan. Nat Clin Pract Cardiovasc Med. 2008;5 Suppl 1:S15-S18. doi:\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1038/ncpcardio0808\u003c/span\u003e\u003cspan address=\"10.1038/ncpcardio0808\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSairenchi T, Iso H, Yamagishi K, et al. Impact and attribute of each obesity-related cardiovascular risk factor in combination with abdominal obesity on total health expenditures in adult Japanese National Health insurance beneficiaries: The Ibaraki Prefectural health study. J Epidemiol. 2017;27(8):354\u0026ndash;359. doi:\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.je.2016.08.009\u003c/span\u003e\u003cspan address=\"10.1016/j.je.2016.08.009\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"},{"header":"Table","content":"\u003cp\u003eTable 1 is available in the Supplementary Files section\u003c/p\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"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":"","lastPublishedDoi":"10.21203/rs.3.rs-3996594/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-3996594/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003ch2\u003eBackground\u003c/h2\u003e \u003cp\u003eMetabolic syndrome (MetS) increases the risk of cardiovascular and lifestyle-related diseases. Therefore, early detection is important to prevent MetS. This study analysed the effects of visceral fat on MetS using health examination. A MetS onset prediction algorithm was developed.\u003c/p\u003e\u003ch2\u003eMethods\u003c/h2\u003e \u003cp\u003e Health examination data were obtained from the Iwaki Health Promotion Project conducted in Aomori Prefecture in Japan, wherein labels indicated the development of MetS within the three years (213 onset and 1320 non-onset cases). The data were divided into training and test data (8:2 ratio), and 18 onset prediction models were developed to support the SHapley Additive exPlanations (SHAP) value. The onset labels and non-invasive input data were used as the output and input variables, respectively. We selected the model with the highest area under the curve (AUC) score when conducting five-fold cross validation, and the AUC of the test data was calculated. Feature impact was calculated based on SHAP.\u003c/p\u003e\u003ch2\u003eResults\u003c/h2\u003e \u003cp\u003eThere were 169 and 1058 people in the metabolic and non-metabolic syndrome groups, respectively. The visceral fat area was significantly higher in the onset group than in the non-onset group (p\u0026thinsp;\u0026lt;\u0026thinsp;0.00001). The cut-off value based on the receiver operating characteristic curve was 82 cm\u003csup\u003e2\u003c/sup\u003e, and the AUC was 0.86. Machine learning was employed on six items reported to contribute to the onset of MetS in addition to visceral fat to build an onset prediction algorithm. The cross-validation AUC\u0026thinsp;=\u0026thinsp;0.90 and test AUC\u0026thinsp;=\u0026thinsp;0.88 indicated a high-accuracy algorithm. The visceral fat was found to be the main factor, as confirmed by conventional feature importance in machine learning.\u003c/p\u003e\u003ch2\u003eConclusions\u003c/h2\u003e \u003cp\u003eVisceral fat is crucial to determining the onset of MetS in the future. A high-accuracy onset prediction algorithm was developed based on non-invasive parameters, including visceral fat.\u003c/p\u003e","manuscriptTitle":"Effect of visceral fat on onset of metabolic syndrome","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2024-03-05 16:22:03","doi":"10.21203/rs.3.rs-3996594/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":"d8c715a7-1166-40b9-a99c-03aead1f7f31","owner":[],"postedDate":"March 5th, 2024","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[{"id":29141380,"name":"Health sciences/Health care/Disease prevention/Preventive medicine"},{"id":29141381,"name":"Health sciences/Medical research/Epidemiology"}],"tags":[],"updatedAt":"2024-03-07T17:56:29+00:00","versionOfRecord":[],"versionCreatedAt":"2024-03-05 16:22:03","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-3996594","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-3996594","identity":"rs-3996594","version":["v1"]},"buildId":"qtupq5eGEP_6zYnWcrvyt","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}