J Diabetes Investig. 2025;16(2):272-284.

Artificial intelligence-based body composition analysis using computed tomography images predicts both prevalence and incidence of diabetes mellitus.

, , , , , , ,

제1저자 김유혁, 윤지완 · 교신저자 조영민

PMID 39576146 PMC11786173 DOI 10.1111/jdi.14365 IF 3.6 내분비대사내과

소속 (7)

Department of Internal Medicine, Seoul National University Hospital, Seoul, Korea.
Department of Internal Medicine, Seoul National University Hospital Healthcare System Gangnam Center, Seoul, Korea.
Department of Radiology, Seoul National University College of Medicine, Seoul, Korea.
Department of Radiology, Seoul National University Hospital, Seoul, Korea.
Department of Internal Medicine, Seoul National University Bundang Hospital, Seongnam, Korea.
Department of Internal Medicine, Seoul National University College of Medicine, Seoul, Korea.
Institute on Aging, Seoul National University, Seoul, Korea.

Abstract

AIM/INTRODUCTION We assess the efficacy of artificial intelligence (AI)-based, fully automated, volumetric body composition metrics in predicting the risk of diabetes.

MATERIALS AND METHODS This was a cross-sectional and 10-year retrospective longitudinal study. The cross-sectional analysis included health check-up data of 15,330 subjects with abdominal computed tomography (CT) images between January 1, 2011, and September 30, 2012. Of these, 10,570 subjects with available follow-up data were included in the longitudinal analyses. The volume of each body segment included in the abdominal CT images was measured using AI-based image analysis software.

RESULTS Visceral fat (VF) proportion and VF/subcutaneous fat (SF) ratio increased with age, and both strongly predicted the presence and risk of developing diabetes. Optimal cut-offs for VF proportion were 24% for men and 16% for women, while VF/SF ratio values were 1.2 for men and 0.5 for women. The subjects with higher VF/SF ratio and VF proportion were associated with a greater risk of having diabetes (adjusted OR 2.0 [95% CI 1.7-2.4] in men; 2.9 [2.2-3.9] in women). In subjects with normal glucose tolerance, higher VF proportion and VF/SF ratio were associated with higher risk of developing prediabetes or diabetes (adjusted HR 1.3 [95% CI 1.1-1.4] in men; 1.4 [1.2-1.7] in women). These trends were consistently observed across each specified cut-off value.

CONCLUSIONS AI-based volumetric analysis of abdominal CT images can be useful in obtaining body composition data and predicting the risk of diabetes.

Keywords

Artificial intelligenceBody compositionDiabetes mellitus

Figures

**Figure 1** Proportions of body segments by age overall (a) and in BMI subgroups (b). All analyses are stratified by sex. From left to right, the colors of the bars denote the following body segments in order: SF, VF, IO, skin, CNS, bone, and SM. Details on the proportions of body segments can also be found in Table S5. CNS, central nervous system; IO, internal organs; SF, subcutaneous fat; SM, skeletal muscle; VF, visceral fat.

**Figure 2** Changes in VF proportion (a) and VF/SF ratio (b) by age. For men, points are represented by circles and connected with solid lines. For women, points are represented by triangles and connected with dotted lines. Points and lines are colored according to BMI. BMI, body mass index; SF, subcutaneous fat; VF, visceral fat.

**Figure 3** Summary of receiver operating characteristic analysis for the prevalence of diabetes in men (a) and women (b). Lines are color coded for each model. AUC, areas under the ROC curves; CNS, central nervous system; IO, internal organs; ROC, receiver operating characteristic; SF, subcutaneous fat; SM, skeletal muscle; VF, visceral fat. **P‐value <0.05 vs VF proportion.

**Figure 4** Median progression time of subjects with NGT at baseline to prediabetes or diabetes by volumetric VF/SF ratio in men (a) and women (b). Lines are color coded for each model. Log‐rank P‐values comparing Kaplan–Meier curves are shown in the lower left corner. Results from the adjusted Cox proportional hazard model are shown in the upper right corner. The numbers below each plot indicate the number of subjects at risk and the cumulative number of censored subjects. *Adjusted for age, body mass index, abdominal obesity (male ≥90 cm, female ≥85 cm), hypertension (systolic blood pressure ≥130 mmHg or diastolic blood pressure ≥ 85 mmHg), hypertriglyceridemia (≥150 mg/dL), and low–high‐density lipoprotein cholesterol (male <40 mg/dL, female <50 mg/dL). CI, confidence interval; HR, hazard ratio; NGT, normal glucose tolerance; P, P‐value; SF, subcutaneous fat; VF, visceral fat.

그림 출처: PubMed Central 공개 논문 (PMC11786173)

Conflict of interest statement

YHK: No relevant relationship. JWY: No relevant relationship. BHL: No relevant relationship. JHY reports the honoraria from Bayer, Philips Healthcare, and Accuzen. JML reports the grants from Bayer Healthcare, Canon Healthcare, Philips Healthcare, GE Healthcare, CMS, Guerbet, Samsung Medison, Bracco, Starmed, RF Medical, and Dongkuk; the consulting fees from Samsung Medison; and honoraria from Samsung Medison, Philips, GE, Bayer, and Guerbet. YMC reports the grant from Daewoong Pharmaceuticals and the consulting fees from LG Chemicals and Sanofi. YMC is an Editorial Board member of the Journal of Diabetes Investigation and a co‐author of this article. To minimize bias, he was excluded from all editorial decision‐making related to the acceptance of this article for publication. Approval of the research protocol: This study was approved by the Institutional Review Board and the Big Data Review Board of Seoul National University Hospital. Informed consent: Written informed consent was obtained from all study subjects at each visit to the health screening program. Approval date of registry and the registration no. of the study/trial: Date: 04/27/2021 and IRB No. 2102–176‐1200. Animal studies: N/A.

그 밖의 정보