Construction of a diagnostic model for fatty liver using human body composition analysis
10.3760/cma.j.cn501113-20230731-00024
- VernacularTitle:利用人体成分分析构建脂肪肝诊断模型
- Author:
Ying ZHANG
1
;
Wentao KUAI
;
Yongzhan ZHANG
;
Yuanshen SONG
;
Denghua HE
;
Jiajia PEI
;
Liang XU
Author Information
1. 天津市第二人民医院，天津 300192
- Keywords:
Fatty liver;
Human body composition analysis;
Diagnostic model
- From:
Chinese Journal of Hepatology
2023;31(12):1277-1282
- CountryChina
- Language:Chinese
-
Abstract:
Objective:To construct a diagnostic model for fatty liver using body composition analysis and further evaluate the diagnostic effect of the model on fatty liver.Methods:726 cases with chronic liver disease who visited Tianjin Second People's Hospital from April 2019 to June 2022 and had body composition analysis tests were retrospectively enrolled and were divided into a fatty liver group (551 cases with fatty liver) and a control group (175 cases without fatty liver) according to the measured values of abdominal ultrasound and controlled attenuation parameter. An independent sample t-test and a non-parametric rank sum test were used for statistical processing. Logistic regression was used to construct a diagnostic model. Hosmer-Lemeshow was used to validate the fit of model. Receiver operating characteristic curve was used to confirm the diagnostic efficiency of the model. In addition, 341 cases of chronic liver disease who visited Tianjin Second People's Hospital were included to further verify the application effect of the model between July 2022 and February 2023.Results:Compared with the control group, the differences in various indicators of body composition analysis in the fatty liver group were statistically significant ( P < 0.05). Basal metabolic rate (X1), visceral fat area (X2), and body fat (X3) were eventually included in the diagnostic model for BCA-FL (body composition analysis-fatty liver)= -7.771+0.002X1-0.035X2+0.456X3 with the Hosmer-Lemeshow test (P=0.059). The measured area under the receiver operating characteristic curve, the sensitivity, and the specificity were 0.888, 0.889, and 0.726, respectively, when the diagnostic threshold value was 0.615 with the Youden index and the receiver operating characteristic curve. In the validated model group, the area under the receiver operating characteristic curve, Youden index, sensitivity, and specificity were 0.875, 0.624, 0.799, and 0.825, respectively. Conclusion:The diagnostic model BCA-FL for fatty liver constructed using human body composition analysis has good diagnostic efficacy and is suitable for screening fatty liver in different basic liver disease populations.
