Related Experiment Video
Updated: Jan 12, 2026
Segmentation and Linear Measurement for Body Composition Analysis using Slice-O-Matic and Horos
Published on: March 21, 2021
Development of a machine learning model for predicting pre-sarcopenia in adults with abdominal obesity: a
Xixiang Wang1, Yu Liu1, Xiuwen Ren1
1School of public health, Capital Medical University; Beijing Key Laboratory of environment and aging; China-British Joint Laboratory of Nutrition Prevention and Control of Chronic Diseases, Capital Medical University, Beijing, 100069, People's Republic of China.
Purpose:
The loss of skeletal muscle mass and function is closely related to various metabolic disorders in vivo. The primary aim of this study is to explore the associations of blood biochemical indexes and age with the risk of pre-sarcopenia and pre-sarcopenia abdominal obesity (PSAO) in middle-aged and older individuals. Using machine learning algorithms, we further aim to identify key predictive factors for pre-sarcopenia and PSAO, and to validate the utility of these factors in risk assessment. The findings are expected to provide scientific evidence for personalized screening and intervention strategies targeting pre-sarcopenia and PSAO.
Methods:
A cross-sectional investigation was conducted on 2427 subjects. Muscle mass was measured by the bioelectrical impedance technique. Participants were categorized into three groups based on skeletal muscle index (SMI) and body composition: the control group, the pre-sarcopenia group and the PSAO group. Shapley additive explanations plots were used to display the contribution and influence of parameters on the model output. Restricted cubic splines were applied to describe the relation of variables with the risk of pre-sarcopenia and PSAO. The fivefold cross-validation was applied for internal validation.
Results:
Subjects in the abdominal obesity group showed higher blood triglyceride (TG) levels than the participants in the control group. In the pre-sarcopenia group, blood TG level negatively correlated with skeletal muscle index (SMI). The results of SHAP showed that the first three variables for the development of PSAO were age, TG, and serum creatinine (Scr). RCS results showed that an increase in age (> 47 years old for the controls, aged 45-50 for sarcopenia patients) and blood TG level (> 1.3 mmol/L for the controls, > 1.35 mmol/L for sarcopenia patients) significantly increased the risk of pre-sarcopenia or PSAO in male subjects.
Conclusion:
The risk of pre-sarcopenia and PSAO is closely related to age, sex, and blood TG level. Age and blood TG are suitable parameters for predicting the risk of pre-sarcopenia and PSAO in Chinese adults.
