Body composition prediction by multivariate statistical modeling

The assessment of human body composition (BC) is important for evaluating health and nutritional status. Increased fat mass is associated with an increased risk of metabolic diseases; declined lean mass or fat-free mass is also directly related to health and particularly with the mortality rate. More importantly, reduction in lean mass occurs together with an increase of body fat during aging; therefore assessing these changes in BC may be important because the study will lead to a pre-diagnosis for the prevention of morbidity and mortality risk. Accurate BC measurements can be obtained from DXA and others, but their applications require fixed equipment and are time consuming. Therefore, they are not convenient for routine clinical examinations. Potential uses of statistical methods for BC assessment have been highlighted, we presented in this book a linear and Bayesian network modeling for predicting simultaneously body, trunk and appendicular fat and lean masses based on anthropometric covariables. The main advantages in our proposed multivariate approach consisted in using very simple covariables and enabling to take into account the correlation structure between the response.

Ph.D in Applied Statistics, AgroParisTech, France, 2013. From 2014, I'm working at Affiliated Zhongshan Hospital of Dalian University as a Research Fellow. My principal research areas are Bayesian methods (especially Bayesian Networks modeling) applied in clinical setting.