Due to the lack of a large sample set, most digital human phantoms do not represent the statistical features of the large population. In our study, over a thousand whole-body CT images were collected from the hospitals across the country. Artificial intelligent methods and statistical shape modelling method was used to learn inter-subject anatomical variations from the segmented structures. The phantom includes over one hundred parameters to adjust the anatomical and physiological features, and it can be registered to personal photo or medical images to achieve individualized modelling.