Through the statistics of digital twin related documents in the WebofScience core collection database, the bibliometric research method is adopted, and the observation points are the number of published papers and the distribution of documents, the cooperation between authors and research institutions, document citations, and the co-occurrence of keywords. This paper also designs a non-parametric anthropometric graph convolutional neural network that does not depend on any parametric body model, only needs to input a body mask image and a small number of anthropometric dimensions to explicitly predict the 3D body vertex coordinates. More than 95% of consumers have a strong desire to try 3D virtual fittings, 80% of consumers support the establishment of virtual fitting rooms in physical stores, and more than 60% of the respondents expressed their willingness to upload their own photos or human body data to construct the efficient model.