| Title | Analysis of Serum Fatty Acids and Vitamin D with Dimension Reduction Methods |
| Authors | Chen, Yifan and Shrestha, Rojeet and Chen, Zhen and Chiba, Hitoshi and Hui, Shu-Ping and Okada, Emiko and Ukawa, Shigekazu and Nakagawa, Takafumi and Nakamura, Koshi and Tamakoshi, Akiko and Minami, Hiroyuki and Mizuta, Masahiro |
| Year | 2020 |
| Volume | Archives of Data Science, Series A 6(1) / 2020 |
| Abstract | Fatty acid plays an important role in human health and fat-related diseases. A comprehensive analysis of diverse fatty acids in serum naturally results in a multi-variable, high-dimensional dataset, and, therefore, multivariate analysis, especially dimension reduction, should be considered to extract useful information. In this study, three basic dimension reduction methods including factor analysis, principal component analysis, and independent component analysis were conducted on total and free fatty acid datasets in a general Japanese population (N=545; men:women=245:300). These analyses successfully characterized fatty acid datasets, reflecting their physicochemical natures, metabolisms, and food sources. Factor analysis and principal component demonstrated the association of 𝜔-3 fatty acids (20:5 and 22:6) with 25- hydroxyvitamin D3 (vitamin D), suggesting fish oil as their common source of vitamin D. We conclude that dimension reductions can serve as a useful tool to extract valuable information from complex datasets of fatty acids and vitamin D in the aspect of health care and disease control. |