[4F2-GS-10] AI application

[4F2-GS-10-01] Building and Evaluating Prediction Models for Skin Constitutions Using Skin Measurement Data and SNP Data of Japanese Women

〇Tatsuya Yoshikawa¹, Kumiko Kikuchi¹, Motoki Ooguri¹, Chie Yasuda¹, Satoshi Amano¹ (1. Shiseido Co., Ltd.)

Keywords:Gene, Single Nucleotide Polymorphism, Skin Phenotypes, Statistical Learning, Regularization

Background: Skin changes daily due to the complex interaction of multiple genetic and environmental factors. Although GWAS (genome-wide association studies) of skin diseases and skin-related genes have been conducted, there have been no reports that have actually measured the skin of Japanese women and analyzed the effects of environmental factors in addition to the polygenic relationship of genes. Objective: To develop an original evaluation algorithm to realize a new DNA testing method that proposes optimal care according to a person's innate skin characteristics (skin constitution) of humans. Results: Five to ten optimal combinations of SNPs (single nucleotide polymorphisms) were detected for each skin type, such as the ease and difficulty of pigmented spots. Conclusion: We constructed and evaluated models using skin measurement data and SNP data of 1,472 Japanese women, and elucidated the relationship between genetic and environmental factors behind differences in innate skin characteristics (skin constitution).