Proposal and Trial of Prediction Methods for Skin Analyses with Chat GPT-4o using Past Skin Data and Subjects’ Daily Data related to the Skin Data as Teacher Datasets

In recent years, aquaphotomics-based research has been actively conducted across various fields to measure, analyze, and collect spectral data related to the moisture content of diverse subjects, including plants, animals, beauty products, concrete, and cement. However, furthermore, particularly in recent aquaphotomics-oriented scientific fields, effectively categorizing theses datasets using diverse system-based methodologies is essential. Aquaphotomics, a subfield of near-infrared spectroscopy (NIRS), has been extensively explored to investigate water-related spectral properties. However, research integrating AI-driven analytical techniques, particularly for human-related datasets, remains limited. In the fields related to aquaphotomics, to date, little to no research has been conducted on utilizing AI including Large language Models (LLM)-based technologies to process and analyze multiple datasets from human subjects. This study seeks to bridge this gap by employing Generative Pre-trained Transformer (GPT)- 4o to process and interpret aquaphotomics data, thereby facilitating more effective and comprehensive clustering into moisture content variations in human skin. Specifically, we divide subjects into four groups based on their skin moisture content and the ratio of epidermal water loss datasets. We then ask, based on the subjects’ basic information, skin care habits, and lifestyle information concerning exercise, sleep, etc., “Which group should we predict that any given, new subject will belong to?” and “What is the probability that those predictions are correct?” We also specify the process of making those predictions. By presenting this set of analytical data in numerical form to both the user and the esthetic salon, we hope that both the user and the esthetic salon will be able to make informed and satisfactory decisions regarding skin care routines, etc. This includes selecting appropriate serums and supplements, which should improve both the accuracy and effectiveness of skin care choices. In the long term, we anticipate that this research will contribute to the development of generative AI applications in skin care science, which is still evolving.