Abstract
Micronutrient deficiencies are underdiagnosed and have a significant negative impact on athletic performance, recovery, and long-term physiological health, yet they are detected through invasive biochemical measurements. This article introduces a non-invasive Artificial Intelligence (AI)-based system for early detection of micronutrient deficiencies by analyzing facial and tongue images. The framework is built around computer vision methods, transformer-based feature extraction (DINOv2), unsupervised clustering, and neural network classification for the detection of visual biomarkers like skin pallor, irregularities in texture, and tongue discoloration. A dual-feature face and tongue approach yields greater predictive accuracy. The system also gives personalized dietary advice based on the principles of precision nutrition. The proposed solution provides a scalable, cost-effective, and accessible way to continuously monitor athletes' nutrition and optimize performance.

This publication has 17 references indexed in Scilit: