Automated Detection of Vitamin Deficiencies in Nails and Skin Via Hyper Vision Transformer Networks

A content of this paper is complementary to Artificial Intelligence application for the web which examine vitamin deficiencies in body organs from the images of these organs. Modern techniques for identifying deficiencies of vitamins require costly laboratory low analysis. Vitamin deficiency is also of wide range, and can be exhibited in one or more visible features and symptoms in different body regions of a human body. Using photographs of skin and nails, the app can provide information on potential vitamin deficiencies and it does so without the need for blood tests. Then, the app helps in generating a list of nutritional sources targeting the discovered deficiency and the proposed consequences concerning nutritional micro-correction. In this study, we use a model, called the Hyper Vision Transformer (HVT), for image classification. This model is applied on total four classes that includes "Normal Nail", "Normal Skin", "Vitamin C Deficiency Nail" and "Vitamin B12 Deficiency Skin". We generate a detailed classification report to assess the performance of the model for identifying different deficiency states from nail and skin images.