Micronutrient Deficiency Detection with Fingernail Images Using Deep Learning Techniques

Micronutrient deficiencies have a significant impact on society's overall health and well-being, necessitating supplementation. Clinical equipment is required to measure micronutrients such as iron, zinc, and vitamins using existing clinical methods. In rural and low-income areas, there is a trade-off between infrastructure, precision, invasiveness, and cost. Invasive blood sampling in paediatric patients is more uncomfortable, in addition to being cost prohibitive. There is a link between micronutrient deficiencies and subjective assessments of anatomical regions like finger nail beds. Artificial intelligence can detect micronutrient deficiency by looking at the colour and metadata of the fingernails. Convolutional neural network models are used in the proposed deep learning model to extract the features of the nails and predict the corresponding deficiencies. Various architectures such as ResNet, SqueezeNet, DenseNet, VGG and custom model are used to detect cutaneous abnormalities such as Melanonychia, Mycotic, and Buea's lines. The real-time images are used to train the model, which is then validated with sample images for increased accuracy. With images collected from web crawlers, the proposed non-invasive model has an accuracy of 94%, and performance can be further improved by training with large variants of data and data augmentation. This prediction mechanism is used in primary health care settings and is linked to frontline workers for quick and easy diagnosis