Relationship between minimal phototoxic dose and skin colour plus sun exposure history: a neural network approach.

Abstract

Before beginning PUVA-therapy it is important to accurately gauge an individual's degree of psoralen photosensitivity. This determination is usually based on an individual's skin phototype or minimal phototoxic dose. Since minimal phototoxic dose is technically complex and time consuming to measure, sun reactivity skin phototype is often used instead; however, it has recently been shown that skin phototype lacks specificity as a predictor of an individual's minimal phototoxic dose. In this study, an artificial neural network was developed to attempt to predict the minimal phototoxic dose from skin colour measurements combined with skin phototype. Our results showed that minimal phototoxic dose was predicted with an error less than 1 J/cm2 in only about half the subjects. In conclusion, minimal phototoxic dose probably cannot be predicted with sufficient accuracy on the basis of skin colour and skin phototype alone.