Machine Learning-Based Prediction of Optical Properties in Chromophoric Organic Compounds

A machine learning method is proposed to predict the properties of chromophores that induce skin inflammation due to UV radiation in sunlight. A publicly available dataset of chromophore properties is utilized to train a lightGBM model, enabling the prediction of the maximum molar extinction coefficient for new organic materials under consideration for development. The suggested approach enhances the efficiency of product development, as per photosafety evaluation guidelines, which specify that additional evaluation is unnecessary when the maximum molar extinction coefficient is below the threshold.