a brief look at machine learning methods in chemical toxicity prediction

Abstract: the most important aim of modern toxicology efforts is based on the detection of chemical compounds which have the toxic potential. by introduction of newly synthesized chemical compounds, it is necessary to evaluate the toxicity behavior of these newly emerged compounds very quickly [1, 2]. usually, conformal prediction strategy is used for this purpose but the most challenge of this area is the uncertainty associated with predicted models in conformal predictor [3] therefore development and introduction of techniques with the ability of toxicity prediction is among the most challenges in scientific communities [3, 4]. nowadays, deep learning is used in order to overcome this problem [3]. deep forward neural networks and graph neural networks are two major deep learning techniques usually applied for computational toxicology. in the present paper challenges and recent advanced based on deep learning in the field of computational toxicology is discussed.