An approach for predicting the effects of endocrine disrupting chemicals on human health using deep learning

Introduction Recent years have seen considerable development in the fields of machine learning and neural networks in particular [1]. New architectures and methodologies allow for the use of data, such as sound and image, and for drawing conclusions based on these. Developments involving neural networks enable us to successfully address issues such as recognition, image categorization and recognition of objects depicted – in certain instances more efficiently than a human agent could [2].

There is ongoing development in neural network architectures [3]. Increased access to data combined with improved computing power have enabled researchers and programmers to find new applications for neural networks at a very rapid rate [4]. For many years, there have been various attempts at devising systems that can depict chemical structures; ‘SMILES’ is one of these systems of depiction which is commonly accepted in scientific cirles [5]. The Simplified Molecular-Input Line-Entry System (SMILES) is a standard which takes the form of a linear string describing chemical structures using short ASCII strings. Most molecular processors can be input with SMILES strings and transform those back to 2D depictions or 3D molecular patterns [6, 7].

Chemical structure images will be used as input in neural networks which will attempt to model their response to human tissue and cells. Our aim was to determine whether this specific information could be used for modelling purposes.