Neural Drug Discovery

In the field of drug production, machine learning methods have already established themselves as a reliable tool for performing various tasks, among which are the prediction of the structure of biologically active molecules and the initial selection of drug candidates. Predictive models based on machine learning have gained great importance almost at every stage of drug development process significantly reducing the financial and time costs for the discovery of a new drug. Despite this, to date there are only a few review papers covering their use in the preparation and production of new drugs. The main goal of this work is to explore the usage of machine learning methods in the field of drug discovery and development as well as to structure the knowledge about some of machine learning methods that are already used in the drug development process. Such an analysis of the current state of affairs in the chosen field will give us an idea of the future prospects for the development of chemical informatics, of the limitations that modern scientists face, as well as possible ways to overcome them. The second part of the work is devoted to experimental work with some of the commonly machine learning algorithms in drug discovery and development. We constructed a Neural Network and managed to tune some hyperparameters on the datasets to outperform some benchmarks.