Study Of Drug Development Based On Machine Learning

In recent years, reckless behaviors of human beings destroy the environment, which, in turn, causes increasing harm to human bodies. Since a lot of new diseases as well as traditional diseases cannot be eradicated, many new drugs are badly needed to be developed. This is a great pressure to workers who are engaged in drug research and development. Development process of traditional drug is very long and complex, and a lot of candidate drugs are eliminated at the stage of clinical trials. Chemical compounds, used as drugs, face a lot of challenges while producing effect on organisms. Many specific drugs show high activity in the experiment, but when they take effect on the organisms directly, unexpected results, even the serious side effects will happen due to the complexity of the organisms and uncertainty of the effects produced by compounds on the organisms. Therefore, it is very necessary and quite important to improve the efficiency of the development of candidate drugs. During the development of drugs, accurate matching of the receptor is strictly required, and thus the efficiency of research andvdevelopment of a candidate drug will be increased. At the same time, the biological mechanism of action is complex, so it is critical to test whether the drug could be suitable for the pharmacokinetic model or not. If the drug is suitable for the model, it will help to reduce the rate of elimination of candidate drugs in clinical trials.This paper, by means of machine learning which is widely used currently, combines with the basic properties of drugs, establishes a prediction model to improve the success rate of drug candidate drugs. Chapter one builds the interaction network, and makes use of the known interaction data and employs network reasoning to predict the interaction between new drugs and targets. Chapter two focuses on the study of inhibition of human cytochrome P450 enzymes, because the human cytochrome enzymes P450 dominates the most oxidative metabolic reactions of drugs in organisms, and the inhibition cytochrome enzyme P450 exhibits or not is directly related to the practical value of drugs. So, in the drug development, the study of the inhibition of this enzyme is very important for candidate drugs. Chapter three establishes a machine learning model to predict the water soluble compounds from the point of the basic compound property-solubility because many drugs on the market belong to the oral drugs, and their absorption and metabolism in organisms are closely related to its solubility.