Application of Machine Learning Technology in the Prediction of ADME- Related Pharmacokinetic Parameters

Background: As an important determinant in drug discovery, the accurate analysis and acquisition of pharmacokinetic parameters are very important for the clinical application of drugs. At present, the research and development of new drugs mainly obtain their pharmacokinetic parameters through data analysis, physiological model construction and other methods, but the results are often quite different from the actual situation, needing more manpower and material resources. Objective: We mainly discuss the application of machine learning technology in the prediction of pharmacokinetic parameters, which are mainly related to the quantitative study of drug absorption, distribution, metabolism and excretion in the human body, such as bioavailability, clearance, apparent volume of distribution and so on. Methods: This paper first introduces the pharmacokinetic parameters, the relationship between the quantitative structure-activity relationship model and machine learning, then discusses the application of machine learning technology in different prediction models, and finally discusses the limitations, prospects and future development of the machine learning model in predicting pharmacokinetic parameters. Results: Unlike traditional pharmacokinetic analysis, machine learning technology can use computers and algorithms to speed up the acquisition of pharmacokinetic parameters to varying degrees. It provides a new idea to speed up and shorten the cycle of drug development, and has been successfully applied in drug design and development. Conclusion: The use of machine learning technology has great potential in predicting pharmacokinetic parameters. It also provides more choices and opportunities for the design and development of clinical drugs in the future.