Compound hepatotoxicity early prediction method based on deep learning and gene expression data

Abstract

The invention relates to the field of computer-aided drug screening, in particular to a compound hepatotoxicity early prediction method based on deep learning and gene expression data, which comprisesthe following steps: (1) mining and preprocessing compound toxicology gene expression data; (2) selecting hepatotoxicity characteristic genes; (3) establishing a deep learning prediction model basedon the expression data of the hepatotoxicity characteristic gene; and (4) performing parameter optimization and performance improvement of the model. According to the method, pharmacogenomics and an artificial intelligence deep learning algorithm are fully combined, so that the limitation of a traditional compound hepatotoxicity prediction method is overcome, and the early prediction of the delayed hepatotoxicity of the compound is systematically realized through a gene level; an efficient, accurate and rapid compound hepatotoxicity prediction method is provided for preclinical toxicity safetyassessment and clinical reasonable medication in a new drug research and development process.

Description

technical field [0001] The present invention relates to the field of computer-aided drug screening, in particular to a method for early prediction of compound hepatotoxicity based on deep learning and gene expression data, which is suitable for early prediction of compound hepatotoxicity based on gene expression data. Background technique [0002] Drug hepatotoxicity is an important factor leading to the failure of new drug research and the withdrawal of clinical drugs from the market. According to statistics, in the process of new drug research and development, the failure rate of candidate drugs due to hepatotoxicity is 37%, and the proportion of drugs withdrawn from the market due to drug hepatotoxicity in clinical application is 18%. Therefore, in the early stage of drug development and clinical use Predicting drug hepatotoxicity is of great significance for improving the success rate of research and development and rational drug use. Due to the complex mechanism of dru...

AI Technical Summary

Problems solved by technology

With the development of information technology, research at home and abroad has begun to use machine learning and compound structural features to establish computational models to predict drug liver toxicity, but it often faces the following problems: 1) It is easily limited by the structure of a single compound, and the structural diversity The prediction of hepatotoxicity of certain compounds is often less accurate; 2) The prediction results lack biological significance, and the prediction results cannot be systematically explained from the biological mechanism of action; 3) Early prediction of delayed drug hepatotoxicity cannot be performed; 4) Traditional The machine learning method cannot automatically learn feature information from big data, and requires a lot of manual feature selection

Images