Disease drug prediction method based on heterogeneous network embedding model

Abstract

The invention discloses a disease drug prediction method based on a heterogeneous network embedding model. The method comprises a data acquisition module, a data preprocessing module, a path design module, a model training module and a result evaluation module. The data acquisition and data preprocessing module constructs a disease-gene-drug heterogeneous network through interaction data acquisition among diseases, genes and drugs. And the path design module finds a shortest path between genes through a shortest meta path strategy to form a random walk path. The model training module performs model training on the random walk data to form vectorized expression, and predicts interaction between drugs and diseases through Euclidean distance. And the result evaluation module evaluates the prediction effect by adopting an ROC curve based on a confusion matrix, and selects the optimal prediction effect through an optimization model. According to the method, through learning topology and potential expression in a biological heterogeneous network, prediction of the drug and disease relationship is realized.

Description

technical field [0001] The invention belongs to the technical field of data mining, and in particular relates to a disease drug prediction method based on a heterogeneous network embedding model. Background technique [0002] Drug development is a difficult and often costly process, characterized by its complex, time-consuming, and expensive nature. Despite a surge in funding for biomedical and pharmaceutical research in recent years, the U.S. Food and Drug Administration approves only a limited number of new treatments each year. There are many factors that lead to the slow progress of FDA's new drug approval, and the classic drug development hypothesis of "one gene, one drug, one disease" is the most critical and often overlooked important factor. Therefore, it is necessary to study from multiple perspectives in the biological system and explore the interactions between complex diseases in order to effectively carry out drug development. The effects of drug targets and d...

AI Technical Summary

Problems solved by technology

[0004] The purpose of the present invention is to provide a disease drug prediction method based on heterogeneous network embedding model, to solve the problem that the biological heterogeneous network with sparse data cannot use meta-paths to predict the relationship between heterogeneous nodes, and at the same time, the invention can eliminate the randomness of clinical trials. Sex, reduce the cycle of drug development

Images