The present invention relates to a method for predicting customized anticancer drug resistance using meta-analysis and biological pathway-based machine learning, and more particularly, a genomic cohort of acquired drug resistance and endogenous drug resistance that is publicly available for automatic detection of acquired drug resistance. Cohorts) information to screen and merge resistance-related cohorts, based on this, a personalized anticancer drug with very high accuracy and high generalizability using penalty regression combined with a personalized path score algorithm A resistance model has been established, and the model of the present invention is capable of predicting acquired drug resistance as well as predicting transferable prediction between endogenous drug resistance and acquired drug resistance. As a result of developing and validating a multivariate predictive model of acquired taxane resistance using the customized anticancer drug resistance prediction method using the meta-analysis and biological pathway-based machine learning of the present invention, the acquired taxane resistance model is 1.000 AUPRC (area under the precision- recall curve), a Brier score of 0.007, sensitivity and specificity of 100%, and AUROC (Area Under Receiver Operating Characterisic) of 1.000 were confirmed to exhibit perfect performance.
展开▼
