Integrated bioinformatic and pharmacokinetic models of high-dimensional drug interactions

点击翻译按钮获取中文摘要

Project Summary Adverse drug events are the fifth leading cause of death in the United States, lead to increased morbidity, and are responsible for a large economic cost on the healthcare system. Polypharmacy is associated with increased risk of adverse events. We hypothesize that individuals taking multiple medications are at an increased risk of drug-drug interactions, leading to clinically relevant adverse events. Through a combination of computational data mining algorithms, statistical inference, and mechanistic pharmacology models, we seek to identify and evaluate clinically significant high dimensional drug interactions (HD-DDIs). We propose a novel frequent close itemset mining algorithm to identify candidate HD-DDIs with adverse reactions from large health record data sets. These HD-DDIs identified by the computational algorithm will be subjected to an innovative empirical Bayes statistical inference to determine this false positive, hence its statistical significance in its potential relevance of each interaction. As a large number of drug interactions are potentiated through the cytochrome P450 (CYP450) system, the mechanistic potential of interactions among multidrug regimens will be evaluated using in vitro metabolism assays. This innovative approach, combining graphical, statistical inference and mechanistic pharmacology models will provide insight into the role of polypharmacy in adverse drug events.