Evaluation of multiple medication exposures concurrently using a novel algorithm

点击翻译按钮获取中文摘要

PROJECT SUMMARY The development of large observational health databases (OHD) has expanded the data available for analysis by pharmacoepidemiology research. The efficiency of these studies may be improved by simultaneously studying the association of multiple medications with a disease of interest. Unfortunately, prior research has demonstrated that it is difficult to distinguish true-positive from false-positive results when studying multiple exposures simultaneously, thus limiting the conclusions drawn from these types of studies and representing a major gap in the field. The objective of this proposal, which is the first step in achieving the applicant's long- term goal of improving the diagnosis and treatment of gastrointestinal diseases using insights derived from OHD, is to evaluate and validate medication class enrichment analysis (MCEA), a novel set-based signal-to- noise enrichment algorithm developed by the applicant to analyze multiple exposures from OHD with high sensitivity and specificity. The central hypothesis of this proposal is that MCEA has equal sensitivity and greater specificity compared to logistic regression, the most widely used analytic method for OHD, for identifying true associations between medications and clinical outcomes. The applicant will complete the following two interrelated specific aims to test the hypothesis: Aim 1 – to calculate the sensitivity and specificity of medication class enrichment analysis (MCEA) and logistic regression (LR) for identifying medication associations with Clostridium difficile infection (CDI) and Aim 2 – to calculate the sensitivity and specificity of MCEA and LR for identifying medication associations with gastrointestinal hemorrhage (GIH). The rationale for these aims is that by reproducing known medication-disease associations without false positives, MCEA can be used to identify novel pharmacologic associations with gastrointestinal diseases in future studies. The expected outcome for the proposed research is that it will demonstrate MCEA as a valid method for pharmacoepidemiology research, opening new research opportunities for the study of multi-exposure OHD. These new research opportunities may lead to more rapid identification of potential pharmacologic causes of emerging diseases and discovery of unanticipated beneficial medication effects, allowing such medications to be repurposed for new indications. To attain the expected outcome, the applicant will complete additional coursework that builds on his Master of Science in Clinical Epidemiology to learn computational biology, machine learning, and econometrics techniques. With the support of this grant and his institution, he will also directly apply these techniques to pharmacoepidemiology applications under the close mentorship of a carefully selected team of faculty with extensive experience in gastroenterology, pharmacoepidemiology, medical informatics, and mentoring prior K-award grant recipients. Through these activities, the applicant will develop the skills necessary to obtain NIH R01-level funding and become a leader in developing novel techniques for application to the epidemiologic study of gastrointestinal diseases.