The MOMENT (Monitoring for Outpatient Medication Effects and New Toxicities) in TIME project will extend the in-progress TIME (Tools for Inpatient Monitoring using Evidence) for Safe and Appropriate Testing grant, #5 R01 LM007995-03, by developing sophisticated text-mining and data extraction tools to examine adverse drug effects in patients presenting for emergency and hospital care. Using a UMLS-based concept identifier enhanced with natural language processing, the MOMENT project will detect: (a) individual clinical manifestations (symptoms and physical exam findings) that are potentially drug related;(b) clinical syndromes (e.g., hepatoxicity, myopathy, renal insufficiency, glucose intolerance);and, (c) and clinical diseases (such as systemic lupus induced by hydralazine or procainamide, or acute myocardial infarction associated with rofecoxib). Innovative aspects of the proposed work include use of advanced natural language processing techniques to abstract concepts representing potential drug effects from patient history and physical examination records and ancillary test reports, the combination of evidence based templates and diagnostic algorithms to detect complex patterns of drug toxicity (e.g., hepatocelluar damage, pulmonary fibrosis, or acute coronary syndrome), advanced statistical methods for determining medication effects in large populations, and utilization of a clinical database containing data on more than 100,000 patients. Public Statement The MOMENT project (Monitoring for Outpatient Medication Effects and New Toxicities) will demonstrate the feasibility of combining informatics applications increasingly present in many healthcare institutions - electronic medical record systems and care provider order entry (CPOE) systems - to create an advanced detection and monitoring system for adverse medication effects. The project will develop tools and algorithms to detect the patient symptoms, physical exam signs, test results, and medical diagnoses that may indicate drug-induced injury.
Showing the most recent 10 out of 27 publications
