Accurate and complete medication lists are critical inputs to effective medication reconciliation to prevent medication prescribing and administration errors. Previous research aggregated structured medication data form multiple sources to generate and maintain a reconciled medication list. Medications documented in clinical texts also need to be reconciled. However, most reconciliation methods currently have limited capability to process textual data and temporal information (e.g., dates, duration and status). Our goal is to pilot and test methodologies and applications in the fields of natural language processing (NLP) and temporal reasoning to facilitate the use of electronic clinical texts in order to improve the """"""""correctness"""""""" and """"""""completeness"""""""" of medication lists. Clinic notes and free-text """"""""comments"""""""" fields in medication lists in an ambulatory electronic medical record system will be considered in the study. An NLP system and a temporal reasoning system will be adapted to automatically extract medication and associated temporal information from clinical texts and encode the medications using a controlled terminology. Multiple knowledge bases will be used to develop a mechanism to represent the timing of medication use, detect the changes (e.g., active or inactive), and then to organize medications into appropriate groups (e.g., by ingredient or by status). The feasibility and efficiency of the proposed methods and tools in improving the process of medication reconciliation will be assessed. Domain experts will serve as judges to assess the success of capturing, coding, and organizing the medications and temporal information and also to evaluate whether our methods are complementary to those currently used for medication management.
Accurate and complete medication information at the point of care is crucial for delivery of high-quality care and prevention of adverse events. Most previous studies aggregated structured medication data from EMR and CPOE (Computerized Physician Order Entry) systems to generate and maintain a reconciled medication list. However, medications in non-structured narrative sources (such as clinic notes and free-text comments) must also be reconciled. Structured data presented in a standard, predictable form can be easily processed by a computer. By contrast, narrative data does not have a well-defined structure, so processing such data is very challenging. Our goal is to pilot and test methodologies and applications in the fields of natural language processing (any system that manipulates text) and temporal reasoning (e.g., identifying the timing of medication use) to facilitate the use of electronic clinical texts in order to improve the correctness and completeness of medication lists. The feasibility and efficiency of the proposed methods and tools in improving the process of medication reconciliation will be assessed.
Zhou, Li; Mahoney, Lisa M; Shakurova, Anastasiya et al. (2012) How many medication orders are entered through free-text in EHRs?--a study on hypoglycemic agents. AMIA Annu Symp Proc 2012:1079-88 |
Zhou, Li; Plasek, Joseph M; Mahoney, Lisa M et al. (2012) Mapping Partners Master Drug Dictionary to RxNorm using an NLP-based approach. J Biomed Inform 45:626-33 |
Zhou, Li; Plasek, Joseph M; Mahoney, Lisa M et al. (2011) Using Medical Text Extraction, Reasoning and Mapping System (MTERMS) to process medication information in outpatient clinical notes. AMIA Annu Symp Proc 2011:1639-48 |