A joint project of the Graduate School, Peabody College, and the Jean & Alexander Heard Library

Title page for ETD etd-03272016-201108


Type of Document Dissertation
Author Smith, Joshua Carl
URN etd-03272016-201108
Title Adverse Drug Effect Detection for Clinical Decision Support
Degree PhD
Department Biomedical Informatics
Advisory Committee
Advisor Name Title
Randolph A. Miller Committee Chair
Dan M. Roden Committee Member
Joshua C. Denny Committee Member
Kevin B. Johnson Committee Member
Qingxia Chen Committee Member
Keywords
  • clinical decision support
  • natural language processing
  • adverse drug effects
  • drug knowledgebases
  • ADE
  • ADR
Date of Defense 2016-03-15
Availability unrestricted
Abstract
Unrecognized adverse drug effects (ADEs) pose a serious clinical problem. They negatively affect patients’ health and quality of life, generate preventable emergency department visits and hospital admissions, and increase healthcare costs. Unfortunately, ADE detection is difficult. As more drugs are approved and new ADEs are discovered, it becomes increasingly onerous for providers to learn and recognize all the potential ADEs their patients might experience. This dissertation project partially addresses these problems. It developed both an ADE knowledgebase and a system to detect potential ADEs in hospital admission notes. The Drug Evidence Base (DEB2) is an accurate, machine- processable medication indication and ADE knowledge base. The DEB2 is automatically derived and updated from reliable public sources. The sources include the National Drug File-Reference Terminology, MEDLINE, MedlinePlus, DrugBank, and the FDA structured product labels (accessed via their SIDER2 extracts). The DEB2 contents require cross- validation from at least 2 of 5 of the sources. The Adverse Drug Effect Recognizer (ADER) is novel clinical decision support system that notifies clinicians about adult inpatients’ previously unrecognized symptomatic ADEs. It uses natural language processing of electronically stored admission history and physical exam notes. The ADER system identifies patients’ preadmission medications and clinical manifestations (diseases, syndromes, symptoms, findings, etc.), and compares them to a set of known ADEs, derived from the DEB2. It alerts appropriate care providers about any potential ADEs shortly after admission. A pilot study on the Internal Medicine wards of Vanderbilt University Hospital showed that the ADER system could help physicians recognize ADEs from outpatient medications presenting at admission. Providers who received ADER alerts were more likely to hold or discontinue inpatient medications after they responded to alerts and more likely to hold or discontinue suspected ADE-causing medications at discharge compared to a retrospective control group. The DEB2 and ADER methodologies have the potential to improve both recognition and treatment of ADEs for any hospital system using electronic records. Addressing previously unrecognized ADEs has the potential to reduce costs and improve patient care.
Files
  Filename       Size       Approximate Download Time (Hours:Minutes:Seconds) 
 
 28.8 Modem   56K Modem   ISDN (64 Kb)   ISDN (128 Kb)   Higher-speed Access 
  smith.pdf 9.07 Mb 00:41:58 00:21:35 00:18:53 00:09:26 00:00:48

Browse All Available ETDs by ( Author | Department )

If you have more questions or technical problems, please Contact LITS.