Record:   Prev Next
作者 Del Fiol, Guilherme
書名 Context-aware prediction of clinician information needs using "infobuttons."
國際標準書號 9780549493051
book jacket
說明 113 p
附註 Source: Dissertation Abstracts International, Volume: 69-02, Section: B, page: 1091
Thesis (Ph.D.)--The University of Utah, 2008
The knowledge that clinicians face numerous information needs during the course of patient care is common place in the medical literature. Although on-line health information resources are now widely available, clinicians list a series of barriers for the effective use of these resources at the point of care. Infobuttons are decision support tools that integrate electronic medical record systems into information resources as an attempt to lower these barriers. In this dissertation, alternative methods to enhance infobuttons are designed and evaluated in three separate studies
The first study addresses the question whether infobutton links that lead clinicians to specific content topics (topic links) are more effective than those links that point to more general content ( nonspecific links). The study concluded that clinicians with access to topic links were able to find answers to their questions more quickly than clinicians who were offered nonspecific links. In addition, the overall results confirm previous evidence that infobuttons are able to provide clinicians with answers to most of their medication-related questions
The second and third studies assess the feasibility of machine learning methods over infobutton usage data as a method to predict the information needs that a clinician is most likely to have in a given context as well as the information resources that clinicians are likely to visit. Classification models developed in both studies demonstrated a good prediction performance, suggesting that these models may further enhance the effectiveness of infobuttons if implemented in a production environment
In summary, this dissertation adds to the knowledge about infobuttons: First, it confirms the effectiveness of infobuttons in answering questions at the point of care. Second, it suggests a novel approach to the implementation of infobuttons, based on machine learning methods, that may lead to further enhancement. Last, this dissertation also suggests that machine learning methods over usage data may also be of use to enhance other clinical decision support systems, in particular when used from a knowledge management perspective
School code: 0240
DDC
Host Item Dissertation Abstracts International 69-02B
主題 Artificial Intelligence
Computer Science
0800
0984
Alt Author The University of Utah
Record:   Prev Next