Recommendations for the medical treatment of patients, by a physician, can be provided as so called Clinical Guidelines. A lot of previous and current research focuses on how to integrate the computerized form of Guidelines, Computer Interpretable Guidelines into clinical practice. Such an integration could be used to provide real time-decision support. The retrospective analysis of adherence to a guideline is also an area of interest. Such an analysis provides insight into guideline quality in terms of treatment success and the general acceptance of the guideline by healthcare professionals. Medical studies are often conducted with a cohort, which is a group of patients who share a common characteristic. In case of this thesis, the commonality is that all patients of the cohort received the same treatment recommended by the guideline we want to evaluate. The goal of our thesis is to enable a retrospective visual analysis of the treatment data of a whole cohort. Furthermore the information is not only visualized but also interaction techniques are provided to the user, to enable exploratory analysis. We extended techniques for single patients and also implemented new approaches for the visualization of all measurements of certain clinical parameters as well as the executed actions and adherence to a guideline by healthcare professionals. For both distinct types of data we developed one visualization technique that aggregates the information and another one that keeps as much detail as possible for each patient. State of the art visualizations for guideline and statistical compliance information were extended to handle the accumulated data within a cohort. The result is a fully functional prototype based on the Java Programming Language and the Prefuse visualization framework. All developed visualization techniques were implemented in the prototype and are available for use. In an evaluation with a domain expert we assessed the usability of the techniques and visual encodings. We found out, that they are mostly intuitive and understandable. Some of them are harder to grasp but only short introductions were necessary for the expert to properly use them as well. During the conduction of the thesis and the evaluation we were able to identify approaches and ideas for subsequent future research and present them briefly.
clinical guidelines
Advisor
Co-Advisor
Abstract
Year of Publication
2014
Secondary Title
Institute of Visual Computing and Human-Centered Technology
Paper
Number of Pages
100
reposiTUm Handle
20.500.12708/8204
Publisher
TU Wien
Place Published
Vienna
DOI
10.34726/hss.2014.23949
J. Unger, “Visual analytics of clinical data and treatment processes for cohorts”, Institute of Visual Computing and Human-Centered Technology. TU Wien, Vienna, p. 100, 2014.
Master Thesis
AC12119313
P. Votruba, Seyfang, A., Paesold, M., and Miksch, S., “Improving the Execution of Clinical Guidelines and Temporal Data Abstraction in High-Frequency Domains”, in Workshop on AI Techniques in Healthcare: Evidence-based Guidelines and Protocols held in conjunction with European Conference on Artificial Intelligence (ECAI-2006), 2006, pp. 112–116.
K. Kaiser and Miksch, S., “Modeling Treatment Processes Using Information Extraction”, in Computational Intelligence In Healthcare (SCI), vol. 48, Springer, 2007, pp. 189–224.
W. Aigner and Miksch, S., “CareVis: Integrated Visualization of Computerized Protocols and Temporal Patient Data”, Artifical Intelligence in Medicine (AIIM), vol. 37, pp. 203–218, 2006.