Poster Presentation
Wednesday, 19 July 2006
9:30 AM - 10:00 AM
Wednesday, 19 July 2006
2:30 PM - 3:00 PM
Mathematical Modeling of Symptom Clusters in Cancer Patients
Hee-Ju Kim, MSN, school of nursing, University of pennsylvania, Philadelphia, PA, USA
| Learning Objective #1: The learner will be able to state what kinds of mathematical methods have been used in identifying symptom clusters in various illnesses ?
|
| Learning Objective #2: The learners will be able to state designing issues in symptom clusters research. |
Symptom clusters are groups of symptoms in which symptoms occur together and are interrelated. Defining symptom clusters becomes a major priority of oncology nursing research, because symptom clusters can be an efficient target for symptom assessment and management. The various methods to define symptom clusters shown in the literature (not only nursing but also medicine and psychology/psychiagry) are mathematically examined with special attention to the properties of those methods that could be used in symptom cluster research in cancer patients. The methods discussed include correlation (including regression analysis), graphical model, factor analysis (including principal component analysis), and cluster analysis. Correlation analysis can show the mathematical evidence of a co-occurring tendency for two or more symptoms. Defining symptom clusters using correlation analysis, however, may involve complicated decision-making procedures. Partial correlation may be more useful than simple correlation given that it shows the true relationship between symptoms. Graphical model can show a more concrete image of the possible clusters of symptoms and may also provide a clue for the reasons they are correlated. The greatest concern about graphical model may be the difficulty in interpreting results in a study with a large number of symptoms. Factor analysis can be used to identify groups of symptoms interrelated due to a common underlying cause. Cluster analysis can be used to find clinical subgroups, and to find groups of symptoms that have similar patterns across subjects and represent dimensions of a collection of symptoms. Both factor analysis and cluster analysis have a weakness in that their solutions are subjective. In designing research to identify symptom clusters, several issues need to be considered such as level of measure, measured aspects of symptoms, the homogeneity of the sample, and variable selection (symptoms). Suggestions for future studies are also included.
See more of Poster Presentations I
See more of The 17th International Nursing Research Congress Focusing on Evidence-Based Practice (19-22 July 2006)