Respiratory knowledge discovery utilising expertise.

BACKGROUND: Significant amounts of medical data are being archived, in the hope that they can be analysed and provide insight. A critical problem with analysing such data is the amount of existing knowledge required to produce effective results. AIMS: This study tests a method that seeks to overcome these problems with analysis, by testing it over a large set of archived lung function test results. A knowledge base of lung function interpretation expertise has been compiled and serves as a base for analysis.
METHOD: A user examines the dataset with the assistance of the knowledge discovery tool. Two pertinent respiratory research questions are analysed (the relative correlation between diffusing capacity and FEV(1) or FVC bronchodilator response, and the effects of BMI on various parameters of lung function), and the results compared and contrasted with relevant literature.
RESULTS: The method finds interesting results from the lung function data supporting and questioning other published studies, while also finding correlations that suggest further areas of research.
CONCLUSION: While the analysis does not necessarily reveal groundbreaking information, it shows that the method can successfully discover new knowledge and is useful in a research context.