Suchergebnisse

This study used an efficient data mining algorithm, called DCIP (the data cutting and inner product method), to explore association rules between the lifestyles of factory workers in Taiwan and the metabolic syndrome. A total of 1,216 workers in four companies completed a lifestyle questionnaire. Results of the questionnaire survey were integrated into the workers' health examination reports to form an attribute database of the metabolic syndrome. Among the association rules derived by DCIP, 80% of those on the list of the top 15 highest support counts are corroborated by medical literature or by healthcare professionals. These findings prove that data mining is a valid and effective research method, and that larger sample sizes will likely produce more accurate associations connecting the metabolic syndrome to specific lifestyles. The rules already verified can serve as a reference guide for the health management of factory workers. The remaining 20%, while still lacking hard evidence, provide fertile ground for future research.

Currently, interpretation of health examination reports relies primarily on the physician's own experience. If health screening data could be integrated with outpatient medical records to uncover correlations between disease and abnormal test results, the physician could benefit from having additional reference resources for medical examination report interpretation and clinic diagnosis. This study used the medical database of a regional hospital in Taiwan to illustrate how association rules can be found between abnormal health examination results and outpatient illnesses. The rules can help to build up a disease-prevention knowledge database that assists healthcare providers in follow-up treatment and prevention. Furthermore, this study proposes a new algorithm, the data cutting and sorting method, or DCSM, in place of the traditional Apriori algorithm. DCSM significantly improves the mining performance of Apriori by reducing the time to scan health examination and outpatient medical records, both of which are databases of immense sizes.