Abstract
Purpose: Metabolic syndrome (MS) has become one of the major health concerns of this century, and profiling MS based on life-style conditions could lead to the successful development of preventative programs useful for public health nurses and professionals. In order to achieve this, life-style profiles for characterizing MS were constructed using 16 health behavior items on life-style questionnaires obtained during routine physical examinations.
Methods: Tree-based models were employed to construct health behavior profiles on the 18,849 examinees who were at least 30 years of age at the time of their exam. The risk of MS among different behavioral profiles was evaluated using odd ratios obtained by logistic regression models. Predictive utilities were assessed by estimating prediction error based on cross-validation methods.
Results: “I often think I eat too much” emerged as a common risk factor throughout all profiles. In addition, “frequency of oral care,” “speed of eating,” and “exercise other than walking” appeared as important behavioral risk factors. Upon evaluation using the test data, it was shown that the possibility for application was high in men's profiles.
Conclusion: Exploratory data analysis using the tree-based model on life-style data obtained from routine physical examinations revealed several health behavior profiles. These profiles may provide many practical and potentially useful methods for developing preventive programs for MS.
