Fig 1.
Hieratical clustering on binary responses for 18 questions.
The cluster consists of similarity of 18 questions (see Methods).
Table 1.
Top Ranked Risk Factors for Each Category and Directions.
Fig 2.
Network of risk factors (selected more than once in risk model profile) to each question.
The network consists of the 38 risk factors, which have been selected more than once in risk model profile (Table 1, S3 Table) and 18 questions (S1 Table) with the LASSO algorithm. Risk factors (blue rectangle) and questions (orange circles) are connected by green lines (negative association) or red lines (positive association). The node size reflects the amount of associated risk factors. The edge width and the color grade reflect strength of the odds ratio from the risk model. This network was visualized using Cytoscape 3.2.0.
Fig 3.
Q-Q similarity network based on common risk factors.
The network consists of the 18 questions connected to each other with a criterion of q ≤ 0.01. Questions (circles) are connected by gray lines. The width of the edge reflects the significance of the q value. This network was visualized by Cytoscape 3.2.0.