Affiliation:
1. Faculty of Science and Technology, Tokyo University of Science, Noda-shi, Japan
Abstract
This study was performed to extract rules and factors for reducing body fat mass and to compare Inductive Logic Programming (ILP) and common classifiers to provide necessary steps for the healthcare system. Many researchers have focused on lifestyle-related diseases; however, few have attempted to extract rules and factors for decreasing body fat mass. The authors obtained lifestyle habits data. This data includes a variety of features (e.g., sleep, exercise, and nutrient intake). These features are easier for patients to understand. ILP and common classifiers are applied to this data. In terms of accuracy, random forest outperformed all other methods, and random forest is suitable for extracting factors among common classifiers. However, in terms of rules, ILP is more suitable than others, because ILP can extract rules covering many positive and negative examples, and it is easy to apply to the healthcare system because these rules cover a range of features.