Purpose The purpose of this study is to introduce the basic concepts and procedures for topic modeling and to explain topic modeling to news articles about dementia-related physical activities. And it is also to discuss the possibility of using topic Modeling in the field of physical education. Methods In this study, the LDA algorithm of topic modeling is explained and the analysis procedure is summarized step by step by text preprocessing, text formatting, and topic number determination. The application cases were selected from 274 news articles about dementia-related physical activities reported in 13 major daily newspapers from 2000 to 2018. Results When the number of topics is 3, the Coherence Score figure is the highest. Topic 1 is about welfare services for dementia patients, Topic 2 is about prevention of dementia, and Topic 3 is about dementia research. The ratio by each subject is Topic 2 (46.0%), Topic 3 (33.2%) and Topic 1 (20.8%) in order of high ratio. Conclusion Topic modeling is an effective methodology to extract potential information excluding subjectivity of researchers. It is expected to be used when searching for information in massive texts in the field of physical education.
PURPOSE This study sought to establish obesity diagnosis criteria by using the Body Volume Index (BVI) by body part extracted through 3D BodyScanner. METHODS The body fat percentage was measured using Dual Energy X-ray Absorptiometer (DEXA) for 225 participants (male = 119, female = 106), and BVI for eight body parts was measured using 3D BodyScanner. Independent t-test and Receiver Operating Characteristic (ROC) analysis were conducted. ROC analysis calculated the Area Under the Curve (AUC), and the optimal cut-point by Youden's J index. Sensitivity, Specificity, Accuracy, Balanced Classification Rate (BCR), and F1-score (harmonic mean of recall and precision) values were calculated to verify the validity of the optimal cut-point. RESULTS A statistically significant difference was observed in BVI by body part according to whether obesity was present for both men and women, and the obese group higher than the normal group. The optimal cut-point for each body part to diagnose obesity was 7.96 for shoulder, 9.79 for chest, 7.15 for upper abdominal, 7.71 for lower abdominal, 14.89 for total abdominal, 9.79 for thigh, 5.70 for calf, and 74.96 for total body volume in men. In case of women, this was 6.04 for shoulder, 9.82 for chest, 4.96 for upper abdominal, 6.23 for lower abdominal, 11.63 for total abdominal, 8.88 for thigh, 4.05 for calf, and 58.15 for total body volume, and the accuracy was 0.6~0.9. CONCLUSIONS BVI is a useful indicator for diagnosing obesity. However, this can be applicable only to Asian adults since there may be differences depending on race or age.
