Prediction of Metabolic Syndrome based on Non-invasive Measurement Features for Chronic Disease Management
ACM International Conference Proceeding Series
Metabolic syndrome is a chronic disease in which metabolism-related abnormal factors are complex. Metabolic syndrome is currently diagnosed by determining the number of abnormal factors based on physical measurements and blood tests. Metabolic syndrome causes complications in patients with diabetes or the cardiovascular system; therefore, prevention and management are particularly important. Metabolic syndrome can be prevented effectively by allowing individuals to manage it themselves; however, blood tests are a major obstacle to the public. Therefore, this study is conducted to devise a method that can easily predict metabolic syndrome without requiring a blood test. A dataset containing data of 69,944 adult Korean men and women is used to develop a predictive model. This dataset contains not only physical measurements and blood test results, but also life logs pertaining to diet, food intake, drinking, and smoking. Using these data, we identify features that contribute significantly to the prediction of metabolic syndrome, except for items associated with blood test. Finally, we propose a predictive model that allows the public to easily manage metabolic syndrome using only non-invasive factors. Furthermore, we investigate methods to improve the predictive performance from the perspective of four subgroups, based on waist circumference and blood pressure.
Ministry of Science, ICT and Future Planning
chronic disease, Metabolic syndrome, non-invasive measurement, predictive model
Applied Data Science; Electrical Engineering
Hyunseok Shin, Simon Shim, Charles Choo, Doosung Hwang, Yunmook Nah, and Sejong Oh. "Prediction of Metabolic Syndrome based on Non-invasive Measurement Features for Chronic Disease Management" ACM International Conference Proceeding Series (2022): 242-248. https://doi.org/10.1145/3543712.3543734