Prediction of Obesity using Metabolites and Genetic Variants in the Korea Association REsource (KARE) cohort
- Title
- Prediction of Obesity using Metabolites and Genetic Variants in the Korea Association REsource (KARE) cohort
- Author
- 남승현
- Alternative Author(s)
- SeungHyun Nam
- Advisor(s)
- 최성경
- Issue Date
- 2024. 2
- Publisher
- 한양대학교 대학원
- Degree
- Master
- Abstract
- Metabolome Genome wide association studies (mGWAS) analyze combining the genetic and metabolic components of complex diseases, including obesity. Despite the success of many mGWAS, previous studies seem to have difficulty understanding biological mechanisms for many complex diseases. In this study, we performed an mGWAS using Korean Chip (KORV1.1) and metabolite chip (AbsoluteIDQTM p180 kit) from the Korea Association REsource (KARE) cohort of Korean Genome and Epidemiology study (KoGES), and the obesity prediction models were constructed using the results. Logistic and linear regression analyses of 1,776 Korean adults controlling for sex, age, area, income, education, drink and exercise as covariates were performed to identify candidate markers associated with obesity-related metabolites and Single Nucleotide Polymorphisms (SNPs). Moreover, we constructed obesity prediction model using metabolic markers and SNPs through 5-fold nested cross- validation and evaluated prediction performance through several values such as area under curve (AUC), Precision, Recall, F1-score, Error rate (ER), Cohen’s kappa, Matthews Correlation Coefficient (MCC) and Area under the precision-recall curve (AUPRC). The AUC of the prediction model using the epidemiological variables, metabolic markers and SNPs was about 26% higher than the prediction model using only the epidemiological variables and other evaluation measures also increased. In addition, some of the significantly selected metabolic markers and SNPs were found to be associated with insulin resistance, which is involved in the expression and complications of obesity. This result shows that metabolic and genetic markers not only contribute to improving the understanding of the biological mechanisms of obesity but also improve the performance of predictive models.
- URI
- http://hanyang.dcollection.net/common/orgView/200000719640https://repository.hanyang.ac.kr/handle/20.500.11754/188848
- Appears in Collections:
- GRADUATE SCHOOL[S](대학원) > APPLIED ARTIFICIAL INTELLIGENCE(인공지능융합학과) > Theses(Master)
- Files in This Item:
There are no files associated with this item.
- Export
- RIS (EndNote)
- XLS (Excel)
- XML