Please use this identifier to cite or link to this item:
|Title:||INTEGRATION OF STATISTICAL AND MACHINE LEARNING METHODS FOR RISK STRATIFICATION IN COMPLEX DISEASES||Authors:||PAVANDIP SINGH WASAN||Keywords:||Machine learning, complex diseases, risk stratification||Issue Date:||2-Feb-2017||Citation:||PAVANDIP SINGH WASAN (2017-02-02). INTEGRATION OF STATISTICAL AND MACHINE LEARNING METHODS FOR RISK STRATIFICATION IN COMPLEX DISEASES. ScholarBank@NUS Repository.||Abstract:||Obesity, a multi-factorial disease with a complex etiology and a prime component of Metabolic Syndrome is investigated in a cohort of approximately 1000 young Asians, by integrating statistical and machine learning tools, namely Logistic Regression, Recursive Partitioning, Random Forests, and Gradient Boosting Machines. Using SNPs from a panel of 34 candidate genes, we identify associations with early onset obesity, and gene-gene interactions between IL-6 and PPARG showing good predictive accuracy within a subgroup of obese subjects. Building on this, we have identified two co-morbid metabolic factors, hypertension and HDL-Cholesterol, among the obese and investigate aspects of body composition that best distinguish the metabolically healthy from the abnormal. Finally we investigate the effects of physical training on the reversal of these disorders and identify genetic and metabolic factors, including SNPs in PPARG, ADRB2, FTO and PLIN1, that influence the differential capacity for the subjects to lose weight and regain metabolic health.||URI:||http://scholarbank.nus.edu.sg/handle/10635/136479|
|Appears in Collections:||Ph.D Theses (Open)|
Show full item record
Files in This Item:
|WasanPS.pdf||6.21 MB||Adobe PDF|
checked on Nov 8, 2019
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.