Metabolic Syndrome Risk Evaluation Based on VDR Polymorphisms and Neural Networks

摘要

The paper presents an intelligent implementation in medical genetics that supports clinical and laboratory practices by evaluating the risk of having metabolic syndrome (MetS) disorder based on its association with genetic variations or polymorphisms in Vitamin D Receptors (VDR). MetS is approximated in this work with irregularities in biochemical measurements of cholesterol and triglyceride levels in patients. The arbitration of this non-linear relation between VDR polymorphism and metabolic disorders is performed using a backpropagation neural network. The development of this risk evaluation system uses a dataset of biochemical and genetic data of 165 anonymous patients. The experimental results suggest that machine artificial neural networks can be successfully employed to evaluate the risk of metabolic syndrome using genetic and biochemical information.