Clinical measurement in assessing obesity and the prediction of biomarkers in association to cardovascular diseases and type-II diabetes

Abstract

There are two main aims in this project. The first aim is to obtain evidence that other obesity measurements are more relevant and provide information concerning clinical conditions than body mass index. This has been achieved by evaluating the effectiveness of bio-impedance analysis using data visualization techniques. The second aim of this project is to investigate the power of several anthropometric measurements in predicting biomarkers. Pearson correlation test was carried out to study the linear relationship as well as the significance of the relationship between two variables and the results were compared to the results of generalized linear model, a linear model of neural network. The results from both methods were found to be consistent, which showed that the data was reliable. Further analysis were carried out using neural network non-linear prediction models such as multilayer perceptron and automatic relevance determination using a leave one-out cross validation method of several input dimensions to predict the non-linear relationship between variables and the performance of the models were studied. The reason for this analysis is to introduce a non-invasive method to monitor obesity and to assess the health risk related to obesity such as identifying the risk of cardiovascular disease and type-II diabetes.