Sng, Bi Xia.
Date of Issue2011
School of Chemical and Biomedical Engineering
This research project aims to create a causal map that not only shows the potential risk factors of Coronary Heart Disease (CHD), but the inter-relationships between these factors as well. This result could provide insights on the causes of CHD, which might improve accurate diagnosis and management of the disease that could be integrated to provide an improved clinical carepathway. Patients’ data were obtained from the Framingham Heart Study and segregated along the Biological Continuum (BC), which comprised of the human body, physiological system, organ, tissue, cell, protein, and gene level. The analysis of the data was challenged with missing values and class imbalance problems that could potentially introduce biasness into the dataset. K-Nearest Neighbor algorithm was used as the imputation method to fill in the missing values and Self-Organizing Map was used as the solution to the class imbalance problem. The completed and balanced data was analyzed with statistical P value, which had the ability to identify potential risk factors of CHD. The variables were analyzed level by level in the BC, where relationships were investigated with a top-to-bottom approach. The risk factors identified formed a causal map that showed how features in different BC levels were related to each other. The results of the causal map formed consisted of clinically identified risk factors of CHD and new potential risk factors that have not been medically proven to be related to the disease. However, this research has presented a novel idea on how risk factors might not only be individually related to CHD, but sharing an inter-relationship between them as well. The resulting causal map has provided great insights on the causes and inter-relationships of CHD risk factors.
Final Year Project (FYP)
Nanyang Technological University