BACKGROUND: The atherogenic lipoprotein phenotype is closely related to determining the risk of Coronary Artery Disease (CAD) and monitoring the treatment process. In particular, it is acknowledged that it is critical not only to have high sdLDL but also to have low lbLDL in determining Pattern B.
OBJECTIVES: The aim of this study was to determine the lipid phenotype using Association Rules Mining (ARM) method based on the concentrations of lipids, lipoproteins, and apoproteins as well as sdLDL.
METHODS: The research type was retrospective; the case-control analytical research method was used. Numerical variables were expressed as mean, standard deviation, median, and min-max values. There is a statistically significant difference between the LDL size categories in terms of TG, LDL, HDL, ApoB, ApoE, sdLDL, and lbLDL distributions. ARM was used to detect lipoprotein phenotype.
RESULTS: There is a statistically significant difference between the LDL size categories in TG, LDL, HDL, ApoB, ApoE, sdLDL, and lbLDL distributions (p_TG<0.001, p_LDL=0.03, p_HDL<0.001, p_ApoB=0.016, p_ApoE=0.004, p_sdLDL< 0.001, and p_lbLDL<0.001). ARM method revealed that the probability of phenotype B was found to be 100% for values in the range of 15.5-109 and 0-31.5 for sdLDL and lbLDL, respectively.
CONCLUSIONS: This study presents a current approach to the detection of lipoprotein phenotypes using ARM, and further supports the close relationship between atherogenic phenotype and sdLDL.