Session P92.8

Evaluation of Risk Factor Selection in Cardiac Risk Stratification

E Yargholy*, S Parvaneh

Islamic Azad University
Tehran, Iran

Careful monitoring (medical history, extensive physical examination, laboratory test, and …) of different populations such as Framingham study has led to identification of major cardiovascular disease risk factors, as well as valuable information on the effect of this factors, such as blood pressure, age, gender, and ….
In the past few years a number of algorithms for cardiovascular risk stratification have been proposed to the medical community but a big question has been remained unsolved: From among Alternative sets of cardiac risk factors which ones are more significant in cardiac risk stratification?
To answer this question, Statistic data of 165 patients in hospitals of two different big cities were gathered. The data included the following variables: sex, age, LDL, blood pressure, and Myocard-brain Creatinine Phosphokinase (cpk-mb) enzyme. The intensity of infarction was determined according to the amount of the enzyme (in three different levels: no infarction, mild infarction, and severe infarction).
Then a simplified cardiac risk stratification model was developed. Sex, age, LDL, and blood pressure were considered as the input and infarction intensity was considered as the output. To draw the input-output mapping of each group, a hybrid neuro-fuzzy classifier, IRIDIA Method for Neuro-fuzzy Identification and Data Analysis, was used. Having obtained a neuro-fuzzy model, we estimated the significance of input variables in making output using Linkens method.
The study demonstrated that the IRIDIA method is efficient in cardiac risk stratification. Applying IRIDIA method led us to consider a distinction between males and females in cardiac risk stratification. Males and females were studied separately because they differ according to the age at which there is risk of infarction, but in the case of blood pressure and LDL, the differences are not considerable. The results of the study directed us to regard subjects of different cities separately. Since life quality and conditions such as air pollution, life style and … differ in these two cities, so risk ranges of different age, LDL, and blood pressure levels are not the same in different cities and various adaptive strategies can be tuned up in terms of a specific population to stratify the risk of cardiac disease. The significance of input variables (blood pressure, LDL, and age) was approximately the same. So our input set has been appropriate and there is no need to decrease the number of variables or change them. Age, blood pressure and LDL, input variables, were sorted out according their significance respectively. With the benefit of this study, the significance degree of a risk factor can be defined according to a specific population and specific region.

(Abstract Control Number: 306)