Abstract TP454: Decision Tree Model for Carotid Atherosclerotic Risks

Abstract

Background: Carotid atherosclerosis is strongly associated with stroke risk. However, we do not know of the severity of the risk in carotid atherosclerosis patients before their ultrasonic evaluations. Purposes: The purpose of this study is to verify the risk factors that influence carotid atherosclerosis and develop a decision tree model. Methods: In this retrospective correlational and methodological study, all 509 adults (329 men and 180 women) who underwent carotid ultrasound at a hospital in Seoul, Korea, for 6 months were enrolled. Baseline data and the results of carotid ultrasound were collected from electronic medical records. Data analyses were performed with SPSS 23 software. Results: A decision tree model for carotid atherosclerotic risks with age, duration of diabetes and hypertension, and LDL values was developed. Various predictive prevalence rates according to 6 subgroups are presented in the model (Figure 1). Misclassification rate, sensitivity, specificity, and area under the curve of this model were 16.6%, 75.6%, 65.2%, and 0.77, respectively. Conclusions: We developed a decision tree model that may be useful for predicting the prevalence rate of carotid atherosclerosis according to individual patient risk factors. Furthermore, nursing management focus on patients in the high-risk group could help prevent stroke.