Multivariate Adaptive Regression Spline in Ischemic and Hemorrhagic patient (case study)

摘要

In 2010, World Health Organization (WHO) predicted cardiovascular cases cause 73% death rate from total of heart function disorder in human. Stroke cause by disorganized blood circulation in human brain that increase of death in Estonia. Based on WHO data, Stroke suffered people that has age between 0 and 64 years old. The limitation of research is Ischemic and Hemorrhagic patients which are type of Stroke in a hospital, Tallinn, Estonia. The aim in the research is to classify modified risk factors of Ischemic and Hemorrhagic whom are alcohol consumption, smokers, physical activity habit, body mass index (BMI), and diet habit. Thus, it applied Multivariate Regression Spline (MARS). The method is non-parametric method to overcome missing value and to increase accuracy. As result, the classification modified risks factor of Ischemic and Hemorrhagic patient using MARS are alcohol consumption, diet habit, smokers, physical activity, and BMI. The MARS model is f(x) = 0,677 + 0,579 x alcohol consumption - 0,780 × diet habit + 0,383 x smoking habit - 0,409 x physical act - 0,045(bmi - 55, 87) + 0, 126(bmi -63.29) - 0,118(bmi - 66,4) + 0,077 (bmi - 74,47). The probability of Ischemic based on the variables is 0,442 and Hemorrhagic is 0,558 respectively. The accuracy of MARS method is 93,65% and misclassification 6,35% respectively.