Prediction model for stroke severity using data mining techniques from administrative data
Keywords:Stroke, Severity, Adminitrative Data, Model, Data Mining
This article presents the introduction of information for administration. (Administrative Data) with 30 independent variables to select the variables that have the power to identify the severity of illness with stroke. The National Institutes of Health Stroke Scale (NIHSS) is divided into 3 levels which are Mild (NIHSS 0-10 points), Moderate (NIHSS 11-20 points) and Severe (NIHSS 21-42 points). Data mining found that there are 5 variables, including periods of high blood lipid levels Atrial Fibrillation (AF), Glasgow Coma Scale (GCS), Barthel Index (BI) and mRS level, can identify the severity of stroke. The accuracy of 85.2 percent, which the classification helps to provide medical resources, including personnel, tools, treatment methods that are different in each group of patients appropriately.