EARLY PREDICTION OF ISCHAEMIC STROKE RISK IN DR. M. SALAMUN HOSPITAL USING DATA MINNG TECHNIQUES

• Main Author: - NIM 13413080 , Oktavianti
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/23621
• Institution: Institut Teknologi Bandung

Along with the increasing number of hospitals, dr. M. Salamun Hospital is required to continuously develop itself to create some competitive advantages for the hospital. Stroke is the leading cause of death in dr. M. Salamun Hospital. Stroke attacks can be prevented if the high-risk patients can be detected early so that doctors can provide appropriate treatment. The ability to predict stroke risk in patients can be a competitive advantage and help doctors in making decisions. <br /> <br /> The study was conducted by applying three data mining algorithms, namely decision tree C4.5, support vector machine (SVM), and artificial neural network (ANN), using personal data and medical record data from 148 ischemic stroke patients and 124 nonstroke patients at dr. M. Salamun Hospital. The initial input attributes used were obtained from the results of literature studies and the addition from neurologist opinion. The risk of stroke is categorized into two classes, namely high and low risk. Based on the result, C4.5 is chosen as the best algorithm in predicting the risk of stroke disease in patients with 90.52% accuracy. The decision tree is then adjusted to obtain a decision tree that suits its implementation purpose based on doctor's suggestions. The final decision tree has an accuracy of 84.42%. The input attribute is reduced to 8 attributes, namely total cholesterol, HDL cholesterol, LDL cholesterol, triglycerides, monocytes, platelets, gender and cardiomegaly. <br /> <br /> This study proposes the implementation of a simple application that uses the decision tree classification rules to predict stroke risk in hospital patients and refer patients based on their needs. This program can be used directly by the hospitals, especially doctors at cardiology, internal medicine and neurology clinics, and administrators at related clinics and medical check-up sections to assist hospitals in planning clinical management while enhancing its competitive advantage.