Research on development of online system for stroke prediction

Abstract

This paper presents a system for online stroke prediction. Firstly, a stroke prediction model is built using Random Forest machine learning method. Next, a mobile app user interface is created to collect patients' input data. Based on the trained model and the collected input data, the stroke risk for patients can be predicted. As a result, patients can take preventive measures to avoid severe outcomes from stroke. The training results indicate that the prediction accuracy exceeds 90% on the training dataset and 84% on the testing dataset. The trained machine learning model is then deployed on server as Application Programming Interface. This allows users can access and check their stroke risk from the mobile application.