A Stroke Prediction System Using Machine Learning

Abstract

Stroke prediction is essential and must be treated promptly to avoid irreversible damage or death. With the development of technology in the medical sector, it is now possible to anticipate the onset of a stroke by utilizing Machine Learning Techniques. Considering the major risk factors for Stroke which include: High blood pressure, Diabetes, Heart and blood vessel diseases, High LDL cholesterol levels, Smoking, Brain aneurysms or arteriovenous malformations (AVMs), Sex, and Age, one can be able to prognosticate whether or not a person is at a risk of getting a stroke using a machine learning algorithm. A study sought to establish stroke distribution patterns and characteristics in patients seeking care at Kenyatta National Hospital (KNH) and Moi Teaching and Referral Hospital (MTRH) in Kenya was done recently, A total of 691 patients with confirmed stroke were recruited [KNH 406 (males: 40.9%; females: 59.1%); MTRH 285 (males: 44.6%; females: 55.4%)] and followed over a 12-month period. Overall, ischaemic stroke accounted for 55.6% of the stroke cases, with women being the most affected (57.5%). The most common vascular risk factors were hypertension at 77.3% (males: 75.7%; females: 78.5%), smoking at 16.1% (males: 26.6%; females: 8.3%) and diabetes at 14.9% (males: 15.7%; females: 14.4%).For this particular project the data used in the training and testing of the model is downloaded from kaggle.com which is an online community platform for data scientists and machine learning enthusiasts. Kaggle allows users to collaborate with other users to find and publish datasets. The model will be created using a random forest classifier algorithm which is a Supervised Machine Learning Algorithm that is used widely in Classification and Regression problems. It will then saved using pickle and deployed using Streamlit, in pycharm.