Analyzing the Occurrence of Stroke using Machine Learning-A comparative Study on Supervised Learning Models

Abstract

Stroke is blood coagulation or bleed in the brain which could cause permanent damage and affects mobility, intelligence, vision, and communication. Stroke is considered a health-related crisis circumstance and can cause long-term neurological damage, complication, and often death. Most of the stroke is classified as ischemic and haemorrhagic. Stroke has been observed to have abnormal ECG signals. Therefore, if the individuals have their bio-signals monitored in real-time, they can get proper treatment rapidly. Most stroke diagnosis forecast systems are based on image processing tools namely CT and MRI, which are costly and hard to use in clinical practices. Stroke is the consequent driving cause for death around the world and quite possibly the most dangerous infection for persons over the age of 65. It causes ill-effects to the cerebrum like “coronary failure” which causes ill-effects to the heart. When a stroke sickness occurs, it causes enormous clinical care and permanent disability, yet in some cases, it also results in death. Like clockwork, someone dies of a stroke every 4 minutes, yet up to 80% of stroke could be averted if the medical specialist could forecast the incidence of stroke in initial phases. In this project, we have designed an ML model for predicting stroke utilizing the KNN algorithm, SVM algorithm, and NB algorithm. Followed by comparing KNN, SVM, and NB algorithms utilizing error and accuracy. Among KNN, SVM, and NB algorithms, the Support Vector Machine algorithm has the highest accuracy of 96.66%. This project hence helps in predicting the stroke effects and provides a customized warning. Therefore, it urges clinical patients to fortify the inspiration of wellbeing prosperity and brief changes in their medical care practices.