Detecting Stroke in Human Beings using Machine Learning

Abstract

In developing and underdeveloped nations, stroke is a leading cause of mortality and disability. Stroke is a life-threatening condition that develops when there is a lack of blood flow to the brain from the carotid arteries and vertebral arteries. Because the brain suffers damage and can quickly expire without oxygen, stroke frequently results in death and can occasionally affect nearby body parts if the patient is not given prompt medical attention. Spasticity, contractures, paralysis, and death are among the effects. According to the World Health Organization, stroke accounts for over 137,000 fatalities per year in the United States alone and over 451,000 deaths per year in Africa. Today, stroke is a medical illness that affects people in practically every region of the world, including industrialized, developing, and undeveloped nations. In general, 1 in 4 adults over 25 will experience a stroke at some point in their lives. This year, 12.2 million people are predicted to experience their first stroke, and 6.5 million of them will pass away as a result. The number of stroke victims worldwide exceeds 110 million. What if this global endemic could be stopped? The world will be safer and life expectancy will rise if accurate stroke prediction technology is developed. We have proposed our research study to develop a solution to predict strokes in people using machine learning. We have employed four models/classifiers to check the accuracy on each of them with same dataset of people and we have achieved great results. The two models gave 98% and 98.29% successful accuracy results which is very close to state-of-the-art methods (99%).

Keywords: SDG, Stroke, Naïve, Bayes, Random Forest, Decision Tree, Neural network, Data Training

DOI: 10.54941/ahfe1003460