Stroke Prediction with Random Forest Machine Learning Model

Stroke is a medical condition associated with either blockage or rupture of blood vessels which prevents the free flow of blood to the brain cells causing the brain cells to die. The dead brain cells cause malfunctions of the part of the body that it controls leading to stroke that can further result in permanent disability. Both ischemic and hemorrhagic stroke though occurring suddenly, are associated with some stroke risk factors such as age, hypertension, and body mass index among others. These two types of stroke are very dangerous to human health and are a threat to life, ischemic stroke occurs more frequently than haemorrhagic stroke. In an attempt to reduce stroke occurrence, medical doctors use stroke biomarkers to predict stroke occurrence and confirm suspected stroke cases using several diagnostic tests. This technique of stroke prediction and diagnosis is highly time consuming, especially at an early stage when decision making is most important and no individual candidate or multimarker panel has proven to have adequate performance for use in an acute clinical setting hence a need for more efficient stroke prediction technique such as machine learning models. Machine learning is one of the modern areas in artificial intelligence that deals with the ability of a machine to imitate intelligent human behavior. This field is widely applied in healthcare services due to the ever-evolving patient dataset that can be used to train machine learning algorithms for pattern detection that enable medical professionals to recognize new diseases, predict treatment outcomes as well as make medical decisions about the risk of developing disease or medical condition like stroke. this paper aims to predict the stroke vulnerability status of patients using a random forest (RF) machine learning model. The model was built on Python programming language using healthcare_dataset_stroke data obtained from the Kaggle machine learning dataset repository. The dataset was properly cleaned and the clean dataset was used to train the random forest machine learning model for efficient prediction of stroke. the results obtained from the random forest model were evaluated using a confusion matrix and it was found that random forest is a very good choice of algorithm for predicting stroke vulnerability as evidenced in its prediction accuracy of 93%.