Stroke Risk Factor Prediction using Gradient Boost Method

Stroke is a major global heart concern, often leading to significant disability or death. Early and accurate prediction of stroke risk can significantly improve patient outcomes. To address this issue, our study employs the Gradient Boosting method to enhance stroke prediction using dataset of 750 records. Key factors analyzed include gender, age, hypertension, heart disease, marital status, work type, residence type, average glucose levels, body mass index, and smoking status; the results identified age as the primary risk factor for stroke, followed by hypertension and smoking history. After preprocessing the data, our model achieves an average accuracy of 77,2% across ten runs, demonstrating strong predictive performance. A decision tree visualization highlights the most critical risk factors associated with stroke. This model aims to assist healthcare professionals in identifying high-risk individuals for early intervention. Additionally, we compare the Gradient Boosting model with other algorithms to determine the most effective predictive approach.