• Nelakurthi Tejaswi
• Kagita Lasya
• Botla Naga Priya
• Satti Sai Sujith Reddy
• Mohammad Shameem

Stroke remains a significant global health concern, prompting the need for accurate predictive models to assess risk and guide prevention efforts. This study delves into the complex relationship between demographic, clinical, and lifestyle factors to better understand stroke risk, employing advanced machine learning techniques. Analyzing a rich dataset encompassing various features like age, gender, hypertension, heart disease, BMI, and smoking status, the research aims to uncover patterns associated with stroke occurrence. Exploratory data analysis techniques are applied to unveil insights into the characteristics and distribution of stroke patients and non-stroke individuals. Visual aids such as heatmaps, distribution plots, and countplots are utilized to illustrate differences in feature distributions between these groups. Furthermore, the study explores methods for handling missing data, ensuring data integrity and completeness. Feature engineering plays a crucial role in enhancing the predictive power of models, with discrete and categorical features undergoing transformation and encoding processes. Label encoding is employed for categorical features, while preprocessing steps are taken to optimize model inputs for discrete features. Additionally, feature selection techniques like mutual information, chi-squared, and ANOVA analysis are used to identify key predictors of stroke risk. Predictive modeling entails the application of advanced machine learning algorithms, focusing particularly on the eXtreme Gradient Boosting (XGBoost) classifier. Model evaluation metrics such as cross-validation scores, ROC-AUC scores, and confusion matrices are employed to gauge predictive performance and generalization ability. Comparative analyses are also conducted to assess model performance across different feature selection and preprocessing strategies. The study concludes with a comparative analysis of predictive models, providing insights into the efficacy of various methodologies in predicting stroke risk. The findings contribute to a deeper understanding of stroke risk factors and inform the development of tailored preventive measures. Ultimately, this research holds implications for public health initiatives aimed at mitigating the burden of stroke-related morbidity and mortality.

Stroke risk prediction, demographic factors, clinical factors, lifestyle factors, machine learning, exploratory data analysis, feature engineering, feature selection, predictive modeling, eXtreme Gradient Boosting (XGBoost), data preprocessing, missing data handling, model evaluation, public health initiatives.

"Understanding Stroke Risk: Exploring the Relationship Between Demographic, Clinical, and Lifestyle Factors ", International Journal of Emerging Technologies and Innovative Research (www.jetir.org), ISSN:2349-5162, Vol.11, Issue 4, page no.j474-j483, April-2024, Available :http://www.jetir.org/papers/JETIR2404962.pdf

"Understanding Stroke Risk: Exploring the Relationship Between Demographic, Clinical, and Lifestyle Factors ", International Journal of Emerging Technologies and Innovative Research (www.jetir.org | UGC and issn Approved), ISSN:2349-5162, Vol.11, Issue 4, page no. ppj474-j483, April-2024, Available at : http://www.jetir.org/papers/JETIR2404962.pdf