Comprehensive Feature Selection Methods for Predicting Diabetes and Stroke Risk

The study titled “Comprehensive Feature Selection Methods for Predicting Diabetes and Stroke Risk Among Married Individuals” explores the relationship between various risk factors and the prevalence of diabetes and stroke among married individuals. The primary aim is to identify key risk factors associated with diabetes and stroke among married, divorced, or widowed individuals. To achieve this, a robust feature selection process is employed to identify the most influential factors among an extensive set of variables. After a thorough literature review, the project proceeds with data preparation. This involves the selection of 21 relevant features based on domain knowledge and data analysis. Additionally, the dataset undergoes a comprehensive cleaning process to ensure data quality and accuracy. To select the most relevant features for diabetes and stroke the cleaned dataset was made to undergo different feature selection methods including some new and innovative methods along with some well-known methods. These features were trained using different machine learning algorithms and classifiers and the most relevant features affecting diabetes and stroke were evaluated based on the performance metrics. Prediction models for both diabetes and stroke were then built using the most relevant features. The innovative methodology, integrating various feature selection techniques, advances the field by enhancing the accuracy and robustness of predictive models. The research findings open doors for further investigations and provide a significant contribution to the healthcare domain.