influence of lifestyle and clinical variables on cardiovascular disease forecasting through machine learning models

Introduction: cardiovascular diseases (cvds) are the primary cause of death and illness globally, highlighting the need for predictive models to facilitate early detection. this research explores the significance of lifestyle factors and clinical metrics through machine learning models to improve the accuracy of cvd risk evaluations. the study utilizes the framingham heart study dataset, selected for its comprehensive clinical and lifestyle information, which includes more than 4,000 participants.methods and materials:the pre-processing phase included handling missing values, where data imputation methods like median substitution were used for several variables, while others with significant missing values were omitted to prevent data bias. outliers in critical continuous features such as systolic blood pressure, bmi, and glucose were identified and managed with capping/flooring techniques to maintain data integrity and mitigate the impact of extreme values. additionally, categorical variables were transformed into numerical representations, and normalization was applied to unify scales across features, thus enabling effective model training. hyperparameter tuning was applied to improve the performance of logistic regression, random forest, and svm models. techniques like bmi categorization and polynomial feature creation were utilized in feature engineering to capture non-linear relationships among variables, while the voting classifier and cross-validation enhanced the model's predictive accuracy and robustness.results:logistic regression achieved 85.29% accuracy during tuning. random forest as an ensemble model yielded a peak accuracy of 87.6% using soft voting. sensitivity was recorded at 87.33%, indicating strong capability in detecting heart disease, while specificity achieved 88%, which helps minimize false positives.conclusion and discussion:machine learning models show considerable potential in predicting heart disease when incorporating both lifestyle and clinical factors. the research emphasizes the significant predictive influence of factors such as age, bmi, glucose levels, smoking habits, alcohol intake, and previous medical conditions like diabetes. both feature types including clinical and lifestyle features play key role in incidence of heart disease. regular physical activity, a balanced diet, smoking status, and alcohol consumption significantly influence on cardiovascular health. additionally, clinical markers such as blood pressure, cholesterol levels, and bmi are essential for assessing heart disease risk. therefore, incorporating both lifestyle and clinical factors into the development of machine-learning based prediction models is vital for accurately identifying individuals at risk and providing preventive measures. in the future, exploring these lifestyle interactions further and testing on broader datasets could make these predictions even more accurate and personalized.