Integrating clinical diagnostic for CVD using machine learning.

Title: Optimizing Cardiovascular Disease Prediction via Azure-Based Machine Learning: A Clinician-Led Analytical Framework​Background: Cardiovascular Disease (CVD) remains a leading cause of global mortality. Despite the established oral-systemic link, a significant diagnostic gap persists between primary dental care and systemic disease prevention. This study explores the feasibility of a clinician-led, cloud-native machine learning framework to bridge this gap, turning routine clinical biometric data into actionable risk stratification.​Methods: A retrospective cohort analysis was conducted using a clinical dataset comprising 11 biometric features (e.g., blood pressure, BMI, and glucose). The research utilized a robust Microsoft Azure architecture. Data ingestion and orchestration were managed via Azure Data Factory. To ensure model reliability on imbalanced clinical data, SMOTE (Synthetic Minority Over-sampling Technique) was implemented. An XGBoost ensemble algorithm was trained and optimized through Bayesian hyperparameter tuning. Model interpretability was addressed using SHAP (SHapley Additive exPlanations) to ensure clinical transparency.​Results: The proposed architecture achieved an overall accuracy of 91.2% and a Positive Predictive Value (PPV) of 93.1%, indicating high reliability for clinical screening. SHAP analysis successfully identified Age, Cholesterol, and Systolic Blood Pressure as the primary risk drivers.​Conclusion: This research demonstrates that cloud-based AI can effectively bridge the gap between clinical observation and predictive diagnostics.