HIPAA-Compliant Patient Risk Stratification Model

Develop a comprehensive patient risk stratification model using Python that anonymizes personal health information while predicting potential health complications. Utilize pandas for data preprocessing, scikit-learn for machine learning, and implement differential privacy techniques to ensure HIPAA compliance. The model should process electronic health records (EHR), extract meaningful features from medical history, lab results, and demographic data, and generate a risk score with 95% confidence interval. Include robust error handling for missing data and create a modular script that can be easily integrated into existing healthcare information systems.