Machine Learning-Optimized Clinical Trial Database

Design a flexible, extensible database schema for managing complex clinical trial data that seamlessly supports both traditional relational storage and machine learning feature extraction. Create a hybrid architecture that allows efficient SQL querying while generating ML-ready feature vectors with minimal preprocessing. Include strategies for handling missing data, versioning experimental protocols, and maintaining full audit trails for regulatory compliance.