Following a hospital merger, patient records existed across three separate EMR systems. AI-assisted patient matching and SSIS pipelines created a unified patient data store without losing historical records.
Three hospitals merged, each running a different EMR system with different patient ID schemes, data formats, and coding standards. Clinical staff had no way to see a patient's complete history across facilities, leading to duplicate tests and fragmented care. Regulatory compliance required a unified patient record within 12 months of the merger.
Used Azure Purview to scan and catalog all three EMR databases, automatically documenting table structures, data types, and identifying common patient demographic fields across systems.
Trained an Azure Machine Learning model on name, date of birth, address, and insurance ID combinations to probabilistically match patient records across the three systems with configurable confidence thresholds.
Developed SSIS packages to extract patient demographics, encounter history, diagnoses, medications, and lab results from each EMR on a nightly schedule with full error handling and logging.
Standardized ICD codes, medication names, and date formats across all three sources using Power Query transformations, resolving coding inconsistencies before loading to the unified store.
Loaded matched and standardized patient records into a central SQL Server database with a complete audit trail, preserving source system identifiers for traceability.
Let's discuss how AI-enhanced Data Wrangling & Integration can deliver similar results for your organization.
Start the Conversation