UNC Health Modernizes Data Infrastructure with Microsoft Fabric to Scale Clinical AI and Research

UNC Health has transitioned its entire data estate to Microsoft Fabric to address a 40% annual growth in data volume and overcome legacy infrastructure failures. This strategic shift to a unified, SaaS-based analytics platform enables the healthcare system to integrate generative AI and governed data models across its 20 hospitals and research facilities. The modernization is critical for the Data & Analytics sector as it demonstrates how large-scale academic medical centers are leveraging cloud-native tools to bridge the gap between operational maintenance and advanced clinical innovation.

UNC Health, a major academic medical system spanning 52 North Carolina counties, faced significant infrastructure challenges as its data storage requirements grew by 40% annually. The situation reached a breaking point in December 2023 when a hardware failure took its production data warehouse offline, highlighting the limitations of on-premises systems for clinical and research missions. Chief Analytics Officer Rachini Moosavi and Enterprise Data Warehouse Manager Shaun McDonald emphasized the need for a scalable, HIPAA-compliant cloud environment that reduced the burden of manual infrastructure maintenance while providing a foundation for conversational analytics and generative AI.

By standardizing on Microsoft Fabric, UNC Health created a single governed analytics platform that integrates with its existing Microsoft ecosystem and utilizes Copilot for accelerated pipeline development. This transition allowed the organization to move production workloads to a unified environment where community analysts can access centralized data models without competing for fixed server resources. Beyond core data management, the platform supports the SHIRE secure research environment, which currently facilitates 25 active studies, and has streamlined supply chain operations by saving approximately two to three weeks of preparation time for contract and vendor evaluations.

A primary achievement of the modernization is the development of the EASI (Extract, Analyze, Synthesize, and Integrate) framework, an AI solution designed to close care gaps by processing unstructured clinical documents. Dr. Isha Mehta, Medical Director of Population Health Informatics, noted that EASI automates the review of complex reports like diabetic eye exams, improving documentation detection accuracy from 83% to 93% during its pilot phase. According to Program Manager Patrizia Dowdell, the solution has reduced chart review time by roughly 50%, saving an average of three minutes per patient and ensuring that abnormal results are surfaced for timely clinical follow-up.