AWS re:Invent 2025 - Advancing Behavioral Health Analytics with Integrated Cloud Services (ARC208)

Advancing Behavioral Health Analytics with Integrated Cloud Services

Overview

• This session discussed the journey of the Commonwealth of Virginia in implementing a modern enterprise data warehouse (EDW) to streamline data management, improve data quality, associate individuals across multiple health systems, and optimize outcomes reporting for critical programs serving vulnerable populations.
• The panel included leaders from Virginia's IT agency (VITA), the Department of Behavioral Health and Developmental Services (DBHDS), and the technology partner Deloitte.

Virginia's IT Landscape

• Virginia is a mid-size state serving 8.7 million citizens across 67 different agencies with diverse missions, operating in 1,700 locations.
• Virginia follows a "shared operating model" where the enterprise standardizes infrastructure, while agencies manage their own data and applications.
• Virginia began its cloud journey in 2018, initially using the cloud as an extension of their private data centers. In 2022, they shifted to a more cloud-native approach, leveraging serverless and other advanced cloud services.

Challenges at DBHDS

• DBHDS faced an antiquated, siloed IT architecture with manual processes, data quality issues, and lack of transparency into data and analytics.
• Key goals were to:
  • Integrate data across 12 facilities, 40 community service boards, and thousands of private providers to create a comprehensive view of individuals' care.
  • Enable real-time data ingestion and processing to support clinical decision-making.
  • Improve data quality and governance to enable data-driven decision making.

The Data Warehouse Solution

• The new EDW platform was designed to:
  • Ingest real-time data from various sources (EHRs, financial systems, etc.) using HL7 standards and APIs.
  • Provide data quality monitoring and feedback loops to source systems for data corrections.
  • Enable entity resolution and master data management to create a longitudinal view of individuals.
  • Incorporate advanced analytics capabilities, including the use of synthetic data and AI/ML models.
• Key technical components:
  • Serverless architecture leveraging AWS services like S3, Glue, and SageMaker
  • Snowflake data warehouse
  • Mirth integration engine
  • Extensive use of APIs and event-driven architecture

Collaboration and Teamwork

• The successful implementation was the result of a highly collaborative effort between DBHDS, VITA, Deloitte, and AWS.
• Challenges included navigating unknown territory, establishing data governance and security protocols, and aligning on a shared vision.
• The team adopted an agile, iterative approach with frequent design sessions and retrospectives to address issues.
• Upskilling of staff was a key focus, with over 60 individuals obtaining AWS certifications in the first 6 months.

Future Opportunities

• The new data platform enables DBHDS to be more proactive in identifying trends, patterns, and opportunities for intervention in behavioral health.
• Plans include integrating with other state agencies (e.g., 911 centers) to create a more comprehensive view of individuals' care and social determinants of health.
• Leveraging AI and synthetic data generation to enhance data quality, metadata management, and predictive analytics.
• Establishing data clean rooms and secure data sharing mechanisms to enable cross-agency collaboration and data-driven decision making.

Key Takeaways

• Transforming legacy IT systems in the public sector requires strong leadership, a clear vision, and a collaborative, iterative approach.
• Embracing cloud-native technologies and serverless architectures can enable rapid modernization and unlock new capabilities.
• Investing in data governance, quality, and integration is crucial for enabling data-driven decision making and improving outcomes for citizens.
• Partnerships between government agencies, technology providers, and academic institutions can foster innovation and build a pipeline of skilled talent.