Overview

The Virginia Department of Behavioral Health and Developmental Services (DBHDS) develops, directs, funds, and monitors the delivery of comprehensive mental health services throughout the Commonwealth of Virginia. In collaboration with state and community partners, DBHDS conducts data reporting and measures the delivery of mental health services in the commonwealth. It also operates twelve behavioral health facilities, including nine residential rehabilitation facilities and three medical and training centers. 

Problem 

DBHDS manages historical data on tens of thousands of patients. Analyzing this data enables the agency to better understand the critical mental health issues facing the people they serve, create optimal treatment plans, and support care staff to implement successful programs. Yet since the data contains sensitive PHI and PII, and is protected under HIPAA, it must be carefully handled. The security team, led by CISO Glendon Schmitz, needed a way for internal teams to analyze and manipulate the data without any chance of PHI or PII being exposed. They also needed a way to share their data with external healthcare providers and pharmaceutical companies to coordinate patient care and conduct valuable research into mental health issues. 

Solution 

After analyzing many data tools, DBHDS chose to implement Howso Synthesizer, the leading synthetic data platform. With Synthesizer, DBHDS generates entirely new data from its original dataset using AI models. These synthetic data sets maintain the statistical properties and characteristics of the original data but eliminate all privacy and security risks, including ransomware attacks, HIPAA violations, and inappropriate data-sharing. DBHDS plans to deploy Synthesizer in several unique ways, including to: 

• Create synthetic production data sets for use in less secure development and test environments 
• Test data quality in a fully secure and compliant way 
• Provide synthetic data sets to partner organizations such as hospitals and universities to conduct joint research on how to improve mental health care 
• Create ML models that accurately predict outcomes of specific treatment protocols 
• Create ML models that predict the best treatment options for specific patients 
• Create ML models that predict per-facility demand to more accurately coordinate bed occupancy, staffing, and state funding 

Results 

DBDHS is in the early stages of deploying Howso Synthesizer but is already seeing measurable impact. All patient data is now fully synthesized and available for use in test and production environments. This means in-house analytics teams can extract valuable information on patient outcomes that could improve care for thousands of people. And by safely sharing patient data with external partners, DBHDS is extending the value of its data to researchers and healthcare providers, who can in turn use it to deliver more effective care for patients everywhere.