Data shapes every aspect of modern life, driving impactful decisions across industries, and nowhere is this more apparent than in mental health. Lyra Health, a leader in mental health benefits, supports millions of people every year with high-quality, evidence-based care. Through a data-driven approach, Lyra has achieved remarkable results, including:

• 26% reduction in average annual claims costs over four years
• A 3:1 ROI for every dollar spent — the highest in the industry
• 54% lower health care costs for children and teens, boosting employee retention
• 70% increase in employee productivity

A key driver behind these results is Lyra’s advanced AI-driven matchmaking software that connects patients with therapists and coaches. This focus on data and AI innovation is no small feat, especially when 82% of enterprise C-suite leaders report that AI is both a top priority and a significant business challenge. Lyra is not just thriving in a complex, highly-regulated industry — it’s changing lives through the innovative use of data. 

During a conversation on theCUBE, Chris Suen, Head of Data at Lyra Health, and I discussed how the Fivetran Managed Data Lake is foundational to Lyra Health’s AI and data integration strategy. These were some of my key takeaways. 

The challenge of health care data: AI, privacy and scale

Health care data brings unique complexities, particularly with large volumes of sensitive patient information that require strict privacy and compliance standards. To address this, Lyra relies on a scalable data lake architecture that supports data-driven insights while maintaining compliance and security. This setup allows Lyra to scale storage and compute independently, ensuring that accessibility, performance and innovation remain uncompromised.

Lyra protects patient information within a virtual private cloud (VPC) to reduce exposure and risks while maintaining control and ownership. Within the VPC, Fivetran securely manages data ingestion and movement into a centralized Snowflake instance. This architecture provides Suen and his team with a scalable data lake and the flexibility to explore new technologies like AI while staying compliant with health care regulations such as HIPAA. 

“Fivetran makes it easy to ingest data from various systems. [...] The value add is that our data engineers don't have to manage that anymore. Before, we had to selectively pull what is not PII into the warehouse. Now we don't have to do that. We can do auto schema detection and pull everything down into our private lakehouse.”
— Chris Suen, Head of Data at Lyra Health

How a managed data lake supports Lyra Health’s goals

At the heart of Lyra’s data strategy is the Fivetran Managed Data Lake Service, which combines the ease of use of a data warehouse with the flexibility and scalability of a data lake. With the ability to move data from over 600 data sources into any major data lake destination, Fivetran automatically normalizes and deduplicates your data before standardizing it on Iceberg or Delta Lake table format. Once landed, Fivetran continuously monitors for changes in the source, optimizes tables for performance and keeps its data lake clean and query-ready at all times —  ensuring the data feeding its AI platform is private, accessible and ready for training. 

Fivetran also supports unstructured data, which is ideal for AI use cases. Suen’s team uses large language models (LLMs) to process customer feedback, sentiment and other unstructured data, further enhancing Lyra’s service offerings. Using an LLM with Fivetran “opens the doors for a lot of processing on the lakehouse itself,” Suen explains. “Having your data in cloud storage allows the diversity of data to be in one place. We can selectively pull in what we need for AI.” This approach saves valuable engineering time and allows Lyra’s team to focus on innovating and scaling their AI models.

The future of data-driven healthcare with Fivetran and Lyra Health

The demand for secure, scalable data infrastructure that supports AI is on the rise, as highlighted by a recent study with MIT Technology Review, which found that 80% of enterprise C-suite leaders view AI as their top priority, while 60% cite data movement and integration as major challenges. 

Lyra Health’s use of Fivetran and Snowflake illustrates how data lakes can unlock new AI models and provide critical insights for advancing mental health support. Fivetran’s support of open table formats, such as Iceberg, enhances flexibility by enabling Lyra to maintain full ownership of its data storage while allowing Snowflake’s compute engine to access and process data efficiency, reducing redundancy and cutting costs. 

For organizations like Lyra Health, Iceberg offers a cost-effective way to store large datasets without redundancy. Snowflake’s support for open-source tools, like the Snowflake Open Catalog, signals an industry shift toward openness and interoperability — a move that Fivetran embraces as part of our commitment to data flexibility and scalability.

Together, Lyra Health and Fivetran are setting a powerful example of how data and technology can empower providers to meet the needs of a changing mental health landscape, all while upholding the highest standards of data privacy and innovation.

[CTA_MODULE]