Softrams’ Data Lake Strategy Helps the Centers for Medicare & Medicaid Services Improve Healthcare

Softrams provides technology solutions and services to federal clients, including the Centers for Medicare & Medicaid Services (CMS), which is the single largest payer for healthcare in the United States. Seeking to improve quality, equity and outcomes in the healthcare system, CMS provides health coverage to more than 160 million people through Medicare, Medicaid, the Children’s Health Insurance Program (CHIP) and the Health Insurance Marketplace.

Data is vital to the Center for Medicare and Medicaid Innovation (CMS Innovation Center) — a part of CMS that develops alternative payment models (APMs) to increase healthcare quality and reduce costs. For the Innovative Design, Development and Operations Contract (IDDOC), Softrams turned to Snowflake’s Data Cloud to help CMS leverage data and drive outcomes.

Operationalizing a single APM is a data-intensive endeavor that involves securely managing large amounts of protected healthcare information (PHI) and personally identifiable information (PII), collecting provider rosters and validating their billing eligibility, monitoring for fraudulent activity and running complex accountability calculations. Multiply these processes across dozens of models, and that’s a lot of data complexity for the CMS Innovation Center to oversee.

But aging systems and technology posed operational risks as the CMS Innovation Center continued onboarding more models and data. The organization’s previous data architecture included multiple on-premises databases that were difficult to scale and manage. “Keeping both repositories in sync was always a challenge, and the significant downside of that is the potential for PHI or PII leakage,” says Vivek Trehan, Director Portfolio Management at Softrams. 

Softrams proposed the development of a user-facing platform to collect healthcare provider data — the 4Innovation (4i) platform — and modernize the organization’s ACO-OS (Accountable Care Organization Operation System) data environment. Seeking to deliver an effective data lake strategy for its client, Softrams experimented with Apache Spark before pivoting to find a more out-of-the-box solution. That’s when the team began working on a proof of concept (POC) with Snowflake’s Data Cloud.

Snowflake’s elastic performance engine and intelligent infrastructure allowed Softrams’ technical staff to focus on implementing a scalable data lake and accelerating development. “We are using the power of Snowflake to abstract away complexity from our developers, so that they can solely focus on the business of the application,” says Salauddin Shaik, Senior Technical Architect at Softrams. 

With Snowflake, Softrams integrated data with microservices-based applications to serve business users’ needs in near real time while also supporting data lake capabilities for processing large volumes of data — around 3 billion rows of data a month — and adhering to modern application lifecycle management. Softrams has also enabled much-needed DevSecOps capabilities, including a continuous integration/continuous delivery (CI/CD) pipeline. “The biggest benefit that we got by moving to Snowflake was the CI/CD capabilities,” Trehan says. “Now with the CI/CD pipeline, it’s just like clicking a button.”

Snowflake’s security features, such as support for PHI data and role-based access control (RBAC), are essential to Softrams’ client engagement. Data governance features in Snowflake make it easier for Softrams to build a reliable data lineage.

Architecting a scalable data lake with the Data Cloud bolstered Softrams’ ability to complete the multi-phase modernization of ACO-OS and, ultimately, deliver a successful data lake platform to CMS as part of the IDDOC program. With Snowflake, Softrams can rapidly promote code from development to test environments instead of waiting days for on-prem technical support. Operations staff run most jobs in less than 10 minutes, saving approximately 40 hours per complex task. Accelerated data processing and streamlined onboarding of additional data sets allow Softrams to respond faster to stakeholders’ evolving needs. “Snowflake has made our lives significantly easier,” Trehan says.

Combining a reimagined ACO-OS, built on Snowflake, with a user-friendly 4i platform helps model participants manage their data and operations. This two-part system supports more than 7,000 active users across six models, serving over 400,000 providers while ensuring 99.99% application reliability. File processing times have improved by more than 80% each week, despite the need to service approximately 25,000 API requests per day. The complete IDDOC solution has proven to be budget-friendly, too.

Consolidating tech sprawl also aligns with the need to prevent information leakage and preserve data reliability. “Now that Snowflake is our single source of truth, we’ve cut down on that risk and know there is inherent data quality and integrity,” Trehan says.

Onboarding additional data sets to Snowflake is making it easier for Softrams to provide an enhanced self-service experience. “Now we’re enabling end users to build their own dashboards and do business intelligence. We’re serving an unmet need that business owners longed for,” Trehan says. Integrating social determinants of health, general ledger information and other data sets in ACO-OS could unlock new insights for a variety of stakeholders.

Leveraging Snowpark to streamline Softrams’ data science initiatives is also on the roadmap. According to Shaik, “We intend to give data scientists the capability to run some of their analyses with Snowpark.”