Bridging data gaps in healthcare’s hidden populations

Photo by usertrmk on Freepik

Opinions expressed by Digital Journal contributors are their own.

Let’s talk about the people who keep falling through the cracks in healthcare data. Hospitals and clinics collect enormous volumes of information, yet the patients who need help the most—especially those in rural and tribal communities—often remain the hardest to identify. Their data is fragmented across electronic health records (EHRs), care manager documentation, state systems, and legacy tools. When information cannot be unified, patterns go unnoticed, outcomes go unmeasured, and system-level failures persist.

For Medicaid programs serving these populations, the challenge is not a lack of data, but the ability to transform fragmented and incomplete datasets into decisions that improve care while meeting regulatory requirements.

Shrie Raam Sathyanarayanan is a healthcare data analyst and population health specialist working at the intersection of data science, health IT, and public health. His work focuses on applying analytics, predictive modeling, and AI to improve care delivery for Medicaid and Native American populations, with an emphasis on real-world implementation rather than theoretical analytics.

Q: People often describe healthcare data as “messy.” What does that mean in the context of Medicaid and rural health?

Shrie Raam Sathyanarayanan: The issue isn’t just inaccurate data—it’s dispersed data. In Medicaid, particularly in rural and tribal communities, patient information is spread across multiple EHR systems, state databases, care management documentation, and legacy platforms. This fragmentation makes it difficult to identify high-risk populations or demonstrate whether interventions are effective. When care teams cannot locate or track patients accurately, care gaps widen. These gaps also affect funding decisions and how services are delivered.

Q: You redesigned care management workflows early in your work. What problem were you addressing?

Shrie Raam Sathyanarayanan: Care managers were spending more time managing spreadsheets than managing patients. Excel-based workflows could not scale as patient populations grew, and outcomes were difficult to track. I developed a centralized database that integrated EHR data with care management activities. This allowed teams to manage more than 1,600 patients while reducing administrative workload by approximately 80 percent. The goal was to design systems aligned with real-world clinical workflows rather than idealized processes.

Q: How did automation change how decisions were made?

Shrie Raam Sathyanarayanan: Automated EMR data pipelines significantly accelerated analytics. Instead of waiting weeks for cleaned datasets, teams received near real-time insights. Processing times dropped by more than 50 percent, allowing teams to monitor high-risk patients continuously rather than retrospectively. Analytics became part of daily decision-making instead of a retrospective reporting exercise.

Q: Your predictive models contributed to reductions in emergency department utilization. How did you ensure responsible use?

Shrie Raam Sathyanarayanan: Models must be grounded in clinical and policy realities. The focus was identifying patients at risk of frequent emergency department use so interventions could occur earlier. Transparency and validation were critical, and outcomes were measured against real-world data. Targeted interventions led to a 19–26 percent reduction in emergency visits, with results published in Professional Case Management. The intent was not prediction alone, but prevention.

Q: Why is data preparation such a critical issue in healthcare analytics?

Shrie Raam Sathyanarayanan: Data preparation is foundational. Coding inconsistencies, missing records, and fragmented patient histories can severely distort risk models. I developed a framework involving iterative data validation, cohort refinement, and clinician review. This process identified errors such as misgrouped diagnoses affecting orthopedic readmission risk models. The framework has since been referenced in peer-reviewed research and applied in AI preprocessing workflows.

Q: What is the broader impact of this work beyond individual programs?

Shrie Raam Sathyanarayanan: These tools and frameworks are integrated into Medicaid care management programs and support NCQA accreditation processes. At a statewide level, they inform decisions about program expansion and resource allocation, particularly in underserved communities. Beyond Medicaid, similar approaches are being adopted by hospitals and research teams to integrate EHRs with wearable and alternative data sources. Reproducibility is the core principle—ensuring insights remain reliable across settings and populations.

Q: What defines your niche within healthcare analytics?

Shrie Raam Sathyanarayanan: My work integrates policy, analytics, and frontline care operations. Medicaid policy constraints, NCQA requirements, and tribal governance structures create unique challenges. By aligning AI-driven analytics with regulatory and operational realities, the goal is to close care gaps rather than unintentionally widen them. This combination of policy awareness and technical implementation is essential for meaningful impact.

Closing: Global impact and future vision

As healthcare increasingly relies on predictive analytics and AI, data integrity will determine whether disparities are reduced or amplified. Shrie Raam Sathyanarayanan’s work demonstrates that rigorous data practices, applied to underserved populations, can improve outcomes while meeting regulatory requirements at scale. His approach positions analytics as a bridge—connecting policy, providers, and communities that have historically been underserved. That integration is where systemic change becomes possible.