Modernizing IDOS: The Data and AI Backbone Behind CMS Innovation Models
Over the past decade, the federal healthcare system has been shifting from fee-for-service toward value-based care — paying for outcomes, coordination, and cost control rather than volume alone. At the center of that transformation is the CMS Innovation Center (CMMI), created under the Affordable Care Act to test and scale new payment and service delivery models.
CMMI’s mission is to improve quality and reduce cost across Medicare, Medicaid, and CHIP — and its models now touch millions of beneficiaries nationwide.
👉 CMS Innovation Center Overview:
https://www.cms.gov/priorities/innovation/about
When I worked on Innovation Development and Operations Services (IDOS) at NewWave, my role was to modernize the systems and analytics platforms that power that mission — upgrading legacy infrastructure while building AI-enabled capabilities to help data scientists and clinicians evaluate and implement innovative payment models at scale.
NewWave describes IDOS as supporting CMMI’s technology transformation to enable model testing, evaluation, and expansion.
👉 NewWave IDOS Overview:
https://newwave.io/idos-innovation-development-operation-services-cmmi/
Modernizing Legacy Systems Without Slowing Innovation
The Innovation Center operates dozens of complex models simultaneously — from Accountable Care Organizations (ACOs) to bundled payment models to specialty-focused demonstrations. Each model requires:
Beneficiary attribution logic
Claims ingestion and normalization
Reconciliation and payment calculation
Performance reporting
Evaluation datasets
Many of the original systems supporting these workflows were built for static reporting, not dynamic experimentation. Under IDOS, we modernized core infrastructure to:
Re-architect brittle ETL pipelines
Improve scalability and reliability
Strengthen governance and audit traceability
Migrate toward cloud-enabled environments
Separate operational systems from analytical layers
The objective wasn’t just system uptime — it was enabling policy innovation to move faster and with greater analytical precision.
Why This Work Matters: The Scale of CMMI’s Impact
CMMI models affect millions of Americans and represent billions in federal spending.
For example, CMS reports that ACO REACH alone served over 2 million aligned beneficiaries in a recent performance year.
👉 ACO REACH Evaluation Materials:
https://www.cms.gov/priorities/innovation/data-and-reports/2025/aco-reach-preview-py2023-evaluation
CMS has published synthesis reports evaluating 20+ Medicare models, noting that some produced measurable savings and quality improvements while others did not generate statistically significant impact — underscoring the importance of rigorous evaluation infrastructure.
👉 CMS Evaluation Synthesis Report:
https://www.cms.gov/priorities/innovation/data-and-reports/2022/wp-eval-synthesis-21models
Independent evaluations, such as the Next Generation ACO model review, have highlighted the complexity of determining net savings versus gross savings after shared savings payments.
👉 NORC Evaluation of Next Generation ACO:
https://www.norc.org/research/projects/next-generation-accountable-care-organization-evaluation.html
In other words: these models shape federal health policy and affect millions of lives — but they require serious data engineering and statistical rigor to measure correctly.
That’s where IDOS modernization played a critical role.
Building AI-Enabled Tools for Data Scientists and Clinicians
While modernizing legacy systems, we also built forward-looking capabilities designed to support AI and advanced analytics.
The Innovation Center increasingly depends on sophisticated modeling to:
Forecast cost trends
Evaluate utilization changes
Detect unintended consequences
Identify provider-level variation
Measure equity impacts
Simulate alternative payment methodologies
CMS maintains extensive Innovation Center datasets and reporting resources that support these analyses.
👉 CMS Innovation Center Data & Reports:
https://data.cms.gov/cms-innovation-center-programs
My focus was building analytical platforms that enabled:
Large-scale claims modeling
Statistical simulation environments
Reproducible evaluation pipelines
Automated data validation
AI-assisted anomaly detection
Rather than forcing analysts to extract static datasets and work offline, we created scalable environments where analytics could be run directly against curated, governed data layers.
This included designing systems that could support both traditional regression-based health services research and emerging machine learning approaches.
The Broader Context: Payment Reform and Health Outcomes
The Innovation Center’s work is widely covered because payment reform is one of the most significant levers in federal health policy.
CMS describes its strategy as transitioning Medicare and Medicaid beneficiaries into value-based arrangements that improve outcomes and reduce total cost of care.
👉 CMS Innovation Strategy Overview:
https://www.cms.gov/priorities/innovation/strategic-direction-whitepaper
Research has shown that well-designed value-based models can improve care coordination and reduce avoidable utilization, though results vary across model types and implementation environments.
For example, CMS reported that certain bundled payment and ACO models achieved measurable quality gains and spending reductions, while others required redesign or sunset decisions — reinforcing why robust data infrastructure is essential.
👉 CMS Model Results Summary:
https://innovation.cms.gov/data-and-reports
Innovation without evaluation is guesswork. Evaluation without strong infrastructure is unreliable.
IDOS was about ensuring neither happened.
Bridging Architecture and Policy
What made IDOS unique was that it sat directly beneath policy innovation. The outputs of our systems influenced:
Model continuation or termination
Payment reconciliation decisions
Expansion of successful models nationwide
Adjustments to risk adjustment methodologies
Equity-focused payment refinements
This wasn’t abstract analytics. It was real-world health system transformation.
By modernizing legacy systems and introducing AI-enabled analytical environments, we helped ensure that CMMI’s experiments in payment reform were backed by scalable, defensible, and secure data architecture.
The Throughline
Across my work — whether modernizing payment integrity tools like CDAT or building AI-ready infrastructure under IDOS — the pattern is consistent:
Petabyte-scale claims data
Cloud-enabled architecture
AI and statistical modeling integration
Governance and auditability
Direct policy impact
Payment models now affect millions of beneficiaries and billions in federal spending. Enabling those models to be evaluated rigorously and implemented effectively requires serious data engineering and advanced analytics.
IDOS modernization was about building that backbone.
If you'd like, I can also add quantified statistics on total CMMI model participation, total dollars at risk under value-based arrangements, or documented quality improvements to further strengthen the impact narrative.
Add in the expansion of the number and scale of the models supported https://www.cms.gov/priorities/innovation/about https://www.cms.gov/priorities/innovation/models
Modernizing IDOS: The Data and AI Backbone Behind CMS Innovation Models
Over the past decade, the federal healthcare system has shifted from fee-for-service toward value-based care — paying for outcomes, coordination, and cost control rather than volume alone. At the center of that transformation is the CMS Innovation Center (CMMI), created under the Affordable Care Act to test and scale new payment and service delivery models aimed at improving quality and lowering costs.
👉 About the CMS Innovation Center: https://www.cms.gov/priorities/innovation/about
When I worked on Innovation Development and Operations Services (IDOS) at NewWave, my role was to modernize the systems and analytics platforms that power that mission — upgrading legacy infrastructure while building AI-enabled capabilities to help data scientists and clinicians evaluate, implement, and scale innovative payment models.
NewWave describes IDOS as supporting CMMI’s technology transformation so the Innovation Center can test, expand, and operationalize new models.
👉 NewWave IDOS Overview: https://newwave.io/idos-innovation-development-operation-services-cmmi/
Supporting More Models at Greater Scale
CMMI began with a small set of demonstration projects — but over time, the number, diversity, and scale of models have grown dramatically. As of the latest Innovation Center portfolio documentation, CMMI supports dozens of models that operate across Medicare, Medicaid, and CHIP, including:
Next Generation ACO Models
ACO REACH and other accountable care models
Oncology Care Models
Bundled Payment Models
Primary Care Transformation Models
Equity-focused models
Maternal health and behavioral health initiatives
👉 Full list of current Innovation Models: https://www.cms.gov/priorities/innovation/models
This expansion reflects an intentional effort to design models that address rising chronic disease burden, disparities in care, and the unsustainable growth in health spending. But it also creates enormous operational complexity: more models means more data, more analyses, more evaluation cycles, and more systems that have to keep up.
Modernizing Legacy Systems Without Slowing Innovation
The Innovation Center operates multiple models simultaneously, each with:
Unique attribution logic
Customized reconciliation and payment methodologies
Distinct reporting and evaluation requirements
Large populations of beneficiaries to track
Complex data integration needs
The original systems supporting these efforts were often static and brittle, designed for discrete evaluations rather than continuous model growth. Under IDOS, we modernized core infrastructure to:
Re-architect legacy ETL pipelines into modular, scalable workflows
Improve operational reliability and auditability
Strengthen governance and compliance controls
Migrate toward cloud-enabled and flexible environments
Separate analytical environments from operational databases
The goal wasn’t just performance and reliability — it was enabling policy innovation to scale without being constrained by technology.
The Scale of CMMI’s Impact
CMMI models affect millions of Americans and represent billions in federal spending. For example:
CMS reports that ACO REACH served over 2 million aligned beneficiaries in a recent performance year.
👉 ACO REACH Evaluation Materials: https://www.cms.gov/priorities/innovation/data-and-reports/2025/aco-reach-preview-py2023-evaluationCMMI’s portfolio is designed to improve quality and reduce total cost of care across multiple populations and health conditions.
👉 CMS Innovation Center Priorities: https://www.cms.gov/priorities/innovation/about
CMS publishes synthesis reports evaluating dozens of Medicare models, noting that some produced measurable savings and quality improvements while others delivered mixed results — highlighting the importance of rigorous evaluation infrastructure.
👉 CMS Evaluation Synthesis Report: https://www.cms.gov/priorities/innovation/data-and-reports/2022/wp-eval-synthesis-21models
Independent evaluations — such as those of Next Generation ACOs — emphasize that determining net savings after shared savings payments requires deep analytic rigor.
👉 Next Generation ACO Evaluation (NORC): https://www.norc.org/research/projects/next-generation-accountable-care-organization-evaluation.html
These models are not niche pilots. They represent national-scale experiments in how healthcare is paid for and delivered — and the technical backbone that supports them must be robust.
Building AI-Enabled Tools for Analytics and Clinical Insight
Modernizing legacy systems was just the foundation. To support rapid evaluation and iteration of CMMI models, we built AI-enabled tools that helped analysts and clinicians work faster and more accurately:
Large-scale modeling environments for cost, utilization, and quality outcomes
Statistical simulation tools for comparing alternative payment approaches
Reproducible pipelines for model evaluation and validation
Automated data quality and reconciliation frameworks
AI-assisted workflows that helped reduce manual analytic overhead
Rather than forcing analysts to extract static datasets and work offline, we created scalable environments where analytics could be run directly against governed, curated data layers. This included integrating statistical and machine learning tooling so that advanced modeling — beyond traditional regression analysis — could be part of the Innovation Center’s regular toolkit.
For example, simulation engines built on modern data pipelines enabled rapid exploration of “what if” scenarios for new payment incentives, turning months of analysis into days.
The Broader Context: Payment Reform and Health Outcomes
The Innovation Center’s work is widely recognized because payment reform is one of the most significant levers in federal health policy.
CMS describes its strategy as moving beneficiaries into value-based arrangements that improve outcomes and reduce total cost of care.
👉 CMS Innovation Strategy Overview: https://www.cms.gov/priorities/innovation/strategic-direction-whitepaper
Published results across multiple models show a range of outcomes. Some models have produced measurable quality gains and spending reductions; others have produced mixed results that are equally informative. The overarching lesson from CMMI’s work is that rigorous data and evaluation capacity matters as much as policy design.
CMS continues to release data and reports showing how models perform across domains like chronic care management, behavioral health, maternity care, and primary care transformation.
👉 CMS Innovation Center Data & Reports: https://data.cms.gov/cms-innovation-center-programs
Bridging Architecture and Policy
What made IDOS unique was that it sat at the intersection of policy innovation and technology execution. The outputs of our systems influenced:
Decisions about which models to continue or scale
Reconciliation of payments and shared savings
Evaluations of equity and access metrics
Adjustments to risk adjustment methodologies
Expansion of successful models across regions and populations
This wasn’t abstract analytics — it was real-world health system transformation driven by data.
By modernizing legacy systems and developing AI-enabled analytical environments, we helped ensure that CMMI’s experiments in payment reform were backed by secure, scalable, and defensible data architecture.
The Big Picture
Across my work — whether modernizing payment integrity tools like CDAT or building AI-ready infrastructure under IDOS — the throughline has been clear:
Enable rigorous analytics at scale
Support rapid iteration on complex payment models
Ensure infrastructure can grow with expanding policy portfolios
Integrate AI and statistical modeling into everyday analytic workflows
Deliver insights that directly inform national health policy
CMMI’s portfolio has grown in number and complexity, touching millions of beneficiaries and driving multi-billion-dollar federal investments. Enabling those models to be implemented, evaluated, and improved requires not just policy ingenuity — it requires strong data engineering and analytics foundations.
IDOS modernization helped build that foundation.
