· By Vironix Health · Industry Insights · 14 min read
CMS Clinical SI Roadmap: Why Outcome Data Is the New Standard for Clinical AI
CMS's Clinical Super Intelligence (SI) Roadmap signals a shift away from activity-based billing and unverified software toward continuous, real-world outcome data. Here is what it means for clinical AI, ACCESS expansion, and between-visit chronic care.

On November 13, 2026, the Centers for Medicare & Medicaid Services (CMS) released A Clinical Super Intelligence Roadmap: Better Answers, Faster Care, and Lower Costs—an eight-page federal framework for the safe deployment of clinical AI across U.S. healthcare, spanning CMS, HHS, and the FDA.
Much of the early coverage has focused on the administration's terminology shift—rebranding artificial intelligence as "Super Intelligence (SI)"—or on potential FDA regulatory reforms. Look closer, however, and a more consequential shift runs through every section of the document: CMS is making continuous, real-world outcome data the price of admission for clinical AI.
Across trust, data infrastructure, FDA oversight, and Medicare reimbursement, the roadmap draws a hard line against two legacy healthcare habits: paying for clinical activity regardless of result, and deploying health technology without proving it measurably improves patient health at lower total cost.
Key takeaways
- Outcomes are the unifying thread: All four pillars of the roadmap (public trust, data foundation, rules of the road, and payment) converge on real-world, post-deployment outcome evidence.
- The end of activity-only incentives: CMS says technology should be deflationary to costs and that it will reward outcomes, not more visits, clicks, or billing, through outcome-aligned payments such as the CMS ACCESS Model.
- A "better than today" evidence bar: CMS says it will not wait for flawless evidence to open pathways, but it will require evidence to keep them open.
- Operational readiness starts now: Practices running outcome-measured RPM, CCM, PCM, and APCM programs today can fund the monitoring infrastructure and build the longitudinal outcome data that future CMS models are designed to reward.
What Is the CMS Clinical Super Intelligence (SI) Roadmap?
The CMS Clinical SI Roadmap is a federal policy framework detailing how HHS agencies plan to scale safe, reliable clinical AI across Medicare, Medicaid, and commercial health tech ecosystems. Rather than issuing a single rule or standalone billing code, CMS sets out how data exchange, FDA market authorization, and alternative payment models will work together.
Two foundational points define the scope:
- Why CMS uses "Super Intelligence (SI)" instead of "AI": Following the White House's July 2025 SI Action Plan and HHS's December 2025 "OneHHS" SI Strategy, the roadmap uses "Super Intelligence (SI)" (explicitly noted as "formerly AI") to cover clinical AI tools—positioning them not as physician replacements, but as "the missing layer between necessary visits."
- Clear guardrails on what CMS will not do: The roadmap does not mandate SI adoption, complements, rather than replaces, clinical judgment where the stakes are highest, does not create a new agency, and will not authorize autonomy that has not cleared a "better than today" evidence threshold for its risk tier.
The Four Pillars—And Why Outcome Data Runs Through All of Them
Most health IT policy frameworks treat interoperability, FDA clearance, and reimbursement as separate tracks. CMS's roadmap ties all four pillars directly to measurable real-world performance:
Roadmap Pillar | Core Objective | How Outcome Data Anchors the Pillar |
|---|---|---|
1. Earn Public Trust | Overcome skepticism through staged, monitored rollout | Requires transparent reporting of real-world clinical performance (surfacing both gains and regressions), alongside informed patient consent, data ownership, and clearer provider workflow and liability guardrails. |
2. Build the Data Foundation | Make longitudinal patient data safely usable at the point of care | Builds on CMS Aligned Networks and the Health Tech Ecosystem (800+ organizations), IAL2 digital identity verification (which helped lift patient digital data access from ~5% to 60% in one year), and the 2026 Medicare App Library—explicitly adding real-time, real-world evidence systems to track post-deployment performance. |
3. Establish Clear Rules of the Road | Replace a 50-state regulatory patchwork with durable national standards | Pairs fast tracks for low-risk wellness and Clinical Decision Support (CDS) tools with a potential competency-based FDA premarket framework (likened to physician credentialing) and possible express federal preemption under the FD&C Act. Crucially, ongoing access relies on post-market surveillance, Predetermined Change Control Plans (PCCPs), FDA's TEMPO pilot, state SI "sandboxes," and clearer liability ownership. |
4. Pay for What Works | Reward measurable health improvement and lower total cost of care | Moves reimbursement away from visits and clicks toward outcome-aligned payments: expanding CMS ACCESS into primary care and additional chronic conditions, integrating SI into Medicare Annual Wellness Visits and preventive care, enabling SI-led accountable care in CMMI models (LEAD, AHEAD) and Original Medicare, fostering SI-led plans in ACA Marketplaces and Medicare Advantage, and testing Medicaid models with states. |
From Billing Activity to Paying for Clinical Outcomes
For health systems, specialty groups, and primary care practices, Pillar 4 ("Pay for What Works") is the clearest signal of where Medicare economics are moving.
CMS delivers a candid critique of prior healthcare digitization: the internet made information abundant and EHRs digitized charts, yet neither consistently reduced the cognitive and administrative burden on patients or clinicians. In CMS's view, clinical software must act as a deflationary force—catching deterioration upstream before delayed care becomes an emergency department visit or avoidable hospitalization.
That philosophy is already taking shape across five payment pathways highlighted in the roadmap:
- Expanding the CMS ACCESS Model: Announced in December 2025 and paired with the FDA's TEMPO pilot in January 2026, the CMS ACCESS Model introduces outcome-aligned payments for technology-supported chronic disease management. In this roadmap, CMS signals that it anticipates extending ACCESS into primary care for additional chronic condition pathways and supporting ACCESS-aligned adoption among other payers. For what ACCESS covers today, see our overview of the ACCESS Model.
- Preventive Care and Annual Wellness Visits: CMS highlights clinical SI as a way to strengthen screenings, longitudinal risk tracking, and post-visit adherence—including delivering components of the Medicare Annual Wellness Visit.
- SI-Led Accountable Care and Original Medicare Reform: CMS anticipates enabling SI-led solutions to participate in CMMI primary care and total-cost-of-care models (alongside capitated models like LEAD and AHEAD) and is exploring broader outcomes-based primary care payment reform in Original Medicare. Notably, CMS specifies that SI-led models must maintain partnerships with human clinicians and facilities for escalation and network adequacy.
- Medicare Advantage and ACA Marketplaces: CMS aims to foster clinical SI-led insurance product options across both Medicare Advantage and the Marketplaces.
- State and Medicaid Partnerships: CMS expects to partner with states to test and validate alternative SI payment structures, including in Medicaid.
The strategic implication is hard to miss: programs built purely to log billable minutes or minimum device transmissions will be poorly positioned as payers shift toward outcome-aligned contracts.
Benchmark AI vs. Real-World Outcome Care
To make the case for clinical SI, CMS cites compelling early studies:
- Diagnosis: OpenAI's
o3ranked the final diagnosis first in 60% of complex cases compared to 15.9% of internal medicine physicians (JAMA Network Open, 2026), while Microsoft'sMAI-DxOdiagnosed 85.5% of challenging cases versus roughly 20% for practicing physicians at lower diagnostic cost (2025). - Treatment recommendations: Google's
AMIEoutperformed primary care physicians across simulated visits (Nature, 2026), and Cedars-Sinai's SI produced optimal treatment recommendations in 77.1% of real urgent-care cases compared to 67.1% by physicians (Annals of Internal Medicine, 2025). - Chronic disease management: A randomized trial (JAMA Network Open, 2023) found that Type 2 diabetes patients using a voice-based conversational AI achieved optimal basal insulin doses faster, with higher adherence and lower emotional distress than standard physician titration.
Those results show what frontier models can do in controlled studies or discrete clinical tasks. Yet any clinician managing heart failure, COPD, chronic kidney disease, diabetes, or complex digestive illness knows there is a wide gap between point-in-time diagnostic benchmarks and moving real-world patient outcomes over months at home.
Delivering outcome-based between-visit care at scale requires solving four operational realities that software alone does not fix:
- Enrollment and sustained retention: Outcome-aligned payment models fail if only a small fraction of eligible patients enroll or if engagement collapses after 30 days. Without continuous longitudinal participation, there is no outcome dataset—and no clinical stabilization.
- Messy, longitudinal physiologic data: Real home monitoring data arrives across comorbidities, irregular schedules, and varying patient literacy—not clean exam vignettes.
- Clinically validated risk stratification: To prevent alert fatigue and meet CMS's "better than today" bar (while guarding against hallucinations, demographic bias, and commercial conflicts of interest), models must separate true clinical decompensation from routine noise.
- Closed-loop human escalation: An accurate risk prediction creates zero value unless a clinical team acts on it the same day and coordinates seamlessly with the patient's physician.
What Outcome-Based Between-Visit Care Looks Like in Practice
CMS acknowledges that trust has to be earned—citing a December 2022 Pew Research Center survey where 60% of U.S. adults were uncomfortable with providers relying on AI for diagnosis or treatment—and says it will come through staged adoption and transparency about what is working in real-world use and what is not.
At Vironix Health, that outcome-first standard is how we have approached remote and chronic care from the start: pairing clinically validated, in-house risk-stratification models with cellular-ready devices, high-touch enrollment, and dedicated clinical monitoring staff so practices can measure both clinical stabilization and program performance over time.
Across active Vironix remote care programs, observational analyses illustrate what structured, between-visit outcome measurement looks like in real clinical populations:
- Sustained blood-pressure stabilization in chronic kidney disease and multi-chronic cohorts: In a 60-day Vironix CKD RPM analysis, the share of systolic readings above 150 mmHg declined from 25–30% to 0–15%. In a longer 150-day Midwest program serving patients with CKD, COPD, and diabetes, readings above 150 mmHg declined from almost 30% to about 5%.
- Statistically significant control-loss reductions in specialty care: In an anonymized gastroenterology virtual care progress report (May 2026), 37 monitored patients showed a statistically significant, persistent ~8-point reduction in systolic blood pressure at 25+ days alongside significant declines in severe (>180), moderate (>160), and mild (>150) systolic control-loss events, while 84 weight-monitored patients maintained stable weight control.
- Engagement built to sustain longitudinal data: Because outcome tracking depends on keeping patients active, Vironix's in-person enrollment model lifts qualified patient enrollment from 15% to 93%, with one gastroenterology practice reporting nearly 90% long-term patient retention in its May 2026 progress report.
- Privacy-preserving longitudinal science: Aligning with CMS's call for secure, real-world evidence systems, Vironix's scientific research with the University of Oxford includes DP-TimeGAN (differentially private time-series EHR generation published at ML4H 2025 and NeurIPS 2025) and ScoEHR (MLHC 2023), alongside our curated Research Library.
(As noted in each linked report, these observational results describe specific participating populations and periods and do not guarantee identical outcomes for every patient or practice.)
Using Today's Medicare Programs as Your Runway to Outcome-Based Care
A common question from practice leaders is whether they should wait for CMS to finalize ACCESS expansions or Original Medicare reforms before investing in between-visit clinical AI and outcome tracking.
Waiting puts practices behind. Existing Medicare care-management programs—Remote Patient Monitoring (RPM), Chronic Care Management (CCM), Principal Care Management (PCM), and Advanced Primary Care Management (APCM)—already provide a funded operational runway today (see our overview of the 2026 Medicare Fee Schedule updates).
When a practice runs these programs with an outcome-driven operating model rather than a checkbox-billing mindset, it accomplishes two things simultaneously:
- Captures sustainable recurring Medicare revenue now without adding administrative burden to clinic staff (practices can model baseline program economics using our Medicare virtual care revenue calculator).
- Builds the longitudinal outcome dataset, retention workflows, and clinical escalation proof that CMS's Clinical SI Roadmap, ACCESS, LEAD, AHEAD, and Medicare Advantage partners are explicitly designed to reward.
Four Questions Every Practice Should Ask Before Adopting Clinical AI
Before signing with any virtual care or clinical AI vendor in 2026–2027, ask:
- What real-world clinical outcomes have you published—not just engagement or billing metrics? Look for longitudinal biomarker stabilization, control-loss reductions, and peer-reviewed validation.
- How do you drive enrollment and long-term patient retention? Outcome-based care breaks down when patients churn after month one.
- Who owns clinical escalation when an algorithm flags deterioration? Confirm how clinician-in-the-loop workflows operate day to day under practice-approved protocols.
- How do you monitor model safety, bias, and drift post-deployment? Ensure your partner is already built for CMS and FDA's continuous real-world evidence expectations.
Frequently Asked Questions About the CMS Clinical SI Roadmap
Is the CMS Clinical SI Roadmap a binding regulation or mandate?
No. The roadmap is a federal policy framework outlining how CMS, HHS, and the FDA intend to align data infrastructure, regulatory oversight, and value-based payment for clinical AI. It creates no new enforcement agency, mandates no immediate adoption, and adds no standalone billing codes on its own.
Why does CMS use "Super Intelligence (SI)" instead of "Artificial Intelligence (AI)"?
The roadmap follows updated federal terminology from the White House's July 2025 SI Action Plan and HHS's December 2025 SI Strategy, explicitly noting "SI (formerly AI)." In clinical operations, it covers healthcare AI broadly—from lower-risk consumer wellness and Clinical Decision Support (CDS) tools to emerging autonomous systems.
How do the CMS ACCESS Model and FDA TEMPO Pilot work together?
ACCESS (announced December 2025) and FDA's TEMPO pilot (launched January 2026) pair reimbursement with regulatory oversight for tech-supported chronic disease care. ACCESS ties Medicare payment to measurable clinical outcomes rather than visit volume, while TEMPO pairs risk-based FDA market access with continuous real-world evidence collection.
How can practices prepare for outcome-based Medicare AI payment models today?
Practices can start now by running outcome-measured RPM, CCM, PCM, or APCM programs for high-risk chronic populations—building longitudinal biomarker tracking, high patient retention, and disciplined clinical escalation workflows while generating compliant Medicare reimbursement today.
Want to build a virtual care program designed around measurable clinical outcomes—not just billing activity? Vironix Health combines peer-reviewed risk-stratification science, cellular-ready monitoring devices, high-touch enrollment, and dedicated clinical staff across RPM, CCM, PCM, and APCM. Contact our team to see how an outcomes-first program fits your practice.
Sources & Primary References
- Centers for Medicare & Medicaid Services (CMS), A Clinical Super Intelligence Roadmap: Better Answers, Faster Care, and Lower Costs (PDF), November 13, 2026.
- CMS Health Technology Ecosystem, Clinical SI Roadmap Overview.
- CMS Innovation Center (CMMI), ACCESS Model.