For several years CMS has struggled with changing, vague, or counterproductive policies for how topay for software-dominant services. This year, the Senate introduce S.1399, which (if passed) would force CMS to develop coherent software reimbursement policies, at least in the OPPS setting. Last July, CMS inserted a Request for Information into both the PFS and OPPS rules. Numerous parties responded (you can find all rulemaking comments online), but CMS has only responded to the extent of saying, "Thanks for getting back to us, we'll be reading that" to stakeholders.
In this blog I'll focus on the responses from Siemens to both the PFS and OPPS rules. To help you, I'll put a zip file in the cloud four documents, the two RFIs and the two Siemens letters.
https://drive.google.com/file/d/1kKdDr14gczk-ATk8MrT3ovoL5WqirZUa/view?usp=sharing
I've read both letters but I can't write a better summary than Chat GPT 5 does, below. Worth reading.
###
AI CORNER
##
Overview
of Siemens Comments to RFI’s on SaaS
Across both
rulemaking venues, Siemens proposes a conceptual shift away from CMS’s
narrow framing of “Software-as-a-Service (SaaS)” toward a broader and more
inclusive category they call Algorithm-Based Healthcare Services (ABHS).
This new term deliberately captures not only cloud-based software tools
but also AI and machine-learning applications embedded[1]
in imaging equipment, delivered via web or workstation, or provided by third-party
analytic vendors. By urging CMS to adopt ABHS as the central regulatory
category, Siemens is effectively attempting to define the vocabulary—and
therefore the policy architecture—through which Medicare will evaluate and pay
for clinical AI. In regulatory affairs, naming the category often determines
how the category will be regulated, and Siemens is moving early to ensure that
the definitions mirror its own wide-ranging digital portfolio.
In both
letters, Siemens argues that AI-driven analytic services must receive separate,
explicit payment, rather than being folded into packaged payments for
imaging or other underlying procedures. They underline that packaging AI
into base procedure codes would immediately suppress adoption and
undercut the value proposition of these technologies.[2]
To prevent this, Siemens requests that
CMS codify formal regulatory text guaranteeing separate payment for ABHS—even
providing draft language for 42 CFR 419.2 in the OPPS comment letter.
This is a significant move: Siemens is no longer asking CMS to “consider”
separate payment, but is proposing the actual legal language that would lock
separate payment into federal regulation. These comments also push CMS to
explicitly shield ABHS add-on codes from OPPS packaging rules—an attempt
to close every possible loophole through which CMS might inadvertently or
intentionally eliminate separate payment in future rulemaking.
Another
major pillar of Siemens’ strategy is the request that CMS automatically
place all new ABHS CPT codes into New Technology APCs for a minimum of five
years, using manufacturer-supplied cost data rather than unreliable
early Medicare claims. Siemens argues that early claims for new AI services are
frequently distorted by incorrect revenue code assignments, slow
hospital adoption, and lack of clear billing pathways. By providing a five-year
protected runway—parallel to the lifespan of Category III codes—CMS would
ensure stable and predictable reimbursement while allowing enough time for
claims data to mature. This approach mirrors the policy logic behind
transitional pass-through payments and NTAP in the inpatient setting. Siemens
also requests that these ABHS codes be exempt from the Universal
Low-Volume APC policy, which otherwise risks downward payment bias for emerging
technologies.
Siemens
devotes substantial attention to the unique cost structure of AI
technologies, emphasizing that ABHS products involve ongoing subscription or
licensing fees, cloud computing costs, cybersecurity infrastructure,
integration with EHRs, and ongoing staff training. These are not capital costs,
and they do not resemble the equipment depreciation and clinical labor inputs
that CMS traditionally uses to calculate reimbursement. Consequently, Siemens
argues that CMS must modernize its resource-based methodology or risk
chronically undervaluing digital health technologies. They also highlight that
AI often generates additional clinical outputs, which increase cognitive
workload for physicians rather than replacing it. Clinicians must review
expanded information, validate algorithmic findings, integrate AI-derived
insights into care planning, document the use of AI, and communicate results to
patients. Siemens leverages this point to argue for meaningful work RVUs under
the PFS, counteracting any CMS assumption that AI reduces professional effort.
Interwoven
throughout these comments is Siemens’ broader strategic posture as both a policy
thought leader and a legislative partner. Siemens explicitly
supports Senate Bill 1399, the Health Tech Investment Act, which
directs CMS to create clear reimbursement pathways for AI. Even though the bill
is unfunded, Siemens recognizes that it carries substantial agenda-setting
force; it can push CMS to create the structural payment models Siemens
wants. By aligning itself with S.1399 and proposing fully formed regulatory
text, Siemens positions itself not merely as a stakeholder responding to CMS,
but as an architect offering CMS a ready-made blueprint for national AI
reimbursement. Siemens also calls for CMS to convene a national public forum on
ABHS, which would cement Siemens’ role as a central voice in federal
policymaking and create a venue where Siemens can influence the conceptual and
technical underpinnings of AI reimbursement.
Ultimately,
Siemens’ comments reveal a comprehensive, proactive attempt to define how
Medicare will recognize, categorize, and reimburse AI-enabled clinical services.
Their strategy seeks harmonization across PFS and OPPS, codification of
separate payment, early placement into New Technology APCs, explicit valuation
of physician work, and modernization of cost inputs. Through legislative
alignment, regulatory drafting, and conceptual reframing of AI as a
clinical analytic service rather than as a software add-on, Siemens
attempts to shape a durable federal policy framework that supports
innovation while aligning closely with the structure and economics of its own
product portfolio.
[1]
I believe SaaS implies “cloud” and skips “embedded.”
[2] For
published thinking on Ai reimbursement policy, see Warshawsky and others. https://www.discoveriesinhealthpolicy.com/2025/11/center-for-medicare-innovation.html
See my earlier blog about Artera and PathAI comments specific to digital pathology,
https://www.discoveriesinhealthpolicy.com/2025/10/pathai-proposes-new-coding-system-for.html
And general blog about the OPPS and PFS comment cycle,
https://www.discoveriesinhealthpolicy.com/2025/10/very-brief-blog-see-search-comments-on.html
