Wednesday, June 10, 2026

Both talks were delivered at the same venue — the HFMA Annual Conference in National Harbor — within a day of each other, which makes the comparison clean: two senior figures in the same administration laying out adjacent but methodologically distinct theories of where health costs come from.

Oz's pitch is operational and programmatic, organized around five levers CMS is actively pulling. Fraud/waste/abuse is his headline: he claims eliminating Medicare fraud alone would double the trust fund's solvency without tax increases, and he frames the current wave as a COVID-era hangover — emergency money flowed out with weak tracking and no clawback, which recruited a new cohort of opportunistic fraudsters who have since stayed in the game. His geographic examples are the familiar enforcement clusters: DME in South Florida, hospices in Los Angeles, personal-care-services overstatement in New York and California. The other four pillars are drug pricing (Most Favored Nation, a projected $600B over ten years, plus $50/month GLP-1 access for obesity-related conditions starting July 1), tech modernization (retiring the COBOL claims system for a cloud platform, the Medicare App Library, and CMS-backed interoperability/HIE data-sharing with ~800 tech firms signed on), preventive health/nutrition (embedding nutrition training into medical-school curricula, the revised food pyramid), and deregulation (the 10-out/1-in executive order, with an explicit shot at quality-measurement burden — "not everything that matters can be measured").

Mulligan's pitch is a single economic thesis rather than a program list. As HHS chief economist and chief regulatory officer (appointed by RFK Jr. in April), he argues costs stem from incentives, not coverage. His framework — "supply-side health economics" — insists that healthcare, health outcomes, and insurance are three distinct things, and that policy debate over-indexes on insurance coverage while ignoring the delivery-side drivers. His concrete target is Medicaid financing mechanics: provider taxes and state-directed payments. He describes the loop where states tax hospitals/nursing homes/managed-care plans, use the revenue to draw federal matching funds, and recycle the combined money back to providers as supplemental payments. His core claim is that this doesn't stay contained within Medicaid — the distortions spill over into commercial prices, employer and marketplace premiums, wages, and Medicare spending — so curtailing these arrangements would lower costs system-wide. He also frames his own role around rigorous regulatory impact analysis and patient empowerment through data access, transparency, and the freedom to evaluate competing medical claims.

Where they overlap. Both locate a large share of excess cost in misaligned incentives and "money flowing the wrong way" rather than in insufficient coverage — Oz via outright fraud, Mulligan via legal-but-distorting financing gimmicks. Both share the administration's deregulatory instinct (Oz's 10-for-1 and anti-measurement framing; Mulligan's cost-benefit-analysis emphasis). And both nod to patient empowerment and data access as part of the cost solution, though Oz operationalizes it (App Library, interoperability) while Mulligan keeps it at the level of principle.

Where each is unique. Oz owns the entire programmatic/delivery agenda — drug pricing, GLP-1 access, IT modernization, nutrition and preventive medicine, medical education. None of that appears in Mulligan's remarks. Conversely, Mulligan owns the Medicaid-financing critique (provider taxes, state-directed payments, federal-match dynamics) and the formal supply-side framework distinguishing coverage from outcomes — territory Oz never enters. Their genres differ too: Oz is a clinician-administrator selling a portfolio of interventions; Mulligan is an economist selling a diagnostic lens.

Do they contradict? Not directly — the articles record no head-to-head disagreement, and their domains barely intersect (fraud and program operations vs. Medicaid financing economics). But there's a latent tension worth flagging for an expert audience. Oz's nutrition/preventive-medicine pillar is fundamentally an investment-and-coverage move — spend now (GLP-1 access, curricular hours, screening) to bend the chronic-disease curve later — which sits somewhat awkwardly against Mulligan's insistence that the endpoint is "better health and lower costs" through incentives rather than coverage expansion, and that the field over-weights coverage. One could read Oz's $50 GLP-1 benefit as exactly the kind of new coverage commitment Mulligan's framework would want subjected to hard cost-benefit scrutiny. It's a difference of emphasis and method, not an on-the-record contradiction, but the two men are clearly reasoning from different first principles: Oz from clinical opportunity, Mulligan from price-distortion economics.

One caveat: these are conference-address summaries filtered through trade-press reporting, so the figures (the $600B MFN projection, the trust-fund "doubling" claim, the 70% chronic-illness cost share) are as-asserted by the speakers and not independently substantiated in the pieces.

Two Senior CMS Officials on "Saving Costs" - June 2026

MedCityNews, June 9, 2026 (Katie Adams): CMS Dr. Oz on Making Healthcare Affordable
https://medcitynews.com/2026/06/dr-oz-cms-healthcare-affordability/

MedCityNews, June 8, 2026 (Katie Adams): Casey Mulligan on Healthcare Cost Incentives
https://medcitynews.com/2026/06/hhs-affordability-healthcare-costs/

For experts, these two pieces in MedCityNews are less “two cost-control plans” than two different levels of the same policy stack. Oz is giving the CMS administrator’s operational/political affordability agenda: fraud, drugs, IT, prevention, deregulation. Mulligan is giving the HHS economist’s causal theory of cost growth: distorted incentives, especially Medicaid financing mechanics, propagate through the whole system.

Core thesis of each

Oz: CMS can make care more affordable by attacking visible sources of excess cost and friction: fraud, waste, drug prices, antiquated claims infrastructure, chronic disease burden, and regulatory overhead. His framework is program-administrative and public-facing: identify large “fixable” domains and mobilize CMS authority around them.

Mulligan: Healthcare costs are not primarily a coverage problem; they are an incentives problem. He wants policy analysis to separate health, healthcare, and insurance, and to ask who pays, who benefits, and what behaviors are induced. His concrete target is Medicaid provider taxes and state-directed payments, which he portrays as fiscal machinery that inflates spending and spills over into commercial premiums, wages, Medicare, and taxpayers.

Where Oz and Mulligan overlap

They overlap most strongly in diagnosis by incentives rather than entitlement expansion. Neither frames affordability as “more insurance coverage” or “larger subsidies.” Both emphasize that current financing and administrative structures generate avoidable cost.

They also overlap on patient control and information, though with different emphases. Oz’s app-library and interoperability discussion points toward consumer access, portability, and data liquidity. Mulligan’s version is more economic: patients need data, price transparency, genuine choices, and freedom to compare competing medical claims.

They share a deregulatory skepticism toward legacy systems. Oz criticizes quality-measurement burden and welcomes a 10-for-1 deregulatory mandate. Mulligan, as chief regulatory officer, focuses on regulatory impact analysis: measuring costs, benefits, distributional effects, and incentive consequences.

They also both imply that federal healthcare costs are embedded in cross-market systems. Oz says Medicare fraud alone could dramatically affect the trust fund. Mulligan says Medicaid payment games spill into commercial prices, employer premiums, wages, Medicare spending, and taxpayers. Both reject siloed accounting.

What is unique to Oz

Oz’s agenda is broader and more operational. His five buckets are:

Fraud, waste, and abuse — especially DME in South Florida, hospice clustering in Los Angeles, and personal care services employment in New York and California. His Covid-era point is politically important: emergency spending and weak tracking taught new actors that healthcare fraud was accessible.
Drug pricing reform — especially Most Favored Nation pricing, with a claimed $600 billion ten-year savings estimate, plus GLP-1 access for certain Medicare beneficiaries with obesity-related conditions at $50/month.
Technology modernization — replacing Medicare’s COBOL-based billing infrastructure with cloud systems, plus the Medicare App Library and data-sharing networks.
Preventive health and nutrition — especially obesity and chronic disease, nutrition education in medical schools, and revised dietary guidance.
Deregulation and measurement reform — less faith in quality-measure proliferation, more concern about administrative burden.

Oz is therefore speaking as an agency head trying to communicate a portfolio: enforcement, payment policy, IT modernization, prevention, and deregulation.

What is unique to Mulligan

Mulligan’s piece is narrower but more theoretically pointed. His distinctive contribution is the “supply-side health economics” frame: do not confuse insurance expansion with health improvement or cost reduction. He wants to analyze delivery-side incentives.

His most concrete policy target is Medicaid provider taxes plus supplemental/state-directed payments. The mechanism he describes is:

States tax hospitals, nursing homes, or managed-care plans; use the revenue to draw down federal matching dollars; then return funds to providers through supplemental or state-directed payments. Mulligan argues this is not benign fiscal plumbing. It changes provider incentives, increases the cost of delivering care, shifts resources, raises commercial prices and employer premiums, affects marketplace premiums and wages, and even spills into Medicare.

That is the more technical article. It is less about “CMS can modernize X” and more about “federal-state financing arbitrage creates systemwide price inflation.”

Do they disagree?

There is no direct contradiction in the excerpts. The articles are complementary, not adversarial.

The closest tension is one of emphasis:

Oz says major savings can come from fraud enforcement, drug pricing, modernization, prevention, and deregulation. Mulligan says the deeper affordability problem is incentive design, especially financing mechanisms that inflate provider payments and prices.

A second mild tension is that Oz’s agenda includes some expanded access/payment policy, such as cheaper GLP-1 access for Medicare beneficiaries with obesity-related conditions. Mulligan’s lens would ask whether that policy improves health enough to justify costs and what incentives it creates. But the excerpt does not show Mulligan criticizing GLP-1 coverage, and Oz frames it as affordability/access.

A third tension is philosophical: Oz’s prevention-and-nutrition discussion is clinician-population-health oriented; Mulligan’s is economist-incentive oriented. Oz talks about chronic disease burden and medical education. Mulligan talks about regulatory impact, price spillovers, and financing distortions. These are different explanatory languages, not contradictions.

Expert read: what this signals

Together, the articles suggest a Trump-era HHS/CMS affordability message built around five themes:

First, anti-fraud enforcement will be framed as trust-fund solvency policy, not just program integrity.

Second, drug pricing remains central, especially through international reference-pricing rhetoric.

Third, CMS modernization is being cast as cost policy, not merely IT housekeeping.

Fourth, chronic disease prevention is being absorbed into affordability policy, with nutrition and obesity as cost drivers.

Fifth, Medicaid financing reform may become one of the most consequential technical battlegrounds, because Mulligan is explicitly tying provider taxes and state-directed payments to commercial premium inflation, employer costs, wages, Medicare spending, and federal taxpayer exposure.

For a policy expert, Mulligan’s piece may be the more important tea leaf. Oz lists the visible agenda. Mulligan identifies a technically specific target with large distributional stakes: the provider-tax/supplemental-payment ecosystem. If HHS acts on that theory, hospitals, states, Medicaid managed care plans, and employer purchasers will all care deeply.

DR OZ ON SAVING COSTS

The 5 Areas Where Dr. Oz Says CMS Can Make Healthcare More Affordable

https://medcitynews.com/2026/06/dr-oz-cms-healthcare-affordability/

CMS Administrator Dr. Mehmet Oz outlined the agency's strategy for making healthcare more affordable during a Tuesday address at the HFMA Annual Conference, touching on everything from Medicare fraud to drug pricing to nutrition.

Dr. Mehmet Oz

CMS Administrator Dr. Mehmet Oz is optimistic about Washington’s ability to bend the healthcare cost curve, he said during a Tuesday address at the HFMA Annual Conference in National Harbor, Maryland.

“It’s not all rosy, but there’s some opportunities. As a clinician, I’ll tell you, if you have an opportunity to fix a problem, it gives you more hope than if you think the issue is terminal. We’re definitely not terminal,” he declared.

During his talk, Dr. Oz outlined a few key areas that CMS is targeting to make healthcare more affordable. Below are the main pillars of the agency’s affordability agenda.

Fraud, waste and abuse

Eliminating fraud from the Medicare program alone would double the trust fund’s lifespan without raising taxes, Dr. Oz argued.

He cited examples of large-scale fraud cases that the federal government has busted in recent months — an outsized number of durable medical equipment suppliers in South Florida, a disproportionate share of the country’s hospices concentrated in Los Angeles, and inflated personal care services employment in New York and California.

“Much of this started during Covid, because what we taught fraudsters in Covid was, we’re going to give a lot of money away from the federal government, and we don’t really have a way of tracking it, so we can’t really tell if you used it the right way, and we can’t get it back if you didn’t. Because of that, we brought a lot of people into the health ecosystem who never thought of defrauding healthcare before — but now that they know it’s possible, they’re loving it,” Dr. Oz remarked.

Drug pricing reform

Dr. Oz highlighted the Most Favored Nation pricing initiative, which requires pharma companies to charge Americans no more than what developed countries pay abroad. He projected that the policy will result in $600 billion in savings over 10 years.

He also noted that Medicare beneficiaries with obesity-related conditions will be able to access GLP-1 medications for $50 per month starting July 1.

Tech modernization

CMS is replacing Medicare’s COBOL-based billing system with a cloud platform, marking the first upgrade in more than 50 years. Dr. Oz said this change will speed up how Medicare processes and adjudicates claims.

He also noted that CMS launched its Medicare App Library in April, which seeks to create a more consumer-driven, app-based health data infrastructure.

Under the initiative, patient data doesn’t stay siloed within individual apps. Instead, participating companies connect to CMS-backed data-sharing networks — including health information exchanges and interoperability frameworks — that allow data to flow directly into clinicians’ workflows. Nearly 800 health tech companies have signed onto the initiative, Dr. Oz said.

Preventive health and nutrition

The country’s high chronic illness burden — particularly obesity — has a huge impact on rising healthcare costs, Dr. Oz pointed out.

To help fix this, CMS is working to embed nutrition education into medical school curricula. Dr. Oz noted that more than 50 schools have pledged 40 hours of nutrition training.

“The problem is that we don’t teach people going through their training about things like nutrition, so they don’t think they matter. It turns out if you want to deal with the chronic illnesses that drive at least 70% of all healthcare costs, you have got to be able to address basic realities of preventive medicine, including nutrition,” he remarked.

Dr. Oz also cited the recently revised food pyramid as a meaningful step toward correcting decades of flawed dietary guidance that he said has contributed to the obesity crisis.

Deregulation

Under a White House executive order, CMS must eliminate 10 regulations for every new one it introduces.

Dr. Oz said he welcomes this mandate. He argued that much of the current quality measurement apparatus creates administrative burden without actually improving care.

“Not everything that you can measure matters, and not everything that matters can be measured,” he stated.

Casey Mulligan on Saving Costs

https://medcitynews.com/2026/06/hhs-affordability-healthcare-costs/

HHS Affordability Czar Says Healthcare Costs Stem From Incentives, Not Coverage

Healthcare affordability czar Casey Mulligan laid out the economic philosophy guiding the Trump administration's approach to healthcare costs during a recent conference address. He argued that provider taxes and state-directed payments inflate healthcare spending far beyond Medicaid and ultimately raise costs for employers and taxpayers.

By Katie Adams on June 08, 2026 8:24 pm

Healthcare spending is one of — if not the largest — financial burdens facing Americans, federal health regulator Casey Mulligan said during an address on Monday at the HFMA Annual Conference in National Harbor, Maryland.

HHS Secretary Robert F. Kennedy Jr. appointed Mulligan as chief economist and chief regulatory officer of the agency in April. Mulligan said that his work at the agency focuses heavily on how to make healthcare more affordable — and that this requires a focus on healthcare delivery, not just insurance coverage

He uses a framework he refers to as “supply-side health economics” — which insists that healthcare, health outcomes and health insurance are distinct concepts.

“Yes, insurance matters, but it’s not the endpoint. The endpoint is better health and lower costs with more control in the hands of patients and families,” Mulligan declared.

In his eyes, policymakers spend too much time debating insurance coverage and not enough time addressing the underlying drivers of patient outcomes and healthcare costs. Moving forward, he thinks patients should have more information and control.

Mulligan believes patients need access to their own health data, transparent information, genuine healthcare choices and the freedom to evaluate competing medical claims.

He described one of his primary responsibilities as conducting regulatory impact analysis. To do this well, he thinks policymakers should carefully measure costs and benefits — as well as who bears those costs and how incentives are created by policy decisions.

As part of his work at HHS, Mulligan has closely examined provider taxes and state-directed payments.

He argued that many states use a financing mechanism in which they tax hospitals, nursing homes, or managed-care plans, and then this tax revenue is used to draw additional federal Medicaid matching funds. Mulligan said the combined funds are usually returned to providers through supplemental payments or state-directed payments.

“Medicaid financing gimmicks spill over to commercial prices, employer premiums, marketplace premiums, wages and Medicare spending,” he remarked.

Mulligan argued that these financing arrangements end up driving up healthcare costs far beyond Medicaid. And he believes provider taxes typically increase the cost of delivering care, while supplemental Medicaid payments encourage providers to shift resources toward Medicaid patients.

Together, these dynamics put increased pressure on commercial insurance premiums and Medicare spending, he said.

Mulligan contended that limiting these arrangements would help lower healthcare costs for taxpayers.

For him, addressing healthcare affordability requires policymakers to focus less on insurance coverage and more on the incentives that actually shape how healthcare is financed and delivered.

Photo: Malte Mueller, Getty I

Names of 87798-billing labs in 2024 vs names of other labs not billing 87798

Claude Opus 6 9 2026

I dug into both tabs. A quick structural note first: the "Clean_Labs" tab is actually the full universe of 1,561 labs that don't bill 87798 (zero overlap with the 590 on the 87798 tab), so the two lists are cleanly disjoint and directly comparable. Two confounds are worth flagging up front, because they shape the raw word counts:

The 87798 list contains the national giants — Quest Diagnostics (≈19 entity variants) and LabCorp / "Laboratory Corporation of America Holdings" (≈20 variants). These are not fly-by-night labs; they're legitimate megalabs that of course run NAAT. So apparent enrichment of "America," "Quest," and "Holdings" in the 87798 group is an artifact of corporate-subsidiary naming, not a fraud signal. I stripped them out and the real signal held.
The clean set carries a large Puerto Rico contingent (~13%, "Laboratorio Clínico…"), which is entirely absent from the 87798 list and inflates the clean group's distinctiveness. Stripped that too.

With those removed, here are the differences that survive.

Entity type is the single sharpest split. The 87798 labs are overwhelmingly LLCs; the clean labs lean toward Inc.

Suffix	87798 labs	Clean labs
LLC	50.5%	30.0%
Inc / Incorporated	27.8%	39.4%
PC / PA / LLP / PLLC	~1.0%	~5.7%

An LLC is the cheapest, fastest entity to spin up and dissolve, which fits the pop-up profile. Notably, the professional-corporation forms (PC, PA, PLLC) that signal physician-owned pathology/medical practices are almost entirely on the clean side.

"Modern molecular" branding vs. traditional/institutional naming. This is the clearest onomastic pattern. After removing the confounds above:

Words like Diagnostics, Molecular, Genomics, Genetic, Precision, DX, Bio, Sciences, Solutions, Advanced, Labs appear in ~39% of 87798 names vs. ~21% of clean names (roughly 2×).
Words signaling an established institution — Pathology, Pathologists, Physicians, University, Hospital, Medical, Associates, Clinic, Family, Community, plus personal surnames — appear in ~33% of clean names vs. ~19% of 87798 names.

So the 87798 set tilts toward the language of a new molecular-testing venture; the clean set tilts toward the language of a practice, hospital lab, or reference lab embedded in a care setting.

Register: "Labs/Lab/Diagnostics" over "Laboratory/Pathology." The 87798 group favors the informal, brand-forward Labs (10.3% vs 4.5%), Lab (11% vs 6.2%), and Diagnostics/Diagnostic (21% vs 10%). The clean group is where Pathology (7.0% vs 2.9%), Physicians, University, and Pathologists concentrate.

The "Dx" tell and coined brand names. "Dx" as a token is ~3.5× enriched (1.4% vs 0.3%), and qualitatively the 87798 list is where you find the coined/portmanteau startup-style names with "Dx" fused on or abstract single-word brands: Sozo Dx, Citizensdx, Helixbiodx, Northstar Dx, Ipsum Diagnostics, Prescience Diagnostics, Proteus Molecular, Truth Diagnostics, Gezunt Diagnostics, Applied Ingenuity Diagnostics. This invented-brand register is much rarer in the clean set, which skews toward eponymous (surname-based) and place-or-institution-based names.

What does not discriminate. I'd caution against over-reading aspirational adjectives — premier, elite, precision, advanced, prime, apex — as a fraud marker. They appear in both groups at similar rates (e.g., "precision" is actually slightly more common in the clean set, 10 vs 4), so they're noise here despite the intuition. Name length and word count are also essentially identical (≈3.6 vs 3.8 words). And explicit state words (Texas, Florida) are only mildly enriched in the names themselves — the TX/FL concentration you've described lives in the address/PECOS data, not the lab name.

The honest summary: the differentiating signal is entity form (LLC) plus a "molecular-startup" lexical register (Diagnostics/Molecular/Genomics/DX/Labs, often as a coined brand), as against the clean set's institutional/eponymous/professional-corporation register (Pathology/Physicians/University/Medical/surnames, plus PC/PA forms).

It's a meaningful descriptive contrast and a reasonable screening prior, but it's correlational onomastics, not a classifier — plenty of legitimate molecular labs (and the national giants sitting right there in your 87798 list) use the exact same modern vocabulary.

If useful, I can build a scored feature set from this (LLC flag + molecular-token flag + Dx flag + absence-of-institutional-token flag) and run it back over the 590 to rank the names by how strongly they fit the pattern — though you'd want to validate any such score against your billing-fingerprint and geographic-concentration signals rather than lean on the name alone.

Monday, June 8, 2026

CMS 2021: Why Not Price Cat III. Why Hard to Price Software.

86 FR 65039 (small section from 65038)

November 19, 2021 Final PFS rule for CY2022

65039-40

For the PFS, we

typically assign contractor pricing for

Category III codes since they are

temporary codes assigned to emerging

technology and services. We followed

this established process for Category III

codes by assigning and listing them as

contractor pricing in Appendix B in the

CY 2018 PFS final rule

#############

Comment: Several commenters

expressed concern that CMS repeatedly

stated that software and analysis fees are

not direct expenses. Commenters

disagreed and stated that software that

is directly attributed to a specific

physician service is a direct expense,

and furthermore that there are multiple

examples of the implementation of such

costs. Several comments provided a list

of current CPT codes that they stated

included software as a direct PE input,

such as CPT code 95905 (Motor and/or

sensory nerve conduction, using

preconfigured electrode array(s),

amplitude and latency/velocity study,

,,,, Several commenters raised the

issue of software as a medical device

(SaMD) and stated that it should be

considered a direct PE expense similar

to other medical equipment.

Commenters stated that even though

SaMD does not require physical space

in an office or administrative staff hours

to maintain it, SaMD does require

ongoing upgrades, improvements, and

security mitigation, as well as the same

regulatory oversight by the Food and

Drug Administration (FDA) as hardware

medical devices. Commenters stated

that the legal, regulatory, and financial

burdens incumbent of a SaMD

manufacturer are no different than those

of hardware medical device

manufacturers.

Response: We appreciate the detailed

feedback from the commenters

regarding the issues surrounding

software and analysis fees. We agree

with the commenters that there have

been occasions in the past where we

have finalized the inclusion of software

as a direct PE expense if it met our

criteria as typical and medically

necessary for the service in question and

could be individually allocable to a

particular patient for a particular

service. For example, we included the

sheer wave elastography software

(ED060) as a direct PE input for CPT

codes 76981–76983 in CY 2019. In this

case, the sheer wave elastography

software was an additional resource cost

added to the general ultrasound room

(EL015) equipment without which the

service cannot be performed. We have

been more hesitant to classify software,

licensing, and analysis fees that are not

associated with physical equipment

used in the performance of a service as

they pose more significant challenges

for our traditional PE methodology.

Therefore, we wish to clarify that

although we have typically considered

software costs to be indirect PE under

our methodology, as these costs were

not individually allocable to a particular

patient for a particular service, there

have been exceptions to this general

principle where software costs have

been included directly in the service

under review.

As we stated in the proposed rule, we

believe that costs associated with

software, licensing, and analysis fees are

not well accounted for in the PE

methodology. Unlike a piece of

equipment, such as the retinal camera,

an analysis fee for software does not

require physical space in an office or

administrative staff hours to maintain it.

These types of costs were much less

prevalent when the Physician Practice

Information Survey (PPIS) was last

administered in 2007 and 2008 and of

course did not exist at all in the case of

AI-based services. We remain concerned

that if we were to consider software

analysis fees and software as a medical

device expenses to be direct costs in all

cases, we may inadvertently allocate too

many indirect costs for supplies that

may not require additional indirect

expenses. The data underlying the PPIS

assumes that direct expenses will

require costs associated with physical

space and physical maintenance that

may not appropriate for these new types

of software. However, we do recognize

that practitioners are incurring resource

costs for purchase of the software and

its ongoing use, which is why we

proposed the crosswalk to CPT code

92325 to capture these resource costs for

CPT code 92229. We believe that the use

of this crosswalk and other similar

crosswalks are the best way to value

services that make use of software,

licensing, and analysis fees at the

moment while we explore ongoing

potential updates to the PE

methodology.

Sunday, June 7, 2026

AI Assesses CMS Pricing History of 92229

[Direct AI research - not verified]

Please use agentic research and determine how CMS set the price for code 92229, retinal scan and auto diagnosis. In the AMA RVU RUC database, they allot about 15 minutes of technician time at 50 cents per minute or about $8, and they alot a trivial capital equipment charge, like 50 cents (a $14000 device with 5 year life used for 15 minutes yields only amortized pocket change). Did CMS see this and nonetheless assign 92229 an artificial value like around 1 RVU ($30-40). How did CMS justify that.

###

Bottom line: Yes. CMS saw the ordinary RUC-style direct practice expense inputs for 92229 — staff time, retinal camera time, etc. — but concluded that the usual PE machinery did not adequately capture the core cost of the service: the autonomous software/AI analysis. CMS first made 92229 contractor-priced for CY 2021, then in CY 2022 created a national price by crosswalking 92229 to CPT 92325, a PE-only eye-code, yielding about 1.36 RVUs and roughly $45.69 national payment in 2022.

Here is the sequence.

In the CY 2021 PFS final rule, CMS described 92229 as retinal imaging with point-of-care automated analysis using AI, with no ophthalmologist interpretation. Therefore, CMS assigned it 0.00 work RVUs and treated it as a practice-expense-only service. CMS explicitly said the RUC had recommended a $25 “per click” analysis fee, but CMS rejected putting that fee into direct PE, saying it looked more like an indirect PE/service fee rather than an item allocated as a direct cost to one patient encounter.

CMS then got pushback. Commenters argued that the AI analysis fee was directly attributable to each patient and that, without the software analysis, there was no 92229 service. CMS accepted the conceptual problem but not the RUC input method: it said AI/software costs should be accounted for, but that AI applications were “not well accounted for” in the current PFS PE methodology. Rather than create a direct PE software fee, CMS finalized contractor pricing for 92229 for CY 2021.

In the CY 2022 proposed rule, CMS made the broader policy admission you are circling around: the PPIS indirect PE data were from 2007–2008, before software, licensing, algorithmic analysis fees, and similar cost structures were typical. CMS said its existing PE method was built around clinical labor, supplies, equipment, and indirect office expenses, and was a poor fit for services whose real cost was a software algorithm or analysis fee with minimal hardware cost.

So for CY 2022, CMS proposed not to rely on the literal camera-amortization-plus-technician-time calculation. Instead, it proposed a crosswalk to CPT 92325, “Modification of contact lens … with medical supervision of adaptation,” because 92325 was a PE-only eye code that CMS believed reflected comparable overall resource costs in the physician office setting. CMS openly acknowledged the services were not the same, but said the total resource costs were similar, and that the crosswalk would recognize the purchase and ongoing use of software that “would not typically be considered direct PE” under the current methodology.

In the CY 2022 final rule, CMS finalized that exact approach. Commenters supported the crosswalk, noting that 92325 was not clinically similar but had a total direct practice cost similar to the RUC-recommended total direct PE cost for 92229. CMS rejected alternative crosswalks such as 95249 and 92977 because those would include additional resource costs and produce a higher valuation. CMS then finalized values for 92229 based on a direct crosswalk to 92325.

The resulting payment was not a “work RVU” payment and not a payment for 15 minutes of technician time alone. It was a PE-only proxy payment for a service whose dominant cost was software/AI analysis. CMS’s own general PFS explanation is that payment is derived from RVUs for work, PE, and malpractice, multiplied by the conversion factor; for CY 2022, the conversion factor was about $33.59. (CMS) Industry reporting based on CMS’s published RVUs put 92229 at about $45.69 nationally in CY 2022, consistent with roughly 1.36 total RVUs. (Eyenuk)

So the clean interpretation is:

CMS did not simply “see $8 of technician time and 50 cents of equipment and then mysteriously pay $40.” Rather, CMS saw that the ordinary PE inputs badly understated an AI service because the real cost was a per-use software/analysis cost. But CMS was unwilling to accept the RUC’s per-click software fee as a direct PE input. It therefore used the old-fashioned Medicare workaround: contractor price first, then crosswalk to a roughly comparable PE-only code.

For your policy point, 92229 is a useful early example of CMS saying, in effect: our PE methodology has no good slot for software-as-the-service, so we will use a pragmatic crosswalk until the PE methodology catches up.

###
Version with citations

Sure — here is the same text in ordinary prose, with the full plain-text source link at the end of each relevant paragraph.

Bottom line: Yes. CMS saw the ordinary RUC-style direct practice expense inputs for CPT code 92229 — staff time, retinal camera time, and similar items — but concluded that the usual practice-expense machinery did not adequately capture the core cost of the service: the autonomous software/AI analysis. CMS first made 92229 contractor-priced for CY 2021, then in CY 2022 created a national price by crosswalking 92229 to CPT code 92325, a practice-expense-only eye-code. The resulting national Medicare payment was about $45–46 in 2022, consistent with roughly 1.36 total RVUs. Source: CMS CY 2021 PFS final rule, https://public-inspection.federalregister.gov/2020-26815.pdf ; CMS CY 2022 PFS final rule, https://www.federalregister.gov/documents/2021/11/19/2021-23972/medicare-program-cy-2022-payment-policies-under-the-physician-fee-schedule-and-other-changes-to-part

Code 92229 describes retinal imaging for detection or monitoring of disease, with point-of-care automated analysis and report, unilateral or bilateral. In the CY 2021 Medicare Physician Fee Schedule final rule, CMS treated 92229 as a practice-expense-only service, with 0.00 physician work RVUs, because the code does not involve a physician or other qualified health care professional interpretation in the usual sense. CMS noted that the RUC had recommended a $25 “per click” analysis fee, but CMS rejected the idea that this software or analysis fee should simply be put into direct practice expense as a patient-specific input. CMS said the “per click” fee looked more like an indirect practice expense or service fee rather than a conventional direct input such as clinical labor, disposable supplies, or equipment time. Source: CMS CY 2021 PFS final rule, https://public-inspection.federalregister.gov/2020-26815.pdf

CMS then received comments arguing that the AI analysis fee was directly attributable to each patient and that, without the autonomous software analysis, there was no 92229 service. CMS accepted the broader conceptual point that AI and software costs had to be recognized somehow, but it did not accept the RUC’s proposed direct-input method. Instead, CMS finalized 92229 as contractor-priced for CY 2021. This allowed Medicare Administrative Contractors to establish payment locally while CMS considered how to handle AI/software costs more generally under the Physician Fee Schedule. Source: CMS CY 2021 PFS final rule, https://public-inspection.federalregister.gov/2020-26815.pdf

In the CY 2022 proposed rule, CMS was more explicit about the methodological problem. CMS acknowledged that the practice expense system was built around physician work, clinical labor, supplies, equipment, and indirect practice costs, and that its underlying indirect practice expense data were old — based on the Physician Practice Information Survey, collected in 2007–2008. That vintage mattered because the current structure did not fit well with newer services whose dominant cost is software, algorithmic analysis, licensing, or a per-use analytic fee rather than a physician’s time or a large piece of capital equipment. Source: CMS CY 2022 PFS proposed rule, https://public-inspection.federalregister.gov/2021-14973.pdf

For CY 2022, CMS proposed a workaround. Rather than price 92229 by simply adding 15 minutes of technician time plus a few cents or dollars of amortized retinal camera time, CMS proposed to crosswalk the service to CPT code 92325, “Modification of contact lens, with medical supervision of adaptation.” The clinical service is obviously different, but CMS regarded 92325 as a practice-expense-only ophthalmology code with a comparable overall resource profile. In effect, CMS used 92325 as a proxy value that could recognize the resource cost of the software analysis without formally accepting a direct “per click” AI fee as a practice expense input. Source: CMS CY 2022 PFS proposed rule, https://public-inspection.federalregister.gov/2021-14973.pdf

In the CY 2022 final rule, CMS finalized that approach. Commenters supported the 92325 crosswalk, even while acknowledging that 92325 was not clinically similar to 92229. CMS rejected alternative crosswalks that would have produced higher payment, because those codes included additional resource costs that CMS did not think were appropriate for 92229. CMS therefore finalized national values for 92229 based on a direct crosswalk to 92325. Source: CMS CY 2022 PFS final rule, https://www.federalregister.gov/documents/2021/11/19/2021-23972/medicare-program-cy-2022-payment-policies-under-the-physician-fee-schedule-and-other-changes-to-part

The resulting payment was not a physician-work payment and not a literal reimbursement for 15 minutes of technician time alone. It was a practice-expense-only proxy payment for a service whose dominant cost was autonomous software analysis. CMS’s CY 2022 conversion factor was $33.59, and industry summaries based on CMS’s published RVUs estimated the national payment for CPT 92229 at about $45.69. Another ophthalmology coding summary reported a national payment of about $45.36. The small difference reflects rounding and locality/payment-file conventions, but both estimates are in the same range. Sources: CMS CY 2022 PFS fact sheet, https://www.cms.gov/newsroom/fact-sheets/calendar-year-cy-2022-medicare-physician-fee-schedule-final-rule ; Eyenuk, https://www.eyenuk.com/us-en/articles/eyeart/2022-cms-pfs-final-rule ; Review of Ophthalmology, https://www.reviewofophthalmology.com/article/medicare-whats-new-for-2022

So the clean interpretation is this: CMS did not simply see $8 of technician time and 50 cents of equipment time and then mysteriously assign a $40-plus payment. Rather, CMS recognized that the ordinary direct practice expense inputs badly understated an autonomous AI service because the economically meaningful component was the software analysis. At the same time, CMS was unwilling to accept the RUC’s proposed $25 per-click software fee as a direct practice expense input. The compromise was first contractor pricing, then a national crosswalk to an existing practice-expense-only eye code. Source: CMS CY 2021 PFS final rule, https://public-inspection.federalregister.gov/2020-26815.pdf ; CMS CY 2022 PFS final rule, https://www.federalregister.gov/documents/2021/11/19/2021-23972/medicare-program-cy-2022-payment-policies-under-the-physician-fee-schedule-and-other-changes-to-part

For policy purposes, 92229 is an important early example of CMS saying, in effect: the Physician Fee Schedule practice expense methodology has no clean slot for software-as-the-service. Until that methodology catches up, CMS may use pragmatic crosswalks rather than literal technician-time-plus-equipment amortization.

Saturday, May 23, 2026

Chat GPT compares OPUS and CHAT regarding SPAC S4 (May 2026)

This presents two AI assessments of the Freenome SPAC S4 at SEC.

The first is Claude Opus 4.7 and the second is Chat GPT 5.5.

COMPARE

Yes. The uploaded comparison document gives enough to make a useful meta-review. Both reviews reach the same broad conclusion, but Opus is more granular, more investor-diligence oriented, and more aggressive about finding commercial complications, while my ChatGPT review is cleaner, more policy-framed, and more client-memo ready, but missed some important deal-specific and competitive details that Opus caught.

Executive comparison

Both reviews agree on the core thesis: Freenome is a CRC-first cancer-screening platform company, not simply an MCED company. Both emphasize that CRC is the reimbursable beachhead; Exact Sciences is the key U.S. commercialization partner; Roche is the broader strategic/platform partner; and the longer-term aspiration is personalized multi-cancer detection built from multiomics, AI/ML classifiers, longitudinal data, and health-system workflows.

Where they differ is in resolution. Opus reads like a diligence note written for a biotech investor or strategic BD team. It drills into transaction proceeds, valuation, PIPE composition, Abbott/Exact change-of-control risk, ADLT pricing analogs, Shield competitive timing, and v2 statistical caveats. My ChatGPT review reads more like a balanced policy/strategy memo for your genomics clients: it is stronger on reimbursement architecture, USPSTF/CMS/MCED framing, and avoiding hype, but it is less sharp on deal mechanics and competitive timing.

Where Opus is stronger

Opus is clearly stronger on commercial execution risk. The most important point it adds is the Abbott acquisition of Exact Sciences. My review treated Exact as a stable channel partner. Opus correctly flags that Freenome’s CRC commercial future may now depend not merely on Exact’s original strategy, but on Abbott’s post-acquisition prioritization. That is a very practical BD question: does Abbott lean into SimpleScreen as a blood complement to Cologuard, or does it slow-walk the Freenome asset while sorting out its diagnostics portfolio?

Opus is also stronger on competitive positioning versus Guardant Shield. My review noted Guardant generally, but Opus makes the sharper business point: Freenome may be 18–24 months behind Shield, which already has FDA approval, Medicare coverage, ADLT pricing, NCCN inclusion, and live commercial experience. That matters more than the abstract platform story. In CRC screening, being second can still work, but only if the channel or product profile changes the adoption curve.

Opus also does better on v2 skepticism. My review said v2 is the performance-improvement story and not yet the regulatory product. Opus goes further: it notes that the v2 data were from a roughly 1,300-sample case-control bridge, not the original prospective registrational setting, and that the paired v1-to-v2 improvement may be more modest than the headline numbers suggest. That is exactly the kind of “neither hype nor attack” nuance you would want in a client-facing memo.

Opus is also better on transaction mechanics: valuation around $1.05B, step-down from prior private valuations, insider/strategic PIPE concentration, minimum proceeds condition being waivable, cash runway into 2028, and likely need for additional capital. My review mentioned cash burn but did not fully connect it to the SPAC mechanics.

Where ChatGPT is stronger

My review is stronger as a policy-facing synthesis. It keeps the frame clear: traditional Medicare generally does not cover preventive screening absent either a statutory benefit or a USPSTF/NCD pathway; CRC is different because it already has a workable screening and coverage architecture; MCED legislation helps future MCED but does not magically solve CRC reimbursement or payer adoption.

The ChatGPT version also gives a cleaner client-ready formulation: Freenome’s moat is not any one biomarker or patent, but the combined package of multiomics assay design, prospective datasets, AI/ML classifiers, partner distribution, payer-aware sequencing, and longitudinal RWD. That is probably the right language for your genomics clients because it avoids both investor-deck exuberance and overly narrow scientific dismissal.

The ChatGPT review is also cautious in a useful way about the MCED law: it reduces the statutory barrier but does not eliminate FDA, CMS, coding, pricing, or evidence requirements. Opus says this too, but ChatGPT places it more centrally in the policy architecture.

Important discrepancy: patent portfolio

There is a meaningful difference between the two reviews on IP. My review says the patent portfolio includes three patent families, four issued U.S. patents, two pending U.S. applications, and one pending PCT application. Opus reports a much broader portfolio: 15 issued U.S. patents, 45 issued international patents, 27 pending U.S. applications, and 107 ex-U.S. pending, with specific families covering non-bisulfite epigenetics, CRC multiomic signatures, protein/autoantibody/RNA, and multi-cancer signatures.

That is a major difference. Opus is likely drawing from a fuller IP table or section of the S-4 that my review undercounted or conflated with a narrower AI/ML patent subset. For any final client memo, I would use the larger Opus IP summary, while still retaining my qualitative conclusion: even a larger portfolio does not automatically create a fortress moat in diagnostics, especially where software, biomarkers, and diagnostic method claims face eligibility and design-around risk.

Best combined interpretation

The best final synthesis would combine the two:

Freenome is a credible but still unproven CRC-first blood-screening company using CRC as a regulatory and reimbursement beachhead for a broader personalized cancer-detection platform. Its strongest assets are the PREEMPT CRC dataset, multiomic assay design, non-bisulfite epigenetic technology, fragment-level AI/ML approach, Exact channel access, Roche strategic relationship, and payer-aware sequencing of indications. Its main risks are FDA timing, v1 versus v2 performance translation, Guardant’s head start, Abbott/Exact prioritization, payer scrutiny of false positives/downstream costs, capital needs, and the fact that MCED legislation helps the future platform story more than the immediate CRC business.

How I would present the two AI reviews

I would say:

Opus is the stronger diligence memo. It catches more concrete business issues: Abbott/Exact, Guardant Shield timing, ADLT anchor pricing, v2 bridging limitations, transaction mechanics, supplier risks, and richer IP detail.

ChatGPT is the stronger executive-policy memo. It gives a cleaner conceptual frame for clients: CRC beachhead, personalized cancer-detection platform, Medicare/USPSTF/statutory reimbursement logic, and a sober distinction between “pathway created” and “coverage guaranteed.”

The best final product should use Opus as the fact-density layer and ChatGPT as the organizing frame.

# # #

Claude Opus 4.7 Reviews SPAC S4 (May 2026)

Freenome / PCSC S-4 — Strategic Briefing

Filing: Perceptive Capital Solutions Corp. S-4, filed April 28, 2026 (658 pages). Proposed business combination valuing Freenome at ~$1.05B post-money equity, with $240M PIPE and a $250M minimum aggregate transaction proceeds condition. Roche will own ~18% of New Freenome post-close; Perceptive PIPE ~10%; existing Freenome holders ~50%. To-date Freenome has raised ~$1.6B; accumulated deficit $1.3B; net loss $219M in 2025; cash and securities $216.7M at YE25; runway "into 2028" with deal proceeds.

1. Commercialization plan — what they actually intend to do

Anchor product: SimpleScreen CRC v1. PMA submitted Q3 2025; FDA action expected 2026; planned commercial launch 2H 2026 if approved. Performance from PREEMPT CRC (~27K evaluable in intended-use population): 81% CRC sensitivity at 90% specificity; 14% advanced adenoma (AA) sensitivity; 64% stage I sensitivity. Trial enrolled ~48K patients across 201 U.S./UAE sites.

v2 upgrade in development. Case-control bridging data presented at ASCO GI 2026: 85% CRC / 22% AA at adjusted 90% specificity, plus a 2.6-fold improvement in limit of detection and increased automation (~95%). Plan: panel-track PMA supplement submitted 2H 2026. This is differentiating data if it survives prospective validation — but the v2 number was generated on a ~1,300-sample case-control study and on the previously-locked v1 classifier, not a prospective registrational study. The v1-to-v2 head-to-head improvement was modest in their own paired analysis (+1.7 pp CRC, +5.4 pp AA), suggesting the larger numbers come substantially from cohort enrichment vs. the PREEMPT intended-use distribution.

Channel strategy is the big bet — Exact Sciences exclusive U.S. license (Aug 2025). Exact gets exclusive U.S. CRC blood rights and pays $75M upfront + up to $700M in milestones + up to 10% royalties (post-FDA) + $20M/yr R&D funding for 3 years + a $50M convertible note. Exact brings ~260K provider relationships, hundreds of health systems, EHR integrations, and a 1,400-person commercial force. Freenome retains rights when blood CRC is co-ordered with other cancer tests (lung + 10+ planned indications) — a deliberate carve-out to preserve their multi-cancer ambition.

Major strategic complication you should flag: In December 2025, Abbott announced it would acquire Exact Sciences. The collaboration "continues to be binding on Abbott," but Freenome's own risk language explicitly warns that if Abbott deprioritizes the contract, timelines and commercialization could suffer. Freenome's CRC commercial fate is now tied to Abbott's diagnostics strategy, not Exact's. Worth a direct question to management: what termination/diligence protections survive change-of-control, and what is Abbott signaling about Shield-vs-SimpleScreen positioning post-close?

Ex-U.S.: Roche License & Option (Nov 2025). Roche Sequencing gets an option (not yet exercised; SBX platform not yet delivered to Freenome as of YE25) for exclusive ex-U.S. rights to kitted CRC and lung assays. $75M option fee + $75M convertible note + up to $100M milestones + up to $24M SBX R&D milestones + low-single-digit to mid-teens royalties. Freenome will port assays to Roche's Sequencing-by-Expansion (SBX) platform — an emerging non-Illumina chemistry. Freenome retains U.S. kitted rights and global centralized rights. The option exercise trigger is staggered: tied to FDA approval/clearance of CRC + lung + additional indications.

Pipeline beyond CRC:

Lung v1 as an LDT in 2H 2026 (no FDA approval needed for LDT today — see §6); PMA pathway via PROACT Lung study (8K enrolled). Discovery-cohort case-control read showed 80% sensitivity at 75% specificity in multi-omic configuration.
Personalized Multi-Cancer Detection (PCD) panels across ~10+ indications (esophageal, liver, pancreatic, gastric, etc.) anchored to overlap with CRC and smoking populations.
Average-risk MCED kept as future optionality — explicitly contingent on evidence, FDA path, payor signals.
Plus exploratory MRD and pharma collaborations (Genentech, Gilead, Novartis, ADC Therapeutics).

TAM claim: ~$50B collective U.S. market across CRC + adjacent overlapping indications. Worth noting: this is derived using "$592 per test pricing similar to the [proposed] Nancy Gardner Sewell Act" — i.e. they are anchoring the multi-cancer TAM on a Medicare MCED rate that doesn't apply to single-cancer tests. I'd treat this number as illustrative, not bankable.

2. Reimbursement — pulling the threads together

This is the section where the company is most candid, and where your USPSTF/statutory framing matters most.

Their own legal/regulatory framing (Risk Factors, lines ~6937–6960; Payer Coverage section ~18717–18820):

"Traditional fee-for-service Medicare generally does not cover screening tests, which are considered preventive services, that are performed in the absence of signs or symptoms of illness or injury, unless there is a statutory provision that explicitly authorizes coverage of the test."

CMS can cover additional preventive services via an NCD process only if the service is recommended USPSTF grade A or B, and "the USPSTF generally waits for regulatory authorization (e.g., FDA authorization) before it considers undertaking reviews of novel technology."

Alternative path: "unless Congress enacts a statutory benefit authorizing coverage for multi-cancer early detection or similar screening tests."

This is the regulatory architecture you're already familiar with — and Freenome describes it accurately.

Where it gets interesting — the Nancy Gardner Sewell MCED Screening Coverage Act:

The bill was signed into law on February 3, 2026, as part of H.R. 7148 (FY2026 Consolidated Appropriations). It creates a Medicare benefit category for MCED screening tests that are FDA-approved (or 510(k) authorized) and that CMS determines coverage is appropriate for, beginning in 2028.

Notice how Freenome characterizes this in the S-4 (which was filed April 28, 2026, ~12 weeks after signing):

Risk Factors language repeatedly frames the statute as something that "may require" or "unless Congress enacts" — written in the conditional, despite passage. This is technically defensible because (a) the law applies to MCED tests specifically, and (b) CMS still has discretion ("if appropriate") and operational rulemaking ahead, and (c) the law takes effect in 2028.
The only explicit references to the Act appear in the strengths section: a TAM-sizing anchor at "$592 per test pricing similar to the … Nancy Gardner Sewell Medicare Multi-Cancer Early Detection Screening Coverage Act taking effect as soon as 2028" (lines 17885–17886). The word "proposed" and the phrase "as soon as 2028" both subtly understate the current state of the law.

The crucial distinction for your clients: The Sewell Act helps MCED tests (which Freenome does not yet have, and which Freenome's lead product is not). For Freenome's single-cancer CRC product, the relevant Medicare pathway is the already-existing NCD 210.3 for blood-based CRC screening — the same NCD that covers Guardant's Shield (which received ADLT status April 1, 2025 at $1,495/test). That NCD requires (i) FDA approval, (ii) ≥74% CRC sensitivity, (iii) ≥90% specificity, (iv) age 45–85, average risk, asymptomatic, q3-year frequency. Freenome states they "are one of only two companies (including Guardant Health) with a blood-based CRC test that has completed FDA validation studies that meet the requirements for CMS coverage." So for v1, the CRC NCD — not Sewell — is the operative reimbursement door.

For Freenome's eventual MCED ambitions, the Sewell Act is genuinely material — it eliminates the prior need to wait for USPSTF A/B recommendation + NCD or for separate legislation. But (a) Freenome doesn't have an MCED product yet, (b) FDA approval of MCED is required, (c) coverage starts no earlier than 2028, (d) CMS pricing for MCED is unsettled.

Coding/MolDx: Freenome flags that CPT codes, Z-Codes, LCDs, and NCDs have not been pursued because no product is commercial yet. They will likely pursue ADLT status post-FDA approval (Shield's $1,495 ADLT rate is the operative anchor for what a similarly-positioned CRC blood test can extract from Medicare in the first nine months).

Commercial payors: No contracts yet, no negotiated rates. Note their honest framing that "tests deployed at scale may face incremental scrutiny from third-party payors given the potential for false positives on an absolute basis and the additional costs associated with confirmatory diagnostic procedures." This is the real near-term private-payor obstacle for any blood-based CRC test — including Shield — and it is independent of FDA status.

PAMA: Next private-payor reporting cycle deferred to 2026; 15% annual CLFS cut cap extended through 2028. This protects Medicare CLFS rates short-term but creates uncertainty in 2027+.

3. Scientific / patent moats — what's distinctive, what's table stakes

Genuinely differentiated elements:

Non-bisulfite, base-level epigenetic sequencing. Most methylation cfDNA assays (including Shield) rely on bisulfite or enzymatic conversion. Freenome's claim of base-level resolution without bisulfite chemistry — preserving DNA integrity and allowing fragment-level features — is technically distinctive and supports their "fragment-level deep learning" (FLDL) architecture. The relevant patent family includes 5hmC sequencing, single-stranded DNA methylation sequencing, and improved methylation sequencing, with three issued U.S. patents and six pending applications, expiring 2031–2045.
Multiomics integration. DNA + RNA + protein + autoantibody. The autoantibody capability came via the 2023 Oncimmune acquisition (EarlyCDT Lung) — that is a meaningful tuck-in, since autoantibody signatures complement low-shedding tumors (prostate, early lung adenocarcinoma) where ctDNA-only approaches underperform. Eight patent families on CRC-specific multiomic signatures, with applications expiring 2041–2045.
PREEMPT CRC dataset. ~48K enrolled / ~27K evaluable in average-risk population, up to 5 years of follow-up on subsets. This is among the largest prospective CRC blood screening datasets in existence — comparable in scale to Guardant's ECLIPSE — and is a real moat if used to train and update classifiers.
CSO (Cancer Signal Origin) algorithms for tissue-of-origin localization, designed to allow modular re-use of assays across indications.

Patent portfolio (summary as of 3/24/26): 15 issued U.S., 45 issued international, 27 U.S. pending, 107 ex-U.S. pending. Earliest expiration 2031; bulk expire 2039–2045. Notable: most of the protein/autoantibody/RNA/multi-cancer signature patents are still pending — issuance risk is real, and the "data moat" claim relies more on accumulated training data + trade secrets than on issued blocking patents.

Table-stakes / not differentiating:

AI/ML classifiers, fragmentomics, cfDNA workflows broadly — every serious entrant (Guardant, GRAIL, Exact, Natera) has parallel programs.
"Multiomics" as a concept is now industry-standard rhetoric.
The base case for CRC (sensitivity ~80%, AA sensitivity ~13–22% at 90% specificity) is comparable to Shield in v1 and modestly better in v2 case-control data — but not category-defining.

Single-source supplier risk: Illumina (sequencers/reagents; agreement runs to 2033, no minimum purchase) and New England Biolabs (sole reagent supplier; minimum purchase commitments, exclusivity warrants issued). The Roche/SBX partnership is partially a sequencing-platform hedge against Illumina concentration.

4. Competitive position — honestly assessed

Dimension	Freenome SimpleScreen CRC	Guardant Shield (FDA-approved July 2024)
FDA status	PMA submitted Q3 2025; awaiting action	Approved July 2024
Medicare coverage	None yet	Covered (NCD 210.3); ADLT $1,495 from Apr 2025
Guideline inclusion	Not yet	NCCN included June 2025
CRC sensitivity (intended-use)	81% (v1) / 85% (v2, case-control)	83%
Specificity	90%	90%
AA sensitivity	14% (v1) / 22% (v2, case-control)	13%
Commercial launch	Targeted 2H 2026	Live since Aug 2024
Sales channel	Exact Sciences (~260K providers) — pending Abbott acquisition	Direct (Guardant)

Freenome is 18–24 months behind Shield in CRC and will launch into a market where Shield already has Medicare coverage, ADLT pricing, and NCCN inclusion. Their two real angles of attack: (1) the v2 advanced-adenoma sensitivity uplift (if it holds prospectively, this is clinically meaningful since AA detection is where blood tests have been weakest vs. stool/colonoscopy), and (2) Exact's commercial reach, if Abbott honors the deal. Both are real but neither is a sure thing.

In MCED, the relevant peer is GRAIL Galleri (LDT, no FDA approval, MCED-LITE bill–dependent, struggling commercially) and emerging entrants — Freenome's risk-enriched PCD strategy (rather than pan-cancer average-risk) is a defensible product-strategy bet that aligns better with current evidence requirements and overlapping screening populations.

5. Deal mechanics and capital-structure notes

Implied EV ~$1.05B post-money. This is the lower end of the pre-deal SPAC range and below Freenome's last private rounds — a meaningful step-down from peak-2021 valuations.
PIPE composition is concentrated in insiders/strategics: Perceptive PIPE Investor and RA Capital pre-commitments, plus existing Freenome stockholders.
Convertible notes (Exact $50M, Roche $75M) convert at 1.5× and 1.2× PIPE price respectively — both above $10, but signaling that strategics priced their conversion conservatively.
Cash runway "into 2028" with proceeds. Given $200M+ annual burn even with collaboration revenue, the company will likely need additional capital before approval-driven inflection unless milestone payments accelerate.
The Aggregate Transaction Proceeds Condition is waivable by Freenome — meaning the deal can close even if redemptions exceed 77.9% and proceeds dip below $250M. This is a soft floor, not a hard one.

6. Other regulatory items worth flagging

LDT Rule vacated. FDA's May 2024 LDT rule was vacated March 31, 2025 by E.D. Texas; FDA rescinded the rule in September 2025. This is favorable for Freenome's planned 2H26 lung LDT launch — they can launch without FDA marketing authorization under current enforcement discretion. Legislative VALID-Act-style proposals remain a watch item.
Centralized U.S. testing model with Brisbane, CA CLIA lab — concentration risk if disrupted.

How I'd frame this for your two audiences

For Freenome (as client): The core strategic narrative is coherent — single-platform, multi-omic, anchored in CRC for reimbursement on-ramp, expanding into overlap-population PCD panels. But three pressure points need direct conversation:

What is the Abbott contingency plan? Exact-as-Abbott may de-emphasize SimpleScreen if Shield (acquired indirectly through Abbott's CRC diagnostics ambitions) is positioned as the lead blood CRC asset — wait, Abbott is acquiring Exact whose stool offering is Cologuard; Shield is Guardant. So Abbott would inherit a portfolio in which SimpleScreen is the blood complement to Cologuard. That's potentially fine — but Abbott's prioritization is unknown.
How will you differentiate v2 vs. Shield V2 in primary care detailing? Both have ~84% CRC sensitivity. AA sensitivity is the only headline differentiator, and v2's 22% is case-control, not prospective.
Make the Sewell Act framing more accurate. The S-4's risk language treats the law as if it might still need to happen. This understates a real positive — and may confuse investors who already track the law. (It also doesn't help your single-cancer CRC plan, but it materially de-risks future MCED.)

For your other genomics clients (objective view): Freenome is a credible, well-capitalized #2 in blood-based CRC entering a market where Guardant has 18–24 months of head start, ADLT pricing, NCCN inclusion, and growing real-world data. The scientific platform has genuine differentiation in non-bisulfite epigenetic sequencing and multiomic integration, but the patent moat leans on pending applications and trade secrets more than issued blocking IP. The deal values the company at ~$1.05B — modest by 2021 standards, defensible given the FDA-pending state of v1, but rich versus the post-IPO multiples now applied to Guardant ($GH) on a per-test-revenue basis. The Sewell Act is favorable for the eventual MCED story but not for the imminent CRC story. The single biggest near-term wild card is the Abbott/Exact transaction — it can either accelerate Freenome (Abbott has scale that Exact alone lacks) or sideline SimpleScreen entirely.

Happy to go deeper on any section — particularly the v1-vs-v2 statistical bridge, the precise ADLT pricing pathway Freenome will likely pursue, or the Roche SBX implications for their long-term sequencing-platform strategy.

CHAT GPT 5.5

Chat GpT 5.5 reviews SPAC S4 (May 2026)

[Repeat with an upload of the 600 page S4 HTMl, rather than as a link or pdf]

Yes. This time I used the uploaded full HTML S-4, not just the shorter SEC investor/425-style materials. The document is the April 28, 2026 Form S-4/proxy/prospectus for the Perceptive/Freenome business combination, with Freenome Holdings as co-registrant.

Re-evaluation: what changes after reading the full S-4

The S-4 makes Freenome look more like a platform-and-financing story built around a CRC beachhead, rather than a simple “MCED test company.” The strongest business logic is: CRC first, because it has the clearest FDA/CMS/guideline pathway; then use CRC infrastructure, Exact Sciences commercialization, Roche platform optionality, and accumulated prospective/RWD data to move toward personalized cancer detection across multiple cancer indications.

The S-4 says this explicitly: Freenome is initially focused on CRC because it is the only population-level screening indication with an established path to coverage and reimbursement, and then wants to deploy a unified assay/automation/informatics infrastructure across future overlapping cancer-screening populations.

1. Commercialization plan

Freenome’s commercial plan has three layers.

First, CRC is the anchor product. SimpleScreen CRC v1 is supported by the PREEMPT CRC study, which Freenome describes as the largest prospective study of its kind. The company says PREEMPT enrolled about 48,000 patients, evaluated about 27,000, and conducted 34,224 tests. In the intended-use population, v1 achieved 81% CRC sensitivity at 90% specificity, with 14% advanced adenoma sensitivity, 31% sensitivity for high-grade dysplasia lesions, and 64% sensitivity for Stage I cancers.

Second, Exact Sciences is the channel partner for CRC. The S-4 describes an August 2025 agreement under which Exact Sciences paid Freenome $75 million upfront, with potential milestones up to $700 million, royalties, $20 million in joint R&D funding over three years, and a $50 million convertible note. This is strategically important because Exact gives Freenome access to a primary-care CRC screening commercial machine that Freenome itself does not yet have.

Third, Roche is the platform/globalization partner. Roche appears not just as a collaborator but as a major equity holder in the pro forma company, with about 17–18% ownership depending on redemption scenario. The S-4 positions Roche as part of the broader strategy for technology collaboration and future commercialization outside Freenome’s own U.S. centralized testing path.

My read: commercially, this is not naïve. Freenome is trying to avoid the classic diagnostic trap of having an elegant test but no ordering channel, no payer logic, and no workflow adoption. Exact addresses CRC channel risk; Roche addresses industrial/platform/global optionality; CRC addresses reimbursement tractability.

The weakness is that the story becomes more complex: CRC-only economics, CRC-plus-other-test economics, Roche rights, Exact rights, U.S. centralized testing, possible future kits, LDT versus FDA-approved versions, and MCED legislation all interact.

2. Science and moat

The scientific moat is not just “blood-based methylation.” Freenome frames the moat as a combination of multiomics, proprietary non-bisulfite base-level epigenetic assay technology, AI/ML classifiers, automation, and longitudinal data.

The S-4 describes a platform that integrates molecular data from blood samples across DNA, RNA, proteins, and other analytes, with ML/DL models designed to optimize sensitivity and specificity. It also says the platform is designed to improve as real-world and longitudinal data accumulate.

The most interesting technical claim is the non-bisulfite, base-level epigenetic assay. This is potentially material because bisulfite conversion can degrade DNA and may be a limiting factor in low-signal early cancer detection. Freenome’s claim is that its assay and informatics architecture can support a common platform with single-cancer and multi-cancer classifiers.

The other significant claim is fragment-level deep learning, or FLDL. Freenome says that as tests receive approval and commercial data scale, it expects to use FLDL and longitudinal RWD tokenization to improve diagnostic accuracy and support multiple cancer-specific and multi-cancer classifiers.

My objective interpretation: Freenome’s moat is strongest if you define it as assay + prospective clinical data + workflow + regulatory submissions + commercial partners + longitudinal RWD. Its patent moat alone looks weaker than the investor narrative might imply.

3. Patent/IP moat: useful, but not impregnable

The S-4 says Freenome’s patent portfolio includes three patent families, four issued U.S. patents, two pending U.S. applications, and one pending PCT application, covering early-stage cancer detection using AI/ML classifiers and implementation of ML/AI to develop disease-detection classifiers, with potential expiration dates between 2039 and 2046 if issued/maintained.

But the document itself is cautious. It notes that diagnostic, cancer-screening, software, and machine-learning inventions may face patent eligibility challenges, and that patents may not issue, may be narrowed, may expire before or soon after commercialization, or may not block competitors from designing around them.

So I would not describe Freenome as having an obvious fortress patent estate. The better formulation is: Freenome has some patent coverage, but the more durable moat may be trade secrets, data scale, validated classifiers, lab automation, payer/regulatory know-how, and partner-controlled distribution.

4. CRC v2: better science, but not yet the regulatory product

The S-4 puts a lot of weight on SimpleScreen CRC v2. v2 is described as a comprehensive upgrade to assay, automation, and AI/ML algorithm components. In a head-to-head performance evaluation, the updated CRC test detected 85% of CRC cases and 22% of advanced precancerous lesions at 90% specificity, with improved APL and CRC sensitivity compared with v1, plus a 2.6-fold reduction in limit of detection.

This is commercially important because APL/advanced adenoma detection is the real differentiator for CRC screening. A blood test that only finds cancers may win adherence but may lose prevention value versus colonoscopy or stool DNA. Freenome appears aware of this and is trying to move from “cancer detection” toward a more prevention-relevant CRC profile.

However, v2 is not yet the initial FDA-approved product. The S-4 says Freenome anticipates submitting v2 data as part of a panel-track PMA supplement in the second half of 2026.

So for client reporting: v1 is the regulatory/commercial beachhead; v2 is the performance-improvement story.

5. Regulatory risk is more concrete than the investor deck tone suggests

The S-4 discloses that Freenome completed a PMA submission for SimpleScreen CRC v1 and received a major deficiency letter from FDA. The letter sought more information on clinical validation, additional analyses of PREEMPT CRC performance data, and analytical validation. Freenome says it submitted a complete response in April 2026 and does not expect the letter to materially affect commercialization timing, but approval remains uncertain.

This matters. A major deficiency letter is not catastrophic, but it means the FDA review is not simply a ceremonial glide path. For an objective memo, I would write: Freenome’s FDA path is advanced but still live-risk.

6. Reimbursement, Medicare, coding, USPSTF, and MCED policy

The S-4 is very clear on the central reimbursement point: traditional fee-for-service Medicare generally does not cover screening tests unless there is a statutory benefit or a preventive-service pathway such as USPSTF A/B plus CMS NCD. It states that CMS can cover some preventive services through an NCD process if the service is reasonable and necessary for prevention/early detection, has a USPSTF A or B recommendation, and is appropriate for Medicare beneficiaries. It also notes that USPSTF often waits for FDA authorization before reviewing novel technologies.

For CRC, Freenome’s path is better than for de novo MCED because CRC already has established screening infrastructure, USPSTF guidance, and CMS coverage logic for blood-based CRC screening. The S-4 explicitly says CRC has an established framework for coverage and reimbursement, including blood-based tests for most individuals 45 and older under current USPSTF and other guidelines.

For broader MCED, the S-4 includes both legacy-style cautionary language and newer assumptions around the Nancy Gardner Sewell MCED legislation. One section still says Medicare coverage may be unavailable unless Freenome obtains USPSTF support/NCD coverage or Congress enacts statutory authority for MCED-like screening tests. But another section uses a market-size assumption tied to a $592 rate proposed under the Nancy Gardner Sewell Medicare MCED Screening Coverage Act taking effect as soon as 2028.

Your note is correct: the MCED bill was signed into law on February 3, 2026. Public summaries describe it as creating a Medicare benefit/category and CMS evidence-based pathway for FDA-approved MCED tests, not as automatic coverage of any specific test. (Representative Mariannette Miller-Meeks) A more precise outside summary says coverage begins as early as 2028, is limited to FDA-approved/authorized MCED tests, and remains subject to CMS coverage determinations. (AZBio)

That distinction is essential for your clients: the law reduces the statutory barrier but does not eliminate the evidentiary, coding, pricing, and CMS implementation barriers.

On coding/payment, the S-4 says Medicare lab payment for similar tests is generally under the CLFS, with rates assigned to billing codes, and notes PAMA’s weighted-median private-payer rate-setting structure. It also discusses payer scrutiny of medical necessity, cost-effectiveness, downstream utilization, false positives, and confirmatory diagnostic workups.

7. The biggest business risks

The first risk is approval timing. Freenome is not yet an approved commercial testing company; the S-4 says it has no commercial products and has not demonstrated commercial-scale manufacturing, revenue generation, or sales/marketing execution.

The second risk is cash burn. Freenome reports net losses of $219.3 million in 2025 and $274.4 million in 2024, and an accumulated deficit around $1.3 billion. That does not invalidate the company, but it makes the post-SPAC financing and milestone story important.

The third risk is payer proof beyond analytical performance. The payer question will not be, “Is the test technically impressive?” It will be: Does it improve adherence, avoid missed cancers, reduce interval cancer burden, avoid unacceptable false-positive downstream workups, and add value compared with FIT, FIT-DNA, colonoscopy, LDCT, and existing guideline pathways?

The fourth risk is multi-cancer expansion credibility. CRC has a path. Lung may have a risk-based screening population and low LDCT adherence. But a personalized multi-cancer menu will require indication-by-indication evidence, payer positioning, and guideline logic. The phrase “one assay, multiple classifiers” is attractive, but each clinical use still needs its own clinical, regulatory, and reimbursement argument.

Final objective framing for other genomics clients

Freenome’s S-4 describes a serious, sophisticated attempt to commercialize blood-based cancer screening through a CRC-first, partner-enabled, data-platform strategy. The plan is credible because it starts where reimbursement and guidelines are most tractable, uses Exact Sciences for CRC commercialization, uses Roche for broader technology/platform optionality, and tries to convert prospective clinical data plus longitudinal RWD into an AI/ML moat.

But the story should not be reported as a simple triumphalist MCED narrative. Freenome remains pre-approval and pre-commercial revenue; its initial PMA received a major deficiency letter; its strongest CRC product version may be v2 rather than the initial v1 PMA product; its IP moat is useful but not obviously blocking; and the MCED law creates a pathway, not guaranteed payment.

The cleanest client-ready sentence is:

Freenome is best understood as a CRC-first cancer-screening infrastructure company trying to become a personalized multi-cancer detection platform; its differentiation lies less in any single biomarker claim than in the combination of multiomics assay design, prospective datasets, AI/ML classifiers, partner distribution, and payer-aware sequencing — with FDA timing, CMS implementation, USPSTF/guideline placement, and real-world clinical utility as the major gating factors.