### Beyond the Bundle: How AI Can Reframe the CMS Skin Substitute Debate
A fierce debate is currently unfolding in the medtech and healthcare policy space, centered on a proposal in the draft 2026 Medicare Physician Fee Schedule (MPFS). The Centers for Medicare & Medicaid Services (CMS) suggested reclassifying cellular and/or tissue-based products for skin wounds (CTPs), often called “skin substitutes,” as “incident-to” supplies. In plain terms, this means their cost would be bundled into the overall payment for a procedure, rather than being reimbursed separately. The backlash has been swift, with manufacturers like Convatec arguing that CMS lacks the statutory authority for such a sweeping change.
While the legal and policy arguments are critical, this conflict highlights a deeper, more systemic issue that we in the AI field see constantly: the struggle of analog regulatory frameworks to keep pace with technological innovation. The core of this dispute isn’t just about payment; it’s about how we define and value technology in medicine. And seen through that lens, this is fundamentally a data problem—one that artificial intelligence is uniquely positioned to solve.
—
#### The Data Blind Spot: When a “Supply” is a Sophisticated Biologic
The “incident-to” classification is typically reserved for low-cost, non-dispositive supplies—think bandages, surgical gloves, or antiseptic swabs. To lump advanced CTPs into this category is to create a false equivalence. Modern skin substitutes are not passive dressings; they are complex, bio-active technologies. Some are cellular therapies designed to actively recruit a patient’s own cells to regenerate tissue, while others are acellular matrices that provide a sophisticated scaffold for healing.
Treating these diverse technologies as a monolithic “supply” is a failure of categorization. It’s like classifying a high-performance GPU as just another “computer component” alongside a simple capacitor. The current system lacks the granularity to differentiate based on mechanism of action, patient outcomes, or total cost of care. It’s a blunt instrument in an era that demands precision.
This is where AI can fundamentally reshape the paradigm. Instead of relying on rigid, predefined categories, we can build dynamic, evidence-based valuation models. Here’s how:
**1. Quantifying Value with Real-World Evidence (RWE)**
The most powerful argument against bundling is that not all CTPs are created equal. A more expensive product might lead to significantly faster wound closure, fewer infections, and a lower rate of amputation, ultimately saving the healthcare system enormous costs down the line. The problem is proving this at scale.
AI and machine learning models are perfect for this task. By analyzing millions of anonymized data points from Electronic Health Records (EHRs), claims data, and patient registries, we can:
* **Identify Causal Links:** Move beyond simple correlation to determine the probable impact of a specific CTP on patient outcomes.
* **Analyze Sub-Populations:** A model can identify which patient profiles (e.g., diabetics with specific comorbidities) benefit most from a particular advanced therapy versus a more basic one.
* **Calculate Total Cost of Care:** AI can track a patient’s entire journey—from initial procedure through follow-up visits, hospital readmissions, and subsequent treatments—to build a comprehensive picture of the true economic value of a given technology.
With this RWE-driven evidence, a manufacturer can present a data-backed case to CMS, proving their product isn’t a simple “supply” but a high-value intervention that reduces long-term system costs.
**2. Building Predictive Reimbursement Models**
Looking forward, AI can help us move beyond static fee schedules altogether. Imagine a system where reimbursement isn’t just based on historical data but is informed by predictive analytics.
A “predictive reimbursement” model could assess a specific patient’s case—their wound characteristics, health history, and genomic data—and forecast the likely outcome and cost with different treatment options. CMS could then use this to implement a value-based payment structure that incentivizes the use of the *most effective* technology for that individual, rather than the cheapest one that fits the bundled payment. This aligns the incentives of the provider, the manufacturer, and the payer toward a single goal: the best possible patient outcome at the most efficient long-term cost.
—
#### Conclusion: Moving from Legal Fights to Data-Driven Policy
The current clash over the MPFS proposal is a symptom of an analog system straining to manage digital-age technology. While legal challenges may provide a temporary solution, they don’t solve the underlying problem. Blunt policy instruments like bundling will always risk stifling innovation and penalizing the development of next-generation therapies.
The path forward is not to argue over outdated definitions but to build a smarter, more responsive regulatory framework powered by data. By leveraging AI to analyze real-world evidence and develop predictive models, we can create a system that truly understands and rewards technological value. The conversation needs to evolve from “Can CMS legally do this?” to “How can we use technology to help CMS create policy that is as advanced as the medicine it governs?”
This post is based on the original article at https://www.bioworld.com/articles/724089-cms-draws-heat-for-skin-substitute-category-change.
