Medicare's Pivotal Shift Towards AI-Driven Healthcare
The Centers for Medicare & Medicaid Services (CMS) has initiated a groundbreaking transformation in its payment landscape, introducing new models explicitly designed to accommodate and incentivize artificial intelligence (AI) in healthcare. This strategic move, largely unnoticed by much of the broader tech world, establishes mechanisms to pay for AI agents that can monitor patients between visits, facilitate check-ins, coordinate referrals, and ensure medication adherence. The new models, particularly the Advancing Chronic Care with Effective, Scalable Solutions (ACCESS) Model and the Wasteful and Inappropriate Service Reduction (WISeR) Model, represent a significant departure from traditional fee-for-service structures, aiming to align payments with patient outcomes and reduce administrative burdens.
Historically, AI integration into Medicare's payment systems has been an awkward fit, with many FDA-approved AI-enabled medical devices lacking a hardware component and the software itself being the service. Medicare has paid for fewer than 5% of the over 1,200 FDA-approved AI-enabled medical devices. The existing payment structures, such as the Medicare Physician Fee Schedule (MPFS) and the Inpatient Prospective Payment System (IPPS), were not originally designed for software-based technologies, leading to challenges in determining appropriate reimbursement and often bundling AI costs into broader procedures. This lack of clear reimbursement pathways has historically limited the adoption of promising clinical AI solutions.
ACCESS Model: Incentivizing Outcomes in Chronic Care
The ACCESS Model, a voluntary initiative launched by the Center for Medicare and Medicaid Innovation (CMMI), is a 10-year program designed to expand technology-supported care for Medicare beneficiaries with chronic conditions. It focuses on prevalent conditions such as high blood pressure, diabetes, chronic musculoskeletal pain, and depression, which affect over two-thirds of Medicare beneficiaries. The model introduces Outcome-Aligned Payments (OAPs), a recurring payment for managing a patient's qualifying condition, with additional payments tied to achieving measurable health outcomes. This approach rewards results rather than the volume of services, fostering flexible, technology-supported care.
The ACCESS Model aims to overcome existing barriers to implementing technological advancements in chronic disease management, encouraging the use of telehealth, wearable monitoring devices, and digital coaching tools. Applications for participation opened in January 2026, with the first performance period beginning July 1, 2026. Notably, beneficiaries can self-enroll or be referred, and the model does not require an initiating visit with an existing provider. Commercial payers representing 165 million members across Medicare Advantage, Medicaid, and commercial coverage have also agreed to align with the ACCESS Model's payment approach, signaling a broader industry shift towards outcomes-based, tech-enabled care.
WISeR Model: AI for Fraud, Waste, and Prior Authorization Efficiency
Running for six years from January 1, 2026, to December 31, 2031, the WISeR Model (Wasteful and Inappropriate Service Reduction) is a voluntary program focused on leveraging AI and machine learning to streamline prior authorizations and combat fraud, waste, and abuse in traditional Medicare. This model targets specific services and items, such as skin and tissue substitutes, electrical nerve stimulator implants, and knee arthroscopy for knee osteoarthritis, which have been identified as having a higher risk of waste or inappropriate utilization. The goal is to ensure timely and appropriate Medicare payment for select items and services, with participating technology companies providing medical necessity recommendations.
The WISeR model operates as a pre-treatment decision process, a departure from post-pay audits. While CMS touts the program as a way to incorporate private sector innovations and expedite prior authorizations, concerns have been raised regarding the potential for incentivizing denials, as model participants receive a percentage of savings from averted wasteful care. Critics highlight that similar technology models in the private sector have led to increased automatic denials without sufficient regard for patient need. Despite these concerns, CMS emphasizes that participating companies must employ clinicians to validate technologically driven determinations, and licensed clinicians, not machines, will make final decisions on denials. The model is being piloted in six states: New Jersey, Ohio, Oklahoma, Texas, Arizona, and Washington.
Challenges and the Path Forward for AI Reimbursement
Despite the promise of these new models, significant challenges remain in fully integrating AI into Medicare's payment systems. One major hurdle is the lack of a clear Medicare benefit category for many AI services, as statutory categories often do not encompass software as a service (SaaS) or algorithm-driven tools. Even when billing codes exist, regional Medicare Administrative Contractors (MACs) typically determine pricing, leading to inconsistencies across regions and limiting widespread adoption. Policymakers are grappling with how to appropriately price AI, whether payments should be separate or bundled, and how to account for AI's ability to augment clinicians and automate tasks, potentially altering the time estimates central to valuing clinical services.
To address these issues, legislative efforts like the Health Tech Investment Act, introduced in April 2025, aim to establish a dedicated pathway for Medicare to reimburse algorithm-based healthcare services (ABHS). This act proposes defining ABHS as a distinct service category, requiring CMS to place qualifying AI services into a "New Technology" Ambulatory Payment Classification (APC), and guaranteeing at least five years of separate reimbursement. Such reforms are crucial to provide clarity and predictability for medical device companies and accelerate the adoption of AI-enabled devices that enhance diagnostics and treatment. Furthermore, the shift towards value-based payment models is seen as a more effective way to incorporate AI advancements without increasing low-value spending.