- $4.3 billion: Annual improper payment rate for skilled nursing facilities in 2025, driven by administrative errors.
- 67%: Rate of overturned Medicare Advantage prior authorization denials when appealed.
- 4.1 million: Prior authorization requests denied by Medicare Advantage insurers in a single year.
Experts agree that the adoption of AI tools like PReP Audits is a necessary response to the escalating algorithmic scrutiny in healthcare reimbursement, though strict human oversight remains critical to mitigate legal risks.
The AI Arms Race in Healthcare: Providers Fight Back Against Algorithmic Audits
BOSTON, Mass. – November 18, 2026 – The modern healthcare reimbursement landscape is no longer just a debate over medical necessity; it has rapidly evolved into a high-stakes war of algorithms. For years, commercial insurers and federal contractors have deployed sophisticated machine learning models to scrutinize claims, flag anomalies, and trigger audits. Today, post-acute care providers are officially arming themselves with equivalent firepower to defend their operations.
Profility, a developer of artificial intelligence solutions for post-acute care, announced the launch of PReP Audits. The new platform is designed to help skilled nursing facility (SNF) operators, contract therapy providers, and reimbursement consultants automate their responses to Additional Documentation Requests (ADRs) and other complex payer audits. By integrating directly into electronic medical records (EMR) and billing systems, the technology aims to level the playing field in an environment where algorithmic scrutiny is threatening billions in earned revenue.
Defending $4.3 Billion in Revenue Integrity
The financial urgency driving this technological shift cannot be overstated. In fiscal year 2025, the Centers for Medicare & Medicaid Services (CMS) reported an 11.8 percent improper payment rate specifically for skilled nursing facilities. That figure represents approximately $4.3 billion in misallocated funds, sitting well above the 6.55 percent error rate for Medicare fee-for-service overall.
Crucially, the vast majority of these payment errors are not rooted in outright fraud, but in administrative friction. Insufficient documentation accounted for 51.5 percent of all improper payments across Medicare last year. For nursing facilities, missing Case Mix Group components, undocumented HIPPS level changes, and absent physician orders are the primary culprits. When an audit request arrives, facilities are typically forced into a frantic, manual scramble to pull years of clinical and billing records to prove the care was justified.
"Audits have become a contest of AI against paperwork. Payers and their auditors use AI to choose which claims to review, while most providers still answer by hand," said Ohad Barzilay, CEO of Profility. "Providers deserve AI on their side. PReP Audits finds the evidence, builds the case, and keeps clinicians in control. It extends our AI suite from securing authorizations while care is delivered to defending claims long after it's been delivered."
The newly launched system automates the heavy lifting of audit defense. Upon uploading an audit letter, the platform reads the specific demands, extracts the exact records requested for the dates under review, and proactively flags missing documentation. It then generates a drafted medical necessity response, complete with citations to applicable Medicare regulations and supporting clinical evidence.
The Healthcare AI Arms Race
The introduction of defensive AI for providers comes at a critical juncture. The healthcare industry is currently entrenched in an arms race where both payers and providers are utilizing sophisticated algorithms to gain a strategic advantage. Major commercial Medicare Advantage plans have already integrated algorithmic decision-making into their claims review processes, leading to unprecedented volumes of prior authorization denials and post-payment audits.
Recent industry analyses reveal that Medicare Advantage insurers denied 4.1 million prior authorization requests in a single year. Yet, when providers have the resources to fight back, they frequently win; approximately 67 percent of appealed prior authorization denials in Medicare Advantage are ultimately overturned. The bottleneck has always been the sheer administrative cost of mounting those appeals.
Federal regulators are also escalating their technological capabilities. CMS recently launched the Wasteful and Inappropriate Service Reduction (WISeR) initiative, a six-year pilot program deploying machine learning for pre-payment reviews and medical necessity determinations. Medicare Administrative Contractors (MACs) and Recovery Audit Contractors (RACs) are simultaneously utilizing deep learning and natural language processing to analyze unstructured EHR data, identifying documentation gaps with ruthless efficiency.
Fighting this level of automated scrutiny with highlighters, sticky notes, and manual chart reviews is no longer a viable business strategy for post-acute operators. To survive the margin compression caused by line-item underpayments and vague downcoded claims, facilities are being forced to adopt matching technology.
Easing Burnout and the Human-in-the-Loop Imperative
Beyond revenue protection, the automation of audit responses addresses a critical operational crisis in post-acute care: chronic clinician burnout. Responding to targeted probe and educate (TPE) reviews or RAC audits traditionally pulls directors of nursing, therapists, and MDS coordinators away from patient care for hours, if not days.
"Profility marks a meaningful evolution in how we work, moving us from manual, paper-based processes to technology-supported ones," said Gemma Magyawi, MDS consultant and Clinical Coach at Continuing Life, a Profility customer. "What I value most is the time it gives back to our clinicians. Rather than manually tracking ADRs and compiling information, they can focus on assessing residents, collaborating with teammates, and providing compassionate care. Although clinical judgment still has the final say, the time saved with this technology is invaluable."
Magyawi's caveat regarding clinical judgment highlights the most significant regulatory hurdle in healthcare AI: compliance and liability. Using generative AI to draft medical necessity appeals introduces substantial legal risks. The Department of Justice recovered a record $6.8 billion in False Claims Act settlements in fiscal 2025, with $5.7 billion originating from the healthcare sector. If an AI tool hallucinates a clinical finding or misinterprets a regulatory code, the provider—not the software vendor—bears the ultimate legal responsibility for submitting a false claim.
The Office of Inspector General (OIG) has issued stark warnings about "automation bias," urging healthcare organizations to ensure that human oversight is rigorous and meaningful. Relying blindly on algorithmic recommendations without independent validation is a fast track to federal scrutiny.
Profility's architecture appears explicitly designed to navigate this regulatory minefield. The platform enforces a strict "human-in-the-loop" protocol. A licensed clinician from the facility must review, edit, and formally approve the AI-generated response before it can be submitted to the payer. This workflow ensures that clinical oversight remains intact, fulfilling OIG compliance best practices while still eliminating the tedious data-gathering phase of the audit process.
As Medicare Advantage grows to cover more than half of all Medicare beneficiaries, the volume of automated denials and documentation requests is mathematically guaranteed to rise. Solutions like PReP Audits, built on the same engine as Profility's widely adopted PReP Authorize platform, represent a necessary evolution in revenue cycle management. In a reimbursement landscape dictated by machines, post-acute providers are finally realizing that the only way to beat an algorithm is with a better algorithm.
Topics & Related
📝 This article is still being updated
Are you a relevant expert who could contribute your opinion or insights to this article? We'd love to hear from you. We will give you full credit for your contribution.
Contribute Your Expertise →