Medicare Started Using AI to Decide What Gets Paid. Here's What It Means for Your Notes.

What WISeR is

WISeR, the Wasteful and Inappropriate Service Reduction model, is a Traditional Medicare demonstration run by the CMS Innovation Center. It uses artificial intelligence and machine learning, paired with human clinical review, to run prior authorization on a set of selected services. It runs from January 2026 through 2031 in six states: New Jersey, Ohio, Oklahoma, Texas, Arizona, and Washington.

An important clarification, because it is widely misunderstood: WISeR targets specific procedures (for example nerve stimulation, spinal procedures, knee arthroscopy for osteoarthritis, and skin substitutes), not routine home health visits or therapy notes. A provider submits a prior authorization request, an algorithm makes the first pass to approve or flag it, and a human clinician reviews every denial. WISeR does not change what Medicare covers. It changes how the paperwork is scrutinized before payment.

So why should a home health or therapy agency care? Because WISeR is the leading edge of a clear direction of travel. Medicare and Medicare Advantage plans are steadily moving documentation review from overworked humans to tireless algorithms. The first pass on your paperwork is increasingly automated, and automation is very good at finding exactly the weaknesses that thin, rushed documentation exposes. Whether the pressure reaches your services through WISeR, Medicare Advantage prior authorization, or routine claims review, the defense is the same: notes that hold up.

Why this raises the bar on documentation

Automated review rewards specificity and consistency, and it punishes vagueness. A note that says "patient tolerated treatment well" gives an algorithm nothing to score. A note that documents objective measures, skilled interventions, and measurable progress toward the plan of care gives it exactly what it is looking for.

The uncomfortable asymmetry is this: on one side you have insurers and auditors using AI to scan for any gap they can find. On the other, you have an exhausted clinician writing from memory at ten at night. You cannot send your people into an automated review armed with a pen. The realistic response is to put the same class of tool on your side of the table, so every note is complete, skilled, and consistent before it is ever submitted.

What automated review tends to flag

• Missing skilled language. Documentation that reads as something an aide could have written, rather than care requiring a licensed clinician's judgment.
• No measurable progress toward goals. The single most common home health denial phrase is that the records do not clearly show progress toward the stated therapy goals.
• Thin or repetitive daily notes. Copy-forward notes that do not reflect the specific visit are easy to flag as not medically necessary.
• Inconsistency across visits. A patient who is "improving" in one note and "declining" in the next, with no reconciliation, invites scrutiny.
• Absent objective data. Vital signs, standardized assessments, distances, assist levels, and other measurable observations that substantiate the claim.

A handful of agencies have already put AI on their side of the documentation review. See how in the free training.

How to prepare your agency now

You do not need to overhaul your EHR or your workflow. You need every note, especially the rushed in-between daily notes, to be defensible by default. Practically, that means:

• Document at the point of care, while the specifics are fresh, rather than reconstructing from memory hours later.
• Write in skilled, objective language that names the clinical reasoning, not just the activity performed.
• Carry patient context across visits so progress toward goals is visible and continuous.
• Standardize the quality floor across every clinician, so a strong note does not depend on which clinician happened to write it.