Even strong diagnosis-related group (DRG) programs leave money in claims they have already reviewed. When one large Medicaid plan put its claims originally reviewed with no findings through a Machinify re-review, it recovered $6 million within four months. Read on to see how the re-review worked and what it means for your own cleared claims.
Inside a Medicaid plan's re-review of already cleared claims
The plan was already working with an outside vendor and an internal review team, and it wondered whether it was still leaving money on the table. Leadership had heard about Machinify's artificial intelligence (AI) capabilities and wanted to see what Machinify could find on its own Medicaid claims.
So, the plan tasked Machinify with two assignments. The first was a head-to-head against the existing external vendor on new claims. The second was a backsweep of claims that had already cleared with no findings.
Machinify Audit's Comprehensive DRG ran at full strength from day one. The program was already mature, built on curated cross-payer data and deep domain expertise. The DRG content, the trained models, and the certified coders and clinicians were already in place, all on one platform instead of a stack of point solutions. That content is shaped by Machinify’s physician governance program. Machinify’s medical doctors (MD) govern all of the clinical content and coding logic the platform applies, so every determination reflects current, physician-vetted standards even without a doctor in the review loop. The program was able to run independent of the plan’s existing workflows, minimizing disruption.
Machinify added to the plan's setup in stages, using a strangler-fig approach, in which new technology grows around the systems the plan already runs, assuming greater responsibility until the legacy system can safely disengage without anything breaking. That means the plan kept its existing vendor and claims system in place while Machinify grew and learned alongside it.
From there, Machinify complemented the plan's internal review team, adding specialized expertise and review capacity where the plan wanted more depth. The platform could run across the full payment spectrum, pre-pay and post-pay, and handle whichever review type the claim called for, with access to the medical record whether documentation arrived electronically or on paper. Selection scored the full population, using the complete episode of care and prior audit outcomes to steer reviewers toward the claims where the DRG code or its severity was most likely wrong. For each claim, Machinify organized the record and surfaced the source documentation inline. Expert coders and clinicians then reviewed to provide a finished, source-backed recommendation instead of building the case from scratch. Because that review runs on the same technology and content every time, the plan got a consistent, repeatable determination regardless of which reviewer touched the claim, not a result that depended on whoever happened to be on the file that day.
The re-review told leadership that claims previously determined to be clean still carried overpayments, and the dollars were still recoverable. In the first two months, Machinify identified$2.5 million in savings. That grew to $6 million within four months. Each finding averaged about $5,000, and Machinify identified savings on 42% of claims the other vendor cleared.
With the re-review delivering, the plan moved ahead onto the next phases of its agreement, starting with the launch of the full DRG program. From there, itemized bill review (IBR) has since gone live, and specialty reviews are running in several of the plan's Medicaid markets.
The plan is now working with Machinify to implement a pre-payment program, designed to drive cost avoidance and reduce administrative burden for the plan and its providers. More subproducts will roll out over the balance of 2026, including our specialty audits, like behavioral health.
This plan's experience is not unusual. Inpatient claims are some of the most scrutinized and costly that a plan pays. Even so, a DRG claim can be selected, reviewed by an outside vendor and an internal review team, and still clear with no findings. Review quality tends to depend on the individual reviewer, and the subtlest DRG errors are the easiest to overlook. Machinify’s AI-driven workflow provides a level of quality that is on par with a plan’s best expert on their best day, with that same standard applied consistently, claim after claim.
Why a clean DRG review still misses material errors
Many DRG reviews run into a few hard limits, and each one lets overpayments through.
The first limit is scope. Some DRG reviews only read the billed codes, not the medical record behind them, so errors only surface once the full inpatient stay is rebuilt.
Sepsis is one of the most common diagnoses billed as the main reason for a hospital stay. When “sepsis due to pneumonia," is noted, the claim is coded with sepsis first, which pays more than a pneumonia claim. But a sepsis diagnosis has to be backed up by the details of the patient’s chart. Sepsis means the body’s response to an infection is causing organ dysfunction. Without clinical evidence of that dysfunction in the patient's chart, the diagnosis isn’t supported. So, Machinify reviews the medical record to see whether the patient truly met the clinical standard for sepsis. If they didn’t, we remove sepsis and list pneumonia as the principal diagnosis. The claim is then paid at the pneumonia rate.
At a glance, the codes on the claim look correct either way. Only a clinical review of the chart reveals the gap.
The second limit is reach. A review examines only the claims it selects. Selection chases the claims that look suspicious, so a claim that shifts the DRG without looking unusual is unlikely to be picked. Capacity narrows it further. Record-based review is slow. Reviewers often reach only a fraction of the volume, and the plan pays the rest without a look.
Even a reviewed claim can still be incorrect. A no-find result means the first reviewer found nothing, not that nothing was there. A deeper look often surfaces hidden overpayments.
Medicaid raises the bar again. Its inpatient DRG methodology is set mostly at the state level, usually APR-DRG with state-specific weights and adjusters, and managed care plans follow the state's rules. A generalist reviewer without that grounding may clear claims a Medicaid specialist would catch. That expertise is hard to build in-house, and the toughest claims are exactly the ones that clear with errors.
These limits sit in how most DRG reviews are designed, so even a strong program can clear claims that still carry overpayments. Most plans accept that and move on.
Why the differences add up
The difference between Machinify’s results and a typical DRG review comes down to a handful of capabilities, and they map to how Machinify measures its own performance: quality, speed, scale, and accountability.
Quality. The same curated content and coding logic apply to every claim, so a determination doesn’t depend on which reviewer happens to be assigned to the file. Machinify also works from the full medical record rather than the billed codes alone, pulling documentation electronically or, where needed, through traditional channels.
Speed. Claims are selected and the record organized far faster than a manual process allows, across the full payment spectrum, pre-pay and post-pay.
Scale. One platform runs DRG alongside billing accuracy, itemized bill review, and specialty audits, so a plan isn’t assembling a stack of point solutions for each review type.
Accountability. Expert-validated, source-backed determinations are built to hold up under scrutiny, which is why audit sustainability, not the raw change rate or dispute rate, is the number that should anchor how a program gets evaluated.
Four lessons a mature program can act on
What worked here is not unique to this plan. The same few things are within reach of any mature program.
1. Re-examine your cleared claims
Many programs treat a no-find result as final, so recoverable overpayments stay on claims that look clean. A record-level re-review can bring those dollars back. For this plan, Machinify’s re-review recovered about $5,000 per finding and changed 42% of the claims.
2. Start with DRG as the first phase of a new audit program
Claims volume doesn’t slow down while most programs take months to ramp up. Machinify’s DRG review arrived pre-built and identified $2.5 million in two months. But DRG is just one of the capabilities within Machinify’s comprehensive Audit product. Starting with DRG allows plans to realize value while expanding into billing accuracy and specialty audits.
3. Judge yield on what holds up
A high change rate means nothing if the findings fall apart after a dispute, so evaluate programs on the yield that sticks. Dispute rate alone is a weak signal; some providers staff teams specifically to dispute every claim because disputing costs them nothing, and a share of appeals get paid out by default when a plan doesn’t have the resources to fight every one. An 8% dispute rate can be entirely appropriate for one plan and a sign of real inconsistency for another. What matters is what holds up after a provider pushes back. For this program, that’s a 99% audit sustainability rate. Reviewing disputes is its own drain on a plan’s resources, especially where dispute volume runs as high as 30-40% of claims. Machinify also applies its technology to the dispute queue, flagging whether a dispute reflects genuinely new information or simply restates the original claim, so a plan’s staff can spend time on the disputes that actually warrant a second look instead of working through all of them by hand.
4. Choose a model that compounds
Static models reach peak performance and then plateau. Machinify's platform learns from every determination, so the more of a plan’s hardest claims it reviews, the sharper it gets. A competitor can license the same models, but not the years of determinations behind them, so the gap only widens the longer the program runs.
Put your cleared claims to a second review
Overpayments only get harder to recover the longer a cleared claim sits, and Machinify Audit's Comprehensive DRG is built for exactly this. Our platform reconstructs the full inpatient episode, applies our clinical and coding expertise with analytics,and reprices it against payer-specific logic. Experts validate the reimbursement impact, not just the billed codes, and catch the material errors a claim-only review leaves behind.
Legacy review sees fragments of a claim. Machinify Audit Comprehensive DRG sees the whole thing.
Contact us to run a no-find re-review of your recently cleared DRG claims.
The content on this page is subject to our Terms of Use.
