Have you ever lost an earring, a ring, or keys at the beach? The hopes of finding them are dim as you survey the vast landscape of sand. And yet we usually make the attempt, because it doesn’t make sense to simply give up on something so valuable.
Most of us rely on a rough guess, subtle clues within the sand, or our hands and feet to dig for our lost treasure.
This is where payment integrity is today. For years, we’ve followed inherited heuristics and used antiquated tools (e.g., legacy edits, spot checks, “that claim looks fine”) to search for what was lost. And while you conduct your haphazard search, you’ve unknowingly stepped over dozens of other unseen buried treasures and lost a few more coins out of your pocket.
In the healthcare world, this is costing billions of dollars from medical fraud, waste, and abuse. To solve this, we need new technology.
And that new technology already exists. While some are still “crawling through the sand” to find incorrect claims and billing errors, the more efficient and determined among us are using a “metal detector” in the form of AI and machine learning.
Where the treasure hides today
According to the Government Accountability Office (GAO), the federal government made an estimated $162 billion in improper payments in fiscal year 2024. But where is that money going?
1. Pharmacy — overspending on high-cost drugs that never help a patient
Pharmacy spending represents a significant component of federal healthcare outlays affected by improper payments.
Waste can occur when high-cost therapies are discontinued early, never administered, or otherwise unused.
When patients don’t show up for treatment, finish treatment earlier than expected, or receive their medication due to shipping errors, drugs are unused and money is wasted. Below shows how our “metal detector” works to efficiently address these issues.
Split-fill rule for first dose
- How it works: Edits any initial specialty‐drug claim greater than $15,000 to pay only 20% of quantity initially. The remainder is released after a second paid claim (administration or refill) arrives within 30 days.
- Waste captured: This addresses abandoned starts and rapid discontinuation of medication.
- Proof: Peer-reviewed studies show that split-fill programs for high-cost specialty medications can reduce waste and unnecessary spending when therapy is discontinued early.
Therapeutic-window denial
- How it works: The system builds diagnosis-to-drug timing windows (for example, chemotherapy cycles or biologic dosing intervals for rheumatoid arthritis) and rejects or pends claims that are submitted earlier than the evidence-based minimum days’ supply allows.
- Waste captured: This approach captures waste caused by stockpiling medications and refilling prescriptions earlier than clinically appropriate.
- Proof: Similar utilization-timing logic that is already in play can be implemented within claims-processing systems to identify early or duplicate billing.
Site-of-care flagging
- How it works: The system crosswalks place of service (POS) code and the provider's national provider identifier (NPI) to the applicable fee-schedule tiers. If the hospital outpatient rate exceeds 125% of the ambulatory surgical center (ASC) or home-infusion median rate, the claim is automatically pended for clinical override.
- Waste captured: This helps to capture outpatient infusion mark-ups.
- Proof: Multiple analyses have shown that care delivered in hospital outpatient departments is often reimbursed at higher rates than the same services provided in physician offices or other non-hospital settings.
High-cost claim with no subsequent spend trigger
- How it works: The system runs a quarterly look-back to identify members with biologic claims of $50,000 or more who have no follow-up drug or office visit claim within 60 days. These cases are referred to the Special Investigation Unit (SIU) for review of potential wasted inventory or fictitious billing.
- Waste captured: This approach captures waste related to phantom claims or discarded doses.
- Proof: Payers commonly use post-payment review processes to investigate high-cost claims with no subsequent evidence of follow-up care.
2. Laboratories — “rainbow” blood panels
Labs are a perennial slice of improper payments as well. Duplicate panels, extra tubes of blood, and repetitive genetic screens all contribute to a large portion of waste.
72-Hour duplicate panel edit
- Logic: The system rejects any Complete Blood Count (CBC), Basic Metabolic Panel (BMP), or Comprehensive Metabolic Panel (CMP) that is billed within 72 hours of an identical paid claim for the same member, regardless of provider. An override requires submission of a K3 modifier along with a medical-necessity note.
- Expected impact: Duplicate laboratory testing is a well-recognized contributor to unnecessary healthcare spending, and this edit is expected to reduce those costs.
- Evidence and notes: Duplicate chemistry panels and repeated CBC, BMP, or CMP orders are a common improper‑payment driver in claims data.
365-Day genomics look-back
- Logic: The system maintains a rolling 12-month history table keyed to the combination of current procedural terminology (CPT) codes and members. Repeat hereditary cancer or pharmacogenetic panel claims are denied unless a new ICD-10 diagnosis meets the required evidence tier.
- Expected impact: The US Department of Health and Human Services (HHS) Office of Inspector General (OIG) has identified substantial improper payments associated with high-cost molecular pathology and genetic testing.
- Evidence and notes: OIG has documented substantial improper payments related to molecular pathology and genetic testing, often driven by insufficient oversight and inappropriate repeat testing.
Reference-price adjustment
- Logic: For the top 50 laboratory CPT codes, the system pays the lesser of the billed charge or 125% of the regional median allowed amount. Providers may appeal this determination by submitting supporting chart notes.
- Expected impact: This approach is expected to generate 10–15% unit-price savings without requiring broad contract renegotiation.
- Evidence and notes: Reference pricing for common laboratory tests is associated with lower prices and reduced overall spending, as well as shifts in testing volume toward lower‑priced labs.
3. Copy-paste care — therapy notes with no pulse
Behavioral health providers can bill hour-long sessions while reusing last week’s narrative—no outcome, no arc.
But, with a metal detector approach, natural-language processing can be used to score clinical note similarity and flag notes that are more than 85% identical across visits.
When this is coupled with a flat PHQ-9 score over time, the approach precisely and efficiently identifies potentially hidden issues.
4. Diagnosis codes that inflate risk scores
The HHS Office of Inspector General’s 2024 audit of Medicare Advantage found that billions of dollars in risk-adjusted payments were driven by diagnoses reported through chart reviews and health risk assessments that lacked evidence of related services in other medical records.
Cross-stream AI that combines claims, electronic health records (EHR), and device data can detect and address these issues before the risk-adjustment engine runs.
How would you recover your lost treasure?
We have three choices moving forward:
- Use the plastic shovel: To continue on as we always have, sorting through the sand manually, to eventually find some of the treasure.
- Stop searching: Accept all the loss as inevitable.
- Pick up a metal detector: Leverage new technologies to find more treasure faster and more accurately.
Despite decades of program-integrity efforts, GAO continues to report persistently high levels of improper payments across federal healthcare programs.
The good news? The right tools are at your disposal:
- You have the data. Every specialty-drug dispense, duplicate lab CPT, and place-of-service code flows through the adjudication engine today.
- Edits flip on like light switches. A 72-hour duplicate-panel rule, a split-fill specialty-drug edit, or a reference-pricing table drops into the rules library in a weekend release.
Stick with manual methods—or fix the problem?
It’s your call.
To learn how Machinify can future-proof your payment integrity strategy, contact us today.
The content on this page is subject to our Terms of Use.
