Not legal advice; educational only. Cases are described according to their public posture; settlements resolve allegations without an admission of liability.
If you want one story that proves an outsider with a billing rule and a claims dataset can do this, at scale, before the phrase “data miner” even existed, it is this one.
Thomas Schuhmann and Leatrice Ford Richards were not employees of any hospital they sued. They were Medicare reimbursement specialists, one a healthcare-reimbursement consultant, one a cardiovascular nurse, who understood, deeply, how hospitals code and bill cardiac procedures. They had no whistleblower’s insider documents. What they had was fluency in one rule and the discipline to read the codes against it.
The rule, and the pattern
Medicare’s National Coverage Determination for implantable cardioverter defibrillators (ICDs), the devices that shock a failing heart back into rhythm, bars implanting them within certain waiting periods: too soon after a heart attack, or after bypass surgery or angioplasty. The waiting periods exist because the medical evidence showed the devices did not benefit patients implanted that early.
Schuhmann and Richards worked from the procedure and diagnosis codes hospitals submit to Medicare and asked a simple, rule-bound question: which ICD implants fell inside the prohibited windows? They did not need a single hospital’s internal files. The timing was legible in the claims data itself, the implant date, the recent cardiac event, and the pattern repeated across the country. This is the asset-misappropriation/false-billing branch of the ACFE (Association of Certified Fraud Examiners) Fraud Tree, found not through a tip but through a code review measured against a coverage rule.
The outcome
The Justice Department took the theory and ran it as a nationwide sweep. By 2015, more than 500 hospitals had settled for over \$250 million, one of the largest coordinated False Claims Act recoveries in health care, and the two relators were reportedly awarded \$38,227,500 for originating it.
| The cardiac-device case | Figure |
|---|---|
| Hospitals that settled | 500+ (across 70+ settlements) |
| Total recovered (2015) | \$250 million+ |
| Reported relator award | \$38,227,500 |
| Insider access required | None, public claim codes + one coverage rule |
Why it worked where pure statistics failed
Compare this to the Integra upcoding cases, which were dismissed because a high coding rate is “consistent with both” fraud and good medicine. The ICD theory did not have that weakness. A device implanted inside a bright-line, rule-defined waiting period is not a matter of clinical judgment that could be lawful or not, it is a claim measured against a specific, published coverage requirement. The data did not merely suggest something might be wrong; tied to the NCD (National Coverage Determination, the Medicare rule that sets coverage limits), it identified claims that should not have been paid.
That is the difference the whole section keeps returning to, and the reason this case is such a clean teacher: the closer your data sits to a specific, identifiable program rule, the harder it is to wave away with an innocent explanation. It is also exactly the logic behind the public-data screen, pick a service, pick the rule, and let the codes show you who is outside it.
The takeaway
Two people, no insider access, one coverage rule, and the public claims data turned into a quarter-billion-dollar recovery and a \$38 million award. It predates the modern “data miner” label, but it is the model the DOJ’s (Department of Justice’s) 2026 FOCUS initiative (the DOJ “Fraud Oversight through Careful Use of Statistics” initiative) is now openly recruiting, and the clearest proof in this section that the skill that matters is not access. It is knowing the rule, and reading the codes against it.
The sweep, and the principle
What turned two analysts’ code review into a 500-hospital recovery was the government’s willingness to run it as a coordinated national sweep. Once the theory was proven against one hospital’s claims, the same coverage rule applied to every hospital’s claims, and the Justice Department pursued them in waves of settlements rather than one trial. That is the multiplier a bright-line rule provides: it scales across an entire industry at once.
It is worth naming the principle directly, because it is the most transferable lesson in this section. A statistical outlier invites the response “there is an innocent explanation.” A claim measured against a specific, published rule does not, the Integra upcoding cases failed on the former, this case won on the latter, and the data underneath was equally public in both.
| The data-miner moment, in numbers | Figure |
|---|---|
| Share of qui tam complaints filed by data miners since FY2024 | 45%+ |
| FY2025 FCA recoveries (record) | >\$6.8 billion |
| FY2025 health-care share | >\$5.7 billion |
| Outside analysts who originated this case | 2 (no insider access) |
Sources: DOJ FOCUS initiative (2026); DOJ FY2025 recoveries.
The ACFE frames professional skepticism as withholding belief from the innocent and the guilty story until the evidence converges; a coverage rule is what makes the evidence converge. It is exactly what a public-data screen is built to exploit, and the reason the government’s 2026 FOCUS initiative now prizes analysis anchored to a documented requirement.
By Noah Green CPA CFE, for Sheepdog Prosperity Partners. Educational only; not legal advice.
Primary sources: DOJ, Nearly 500 hospitals pay over \$250M to resolve ICD allegations · 31 U.S.C. § 3730 · ACFE, Fraud Tree