Still Billing, Still Free: 10 Healthcare Providers Our Algorithms Flagged — Who Haven't Been Caught

Across three of our healthcare data platforms — OpenMedicare, OpenPrescriber, and OpenMedicaid — we've built machine learning models trained on confirmed fraud cases from the HHS Office of Inspector General's exclusion list. These models have flagged hundreds of currently-practicing providers whose billing patterns closely match those of convicted fraudsters. None of the providers below have been charged with fraud. But the numbers demand attention.

The Full Scale of What We Found

These 10 providers are the tip of the iceberg. Across our three platforms, our machine learning models have flagged $3.5 billion in suspicious billing:

That's $3.5 billion in billing by providers whose patterns match those of confirmed, convicted healthcare fraudsters. For context, the DOJ's entire Healthcare Fraud Strike Force recovered $1.9 billion in 2024. Our models suggest the amount of fraud-patterned billing currently flowing through Medicare and Medicaid is roughly double what enforcement recovers annually.

The Smoking Gun: Cost Per Beneficiary

The single most damning metric in our analysis is cost per beneficiary — how much each provider spends per patient compared to their peers. Legitimate high-volume providers might have lots of patients. But when a provider is spending 10-30x more per patient than their peers in the same specialty, volume can't explain it.

Peter Alpert is the standout: $33,434 per beneficiary. That means for every patient he prescribes to, Medicare Part D pays $33,434 in drug costs in a single year. The typical infectious disease specialist generates $1,000–2,000 per patient. He is spending 17 to 33 times what his peers spend — on a per-patient basis — while carrying a 95% ML fraud score and a 39.4% brand-name prescribing rate (nearly 3x the average).

How We Built the Models

Our approach is straightforward: we trained machine learning classifiers on the billing patterns of providers who were actually caught and convicted of healthcare fraud — sourced from the HHS-OIG List of Excluded Individuals/Entities (LEIE) and DOJ prosecution records. Then we asked: who else bills like them?

What makes this analysis unique: four providers appear on multiple platforms simultaneously, flagged independently by different models analyzing different datasets. When separate algorithms looking at separate data reach the same conclusion, the signal gets much harder to ignore.

---

1. Remy Zockazock, MD

Internal Medicine · Katy, Texas · NPI 1871637157

View on OpenMedicare →

Dr. Zockazock is the most alarming case in our entire database. He is independently flagged on OpenMedicare with a 92% fraud probability — one of the highest scores in our entire dataset.

$4.1 million in Medicare payments, 48,344 services, and 11,412 beneficiaries, with a fraud probability of 92%. His services-per-day rate of 132 means he's billing Medicare for a patient interaction roughly every 4 minutes of a 9-hour workday. That volume alone is a major red flag — most internal medicine physicians see 20-30 patients per day.

He is currently still practicing and accepting patients in Katy, TX. He has positive patient reviews on Healthgrades and ZocDoc. He has not been charged with any crime.

---

2. Najmuddin Karimjee, MD

Internal Medicine · Houston, Texas · NPI 1285660241

View on OpenMedicare → · View on OpenPrescriber →

Dr. Karimjee is the single highest-cost provider in our OpenPrescriber ML-flagged dataset. Fourteen million dollars in Medicare Part D drug costs in a single year, with a 97% machine learning fraud score. That's an internal medicine doctor in Houston prescribing more drugs than most entire clinics.

To put $14 million in context: the average internal medicine provider generates roughly $200,000-$400,000 in Part D prescribing costs annually. Dr. Karimjee is generating 35 to 70 times the average. His 113,639 claims work out to roughly 311 prescriptions per day, every day of the year.

He has practiced for 37 years, accepts Medicare and Medicaid, and is affiliated with multiple Houston-area hospitals including HCA Houston Healthcare Tomball and Huntsville Memorial Hospital. He has not been charged with any crime.

---

3. Terence Frinks, MD

Family Practice · Atlanta, Georgia · NPI 1285761833

View on OpenMedicare → · View on OpenPrescriber →

A family practice doctor in the Atlanta area writing 476 prescriptions per day. Our model gave him a perfect 100% score — every single decision tree flagged him. $10.2 million in drug costs from a single family practice.

For comparison, the average family practice physician writes roughly 20-30 prescriptions per day. Dr. Frinks is writing 16 to 24 times the normal volume. Even accounting for a large patient panel, this level of prescribing would require seeing a patient and writing a prescription every 60 seconds for an entire workday.

He has been in practice for over 25 years, is affiliated with Northside Hospital - Gwinnett, and is currently accepting new patients. He has not been charged with any crime.

---

4. Vadim Baram, MD

Psychiatry · St. Louis, Missouri · NPI 1124007380

View on OpenMedicare → · View on OpenPrescriber →

Dr. Baram's profile is distinctive because of the extraordinary cost per claim. While other flagged providers generate high costs through massive volume, Dr. Baram generates $9.3 million through only 22,648 claims — a cost per claim of roughly $411. The typical psychiatrist's cost per claim is under $50.

This pattern — moderate volume but extreme cost — is consistent with prescribing expensive brand-name medications when generics are available, or prescribing high-cost specialty drugs at unusual rates. His 14% brand-name rate doesn't fully explain the disparity, suggesting the specific drugs being prescribed may be extremely high-cost specialty medications.

He is board-certified, serves as Chairman and Medical Director of Psychiatry at South City Hospital in St. Louis, and has over 17 years of experience. He is affiliated with Mercy Hospital St. Louis and SSM Health DePaul Hospital. He has not been charged with any crime.

---

5. Peter Alpert, MD

Infectious Disease · Bronx, New York · NPI 1104916907

View on OpenMedicare → · View on OpenPrescriber →

Dr. Alpert presents the most extreme cost-per-claim ratio in our top 10: $9.2 million on just 8,557 claims works out to roughly $1,075 per prescription. This is 20-40x the typical cost per claim for infectious disease specialists.

His 39.4% brand-name rate is the highest on this list — nearly three times the average. In infectious disease, brand-name preferences can be legitimately high for certain antiretrovirals and antifungals. But $9.2 million from a single provider with fewer than 9,000 claims is a severe statistical outlier by any measure.

He also carries 4 risk flags on our OpenPrescriber risk scoring system, including high cost outlier and high brand preference. He graduated from New York Medical College in 1987 and is affiliated with Montefiore Medical Center. He has not been charged with any crime.

---

6. Edwin Yau, MD

General Practice · California · NPI 1316291925

View on OpenMedicare → · View on OpenPrescriber →

Dr. Yau is flagged on both OpenMedicare and OpenPrescriber, and carries the most risk flags of anyone on this list: six separate indicators:

1. High cost outlier — spending far above peers
2. High brand preference — prescribing expensive brand-name drugs over generics
3. Opioid-benzodiazepine co-prescribing — an FDA black-box warning combination
4. Antipsychotic prescribing to elderly patients — another FDA black-box warning
5. High long-acting opioid use — associated with higher abuse potential
6. Extreme fills per patient — patients receiving far more prescriptions than normal

The combination of opioid-benzodiazepine co-prescribing and elderly antipsychotic use is particularly concerning. The FDA warns that combining opioids and benzodiazepines dramatically increases the risk of respiratory depression and death. Antipsychotic use in elderly dementia patients carries a similar warning due to increased mortality risk.

On the Medicare billing side, $3.8 million in payments with 60,161 services (165 per day) and 18,999 beneficiaries. He has not been charged with any crime.

---

7. Terrance Hughes, MD

Physical Medicine and Rehabilitation · Georgia · NPI 1659378743

View on OpenMedicare →

Dr. Hughes bills Medicare for an average of 317 services per day. That's a service every 1.7 minutes for a 9-hour workday. Physical medicine and rehabilitation typically involves hands-on patient interaction — physical exams, therapeutic procedures, injections. The idea that one physician can deliver 317 of these per day strains credulity.

He is the highest-dollar provider in our Medicare "still out there" dataset at $5.3 million, with 115,777 total services and 45,889 unique beneficiaries. The beneficiary count is also extraordinary — seeing nearly 46,000 different patients in a year would mean over 125 unique patients per day.

It's possible Dr. Hughes runs a large multi-provider clinic billing under his NPI, which could legitimately inflate these numbers. But our model evaluates individual NPI-level patterns, and his match confirmed fraud profiles at 88%. He has not been charged with any crime.

---

8. Stephen Kelly, MD

Family Practice · Oklahoma City, Oklahoma · NPI 1134167455

View on OpenMedicare → · View on OpenPrescriber →

Dr. Kelly is the most aggressive opioid prescriber on this list. More than half of everything he prescribes — 53.6% — is opioids. The national average for family practice is under 5%. He has a risk score of 70 out of 100 on our OpenPrescriber risk scoring system, with five flags:

1. Extreme opioid rate vs. peers — 10x the specialty average
2. 95th percentile opioid prescribing — top 5% nationally
3. High long-acting opioid use — associated with higher overdose risk
4. Extreme fills per patient — 13+ prescriptions per patient per year
5. High cost outlier — $3.2M total

His 23,091 total claims and 1,730 beneficiaries mean each patient is averaging over 13 prescriptions per year — primarily opioids. In the context of an opioid epidemic that has killed over 600,000 Americans since 1999, a family practice doctor writing this volume of opioid prescriptions represents an extraordinary public health concern.

He has been in practice for over 25 years in Oklahoma City. He has not been charged with any crime.

---

9. Fariborz Mortazavi, MD

Hematology-Oncology · California · NPI 1689782708

View on OpenMedicare →

Oncology fraud is among the most harmful forms of healthcare fraud, because it can involve administering unnecessary chemotherapy to patients who don't have cancer — or overtreating patients who do. Dr. Mortazavi's billing shows 63,625 services in a year (174 per day) with $2.8 million in payments and a 90% fraud probability.

Hematology-oncology is a field where high per-patient costs are expected — chemotherapy drugs are expensive, and treatment regimens involve many visits. But 174 services per day is an extreme outlier even for this specialty. The typical oncologist sees 15-25 patients per day.

Oncology has been one of the most fraud-prone specialties historically. The DOJ has prosecuted numerous oncologists for administering unnecessary chemotherapy, upcoding treatment levels, and billing for drugs that were never administered. He has not been charged with any crime.

---

10. Andrew Graf, MD

Internal Medicine · Pennsylvania · NPI 1316108269

View on OpenMedicare → · View on OpenPrescriber →

Dr. Graf rounds out our list with another perfect 100% ML score — the unanimous verdict of all 20 decision trees. $7.5 million in Part D prescribing costs, 132,160 claims (362 per day), and a 1.4% opioid rate that suggests this isn't about pills — it's about sheer volume.

362 prescriptions per day is physically impossible for a single physician to personally evaluate, prescribe, and supervise. Even if each prescription took only one minute — no patient interaction, no chart review, just signing — that would take 6 straight hours of non-stop signing. With any clinical involvement, the numbers don't work.

He has not been charged with any crime.

---

All 10 Providers: The Complete Comparison

The Bigger Picture

These 10 providers represent $69 million in billing that our models flagged as matching confirmed fraud patterns. But they're 10 out of 3,000+ flagged providers billing a combined $3.5 billion. Every one of them is still practicing. Every one of them is still billing taxpayer money.

The HHS Office of Inspector General estimates that improper payments in Medicare totaled $51 billion in 2024. Medicare fraud prosecutions, while increasing, result in conviction of only a fraction of offenders. The DOJ's Healthcare Fraud Strike Force has charged over 5,000 defendants since 2007 — but our models suggest the number of providers exhibiting fraud-like patterns is an order of magnitude larger.

We're not law enforcement. We can't determine intent, investigate patient records, or bring charges. What we can do is make the public data visible and searchable. Every provider on this list is billing public money — your tax dollars and mine. The billing data is public. The patterns are public. The question is whether anyone is looking.

Explore the Data Yourself

Related Investigations on OpenMedicare

• The Algorithm Knows — How we built the ML model that found 500 unflagged providers
• 500 Providers Still Billing Like Fraudsters — The full investigation with searchable data
• The 4,636 Impossible Doctors — Providers billing for more services than hours in a day
• Medicare Fraud Statistics 2025 — The scale of healthcare fraud by the numbers
• Pain Management Fraud — The specialty at the center of the opioid crisis
• Million-Dollar Flagged — Providers with 7-figure billing and high fraud scores
• Still Out There: Interactive Explorer — Search all 500 flagged providers by state, specialty, and score
• Top 100 Suspicious Providers — The highest-scoring providers ranked

Related on OpenPrescriber

• ML Fraud Detection — Full methodology, tier breakdown, and top 100 table
• Risk Explorer — 6,706 providers with elevated risk scores, searchable and filterable
• Fraud Hub — Overview of prescribing fraud detection across all providers

All data sourced from CMS public use files and HHS-OIG exclusion lists. Models trained on confirmed fraud cases. Being flagged does not imply guilt — it indicates statistical anomaly relative to confirmed fraud patterns. If you are a provider listed here and believe the data is inaccurate, please contact us.