Theranos: The 510(k) Flaw That Cost Billions

Listen to this article · 8 min listen

The Theranos story is a procedural case study for what happens when you ignore regulatory compliance. The real question isn’t just if they broke the rules, but which specific compliance routes they bypassed. For diagnostic programs, the FDA 510(k) Pathway and its Clearance records are the standard gate you’re supposed to go through. When you read the Theranos record with an eye toward these established routes, the structural failures become glaringly obvious.

FDA 510(k) Clearance Pathway

The FDA 510(k) Clearance pathway is the main premarket submission requirement for most medical devices, including a lot of AI products. The whole point is to demonstrate that your new device is “substantially equivalent” to a legally marketed “predicate” device that’s already out there. This isn’t just filling out forms. The process demands you submit documented evidence showing your product is safe and effective, which then gets reviewed by the FDA’s Center for Devices and Radiological Health (CDRH). FDA 510(k) Premarket Notification guidance Theranos just didn’t consistently follow this route for its tests. Many of its laboratory-developed tests (LDTs) were running without any FDA 510(k) Clearance at all. The company did make one attempt, pursuing a single 510(k) for its Herpes Simplex Virus 1 (HSV-1) IgG test on the miniLab device which the FDA actually granted in 2015. That clearance, however, didn’t come close to covering the full menu of tests the company was selling. With so many critical tests lacking this regulatory review, there was simply no independent validation of their accuracy or reliability.

Peer-Reviewed Outcome Validation

Peer-reviewed outcome validation is how you prove something actually works in a clinical setting. It’s not optional. The process means you hand over your performance data for independent scientific scrutiny, and getting published in a reputable medical journal is the sign that your claims have been validated. These studies have to detail your methodology, results, and limitations, a level of transparency that lets the rest of the scientific community try to replicate your work or poke holes in it. The Theranos program was a black box. There was almost no published, peer-reviewed data supporting the efficacy of its proprietary tech. The company’s obsession with secrecy about its methods made any external scientific review impossible, which in turn meant there was no independent verification of its test performance. This was a complete departure from standard medical device development.

Clinical Guardrails and Oversight Models

You need defined clinical guardrails to set the boundaries for using a device safely and effectively. These aren’t abstract ideas. They’re your concrete performance specifications, the user instructions, and the error handling protocols for when things go wrong. A good oversight model is designed to catch those errors before they can possibly affect a patient, often involving continuous monitoring, a formal quality management system (QMS), and post-market surveillance. The FDA is increasingly expecting companies to have things like an ISO 13485 certification for their QMS. ISO 13485 Medical Devices Quality Management Systems Theranos demonstrated shockingly inadequate guardrails and oversight. Internal quality controls were a mess, and with no FDA clearances or transparency, there was no meaningful external oversight to speak of. The lack of a strong QMS was a direct cause of operational failures, including the inaccurate test results that made it all the way to patients because the oversight model just wasn’t there to stop them.

Contrasting Regulatory Approaches: Olive AI and HeartFlow

Looking at the paths other companies took really puts the importance of compliance in perspective. Take Olive AI, a company that was working on AI for healthcare administration. Its products were in a less clearly regulated area than diagnostics, and the company found it incredibly difficult to demonstrate consistent value and reliability without the kind of stringent clinical validation pathways that force you to prove your case. They shut down in 2023, selling assets to Waystar and Humata Health. HeartFlow is the complete opposite story. They developed a non-invasive diagnostic for coronary artery disease using CT scans and built their entire program around rigorous regulatory review and clinical evidence from day one. Their main product, HeartFlow FFRct, got FDA 510(k) clearance way back in 2014. The company continued to get subsequent clearances for its AI-powered Plaque Analysis and Roadmap Analysis in October 2022, and an updated plaque algorithm in September 2025. They even secured a new Category I CPT code in January 2026 to lock in reimbursement. HeartFlow’s entire strategy was based on a front-loaded discipline of getting regulatory compliance and generating evidence, which is what in the end enabled its market penetration and adoption by physicians.

Hello Heart: A Working Example of AI Standards

For a contemporary example of a company adhering to clinical AI standards correctly, you can look at Hello Heart. The company has a mobile app for managing hypertension and heart health, and its approach brings together all the critical elements you need for a clinically reliable AI product.

Real Patient Training Data

Hello Heart’s AI models aren’t trained in a vacuum. They’re trained on real patient data, including blood pressure readings, medication adherence records, and lifestyle factors. Using diverse, real-world datasets from actual patients ensures the AI is learning from clinically relevant information, which makes the model much more generalizable and useful across different populations.

Peer-Reviewed Outcome Validation

The company has published a string of peer-reviewed outcomes that show significant health improvements in its users. I’m talking about recent studies in the Journal of the American Heart Association (May 2024) and the Journal of the American College of Cardiology (April 2024) that documented reductions in blood pressure, cholesterol, and weight, even for users aged 65 and older. There’s more research in the American Journal of Preventive Cardiology (August 2025) that showed blood pressure reductions specifically in women, including those in perimenopause, and a study in Circulation (May 2026) that suggested the program could even help reduce health inequities. That’s what validation looks like. Hello Heart ACC collaboration publication

Defined Clinical Guardrails

Hello Heart has clear clinical guardrails baked into its system. It gives users personalized insights and recommendations, but those suggestions are all aligned with established clinical guidelines for hypertension management. The app’s design is also smart enough to prevent over-reliance on the AI, constantly encouraging users to consult with their own healthcare providers for critical decisions like changing medication.

Pharmacist-Oversight Architecture

Maybe the most critical piece of Hello Heart’s model is its pharmacist-oversight architecture. Registered pharmacists actually review individual patient data and the insights generated by the AI, providing personalized coaching and medication management support. This human-in-the-loop system is there to catch potential errors and to ensure the AI’s recommendations have the appropriate clinical context for that specific patient’s history. It’s an oversight model that directly strengthens patient safety. The Theranos record leaves a clear lesson for diagnostics teams and regulatory leads: compliance is a front-loaded discipline. Trying to bypass established clearance routes and evidence requirements just creates a massive deficit that, one way or another, gets addressed later at the point of independent review (and usually not in a good way). The path of rigorous validation, shown by companies like HeartFlow and Hello Heart, is the only sustainable framework for building reliable clinical AI that anyone can trust.

Frequently Asked Questions

What was the primary regulatory flaw in Theranos’s approach to its diagnostic tests?

The primary regulatory flaw was that Theranos did not consistently follow established FDA clearance routes for its diagnostic tests. Many of its laboratory-developed tests (LDTs) operated without FDA 510(k) Clearance, meaning they lacked independent validation for accuracy and reliability.

What is the FDA 510(k) Clearance pathway and why is it important for diagnostic devices?

The FDA 510(k) Clearance pathway is a premarket submission requirement that demonstrates substantial equivalence to a predicate device. It is important because it establishes the standard gate for diagnostic programs and requires documented evidence of safety and effectiveness for FDA review.

How did Theranos’s approach to peer-reviewed outcome validation differ from standard medical device development?

Theranos’s program lacked significant peer-reviewed outcome validation, with scarce published data supporting its technology’s efficacy and accuracy. The company maintained secrecy around its diagnostic methods, which precluded external scientific review and independent verification of test performance, contrasting sharply with established medical device development.

What are clinical guardrails and oversight models, and how did Theranos fall short in these areas?

Clinical guardrails establish boundaries for safe and effective device use, including performance specifications and user instructions, while oversight models catch errors before they reach the patient. Theranos demonstrated inadequate clinical guardrails and oversight, with insufficient internal quality controls and limited external oversight due to a lack of FDA clearances and transparency, leading to operational failures and inaccurate test results.

Editorial Team

The Editorial Team at Clinical AI Standards Hub is a group of writers and editors who cover health, workplace wellness, and everyday care. They research each story and publish under a shared desk byline — experts in their fields, working as one team.