Bench, Simulation, and Clinical Evidence: What Each Proves

By the Connected Diagnostics Evidence Team

Bench testing, simulation, and clinical evidence: the short answer

Bench testing, simulated-use testing, and clinical evidence each answer a different question about a connected diagnostic device. Bench testing checks whether the hardware or software meets a technical spec in the lab. Simulated-use testing checks whether a realistic user can operate it correctly. Clinical evidence checks how it performs with real patients and clinicians.

A device can pass one of these and still leave the other two questions unanswered. That’s why a single study result is never proof of every claim made about a device — the type of evidence has to match the type of claim.

Terms to know

• Bench testing (nonclinical testing): Laboratory testing of a device’s physical or software performance — for example, image resolution or whether software correctly flags a result — done outside of any real patient or user.
• Simulated-use testing: Testing where people representing real users try the device in a realistic setup, such as a mock exam room, without an actual medical decision riding on the result.
• Clinical evidence: Data gathered from real patients and real clinical use, used to show a device is safe and effective for its stated purpose.
• Pivotal clinical investigation: The specific, formally designed clinical study a company runs to support FDA marketing authorization — not just any patient trial.
• Intended use: The specific medical purpose a device is cleared or approved for. A study only supports claims that match this stated purpose — see our guide to intended use and FDA classification for how to check a device’s actual cleared purpose.

What each type of evidence can — and can’t — tell you

Bench testing

• Can tell you: Whether the device’s hardware and software meet a technical specification under controlled conditions. FDA’s device-marketing guidance points device makers to a separate document specifically for “non-clinical bench performance testing,” because this type of testing has its own standards apart from clinical study standards.
• Can’t tell you: Whether a real person can use the device correctly, or whether it changes a patient’s actual health outcome. A device can perform perfectly on a lab bench and still be used incorrectly in practice, or measure something that doesn’t end up mattering for a real diagnosis. That gap between lab performance and real-world use is a recurring theme across connected exam devices — see our broader look at evidence and limits for digital otoscopy devices.

Simulated-use testing

• Can tell you: Whether a person representing a real user — a caregiver, a clinician, a patient — can physically operate the device and follow its instructions correctly in a realistic setup. Usability problems often only show up once an actual person, not a lab technician, tries the device.
• Can’t tell you: How the device performs across a real, varied patient population, or whether it produces a correct result when disease presentation is messy, as it is in real practice. We’re describing simulated-use testing here as a general, well-established evaluation concept. The two source pages for this article don’t spell out simulated-use testing in detail, so we’re not attributing specific claims about it to either one.

Clinical evidence

• Can tell you: How the device performs with real patients and real clinicians — things bench and simulated testing structurally cannot capture. FDA’s training material on pivotal clinical investigations notes that devices often require “a particular skill level for the user that might involve training,” that design problems can surface only through real human interaction, and that patient responses, including the placebo effect, can affect results in ways a lab test never will.
• Can’t tell you: Everything, either. FDA’s marketing-guidance page confirms that not every device needs clinical evidence at all — only PMAs, HDEs, and some 510(k) and De Novo submissions require it. Whether a clinical study exists is tied to what type of submission a specific device went through, not to how serious the device looks to a reader. If you want to check what accuracy numbers from a clinical study actually mean, our guide to reading a diagnostic-accuracy study walks through that.

A step-by-step field guide: matching a claim to the right evidence

1. Identify the exact claim. Write down, in one sentence, what the device or its marketing actually claims — “detects X,” “is comfortable to use,” “improves outcome Y.”
2. Ask what kind of claim it is. Is it about hardware or software performance (bench), about whether a person can use it correctly (simulated use), or about a real health outcome or diagnostic accuracy in real patients (clinical)?
3. Look for a matching study type. A performance-spec claim needs bench data. A usability claim needs simulated-use or human-factors data. A diagnostic-accuracy or outcome claim needs clinical evidence.
4. Check whether the evidence type matches the claim type. If a usability or outcome claim is being supported only by a bench-testing result, then that’s a mismatch — the claim is going further than that specific study can support.
5. Confirm the study supports the device’s actual intended use. If a clinical study was run for one intended use, then it doesn’t automatically support a different or broader use of the same device.
6. Note what’s missing. If you can’t find any public information about which type of testing backs a specific claim, then treat the claim as unverified rather than assuming the strongest type of evidence exists.

Where the two FDA sources actually land — and where they don’t

FDA’s device-marketing guidance page confirms that bench testing has its own dedicated guidance document, and that clinical evidence is required only for specific submission types — PMA, HDE, and some 510(k) and De Novo requests — under Investigational Device Exemption and Good Clinical Practice rules. FDA’s training transcript on pivotal clinical investigations confirms the purpose of clinical studies: capturing user skill, human-factors issues, and real patient response that bench testing cannot. It also describes how bias, controls, and site selection are managed in those studies.

Neither source page provides a side-by-side breakdown of what each study type answers, and neither goes into detail on simulated-use testing specifically. This article builds that comparison from what the two pages do establish, plus general, non-source-specific description of simulated-use testing as a recognized step in device evaluation. Readers who want the primary language should read the two source pages directly rather than relying on any single summary, including this one.

What this guide doesn’t cover

This guide does not evaluate any specific device, brand, or product, and it is not a substitute for reading a device’s actual FDA clearance or approval documents. It does not tell you whether any particular device is safe, effective, or appropriate for you or someone you care for. If you have a health concern, symptom, or urgent question, contact a qualified healthcare provider or, for a medical emergency, call your local emergency services.

Medical and editorial disclaimer

This article is independent educational content published by Connected Diagnostics Evidence. It is for general information only and is not medical advice, a product recommendation, or a substitute for professional medical judgment. Connected Diagnostics Evidence is not affiliated with, and does not represent, endorse, or continue the operations, products, research, or clinical relationships of any company that previously operated this domain. This page does not diagnose, treat, or recommend treatment for any condition.

By Connected Diagnostics Evidence Editorial Team. Last updated September 9, 2026.