Subscribe

Does Medical AI Need FDA Review?

Use intended use, risk, and predicate status to understand whether medical AI may follow 510(k), De Novo, PMA, or a planned-change review.

The FDA does not review software simply because it uses AI. The first question is whether the software function meets the legal definition of a medical device based on what the developer says it is intended to do.

A function intended to diagnose, treat, prevent, or mitigate disease may require FDA review. So may software intended to affect the structure or function of the body. The intended use, patient population, output, risk, and role of the clinician matter more than the AI label.

Four Regulatory Terms To Know

  1. 510(k) clearance: the manufacturer shows that the device is substantially equivalent to a legally marketed predicate for the stated intended use. It is not a finding that the new product is better than the predicate.
  2. De Novo classification: this route can create a new device classification for a novel low-to-moderate-risk device when there is no suitable predicate.
  3. Premarket approval: PMA is the review route for many high-risk Class III devices and requires valid scientific evidence supporting safety and effectiveness for the intended use.
  4. Predetermined change control plan: a PCCP can be included in a device marketing submission to describe specified future software changes and how the manufacturer will develop, validate, and control them. It is not a separate route around FDA review.

What The FDA List Tells You

The FDA says it had authorized more than 1,600 AI-enabled medical devices for marketing in the United States as of September 2026. Its public AI-Enabled Medical Device List links each listed device to the relevant authorization database.

The FDA also says the list is not comprehensive and is updated periodically. A product can be authorized without appearing immediately, and the list does not replace the decision summary, labeling, or instructions for use.

What To Verify Before You Rely On A Device

  • The exact device name and software version
  • The intended use and patient population
  • The required input, imaging device, or data source
  • The output and the clinician's required role
  • The authorization route and decision date
  • The current labeling, warnings, and modification history

Clearance or approval does not establish that a device outperforms a clinician, fits every patient group, or will work the same way in a different clinical setting. Match the public evidence to the workflow where the device will be used.

Use The Official Starting Point

Read the FDA's current AI-enabled medical device overview for the regulatory framework. Use its PCCP guidance for planned AI software changes.

This guide is educational. Regulatory strategy depends on the exact product, claims, risk, evidence, and current FDA requirements.

Key Points

  • Start with the software function's intended use, patient population, output, and risk
  • 510(k), De Novo, and PMA are different review routes with different regulatory jobs
  • A PCCP describes specified future changes and how they will be controlled
  • The FDA list is useful but not comprehensive and does not replace current labeling
  • Authorization does not establish superiority or local clinical fit

Articles In This Area

10 min read

FDA-Cleared AI Medical Devices Explained

What FDA clearance means for AI medical devices, how the regulatory pathways work, and what clinicians need to understand before deploying FDA-cleared AI tools.