Substantial equivalence
The US regulatory route by which most AI medical devices reach the market, requiring only that they match an already approved device rather than prove they help patients.
To sell a medical device in the United States, a maker usually does not have to show that patients get better. It has to show the device is substantially equivalent to one already on the market, meaning it performs comparably at the same technical task. For an AI tool that means matching or beating an existing product at spotting a nodule, flagging a heart rhythm or measuring a structure on a scan.
The gap this leaves is easy to miss. Being better at detection is not the same as improving health, because a tool can flag findings that would never have caused harm, or send results to a clinician who is already overloaded. An analysis published in PLOS Digital Health in August 2026 found that of 1,357 FDA-authorised AI medical devices, only three had ever been assessed on patient-centred outcomes such as mortality or hospitalisation. So “FDA cleared” on an AI health product is a statement about equivalence, not about benefit.