Artificial intelligence has quietly become a routine part of American clinical practice. The U.S. Food and Drug Administration has now cleared well over 1,400 AI- and machine-learning-enabled medical devices, up from just 221 as recently as 2023 — and the pace of new clearances keeps accelerating rather than leveling off.
A milestone years in the making
Radiology remains by far the biggest category, accounting for roughly three-quarters of all AI device clearances to date. That reflects both the maturity of medical imaging AI and the relatively well-defined regulatory pathway for tools that flag or triage scans rather than make final diagnostic calls on their own. Health-focused AI companies pulled in more than $10 billion in venture funding in a single recent year, a jump of roughly a quarter over the year before, underscoring how much capital is still flowing into the space even as the market matures.
Most of these devices reach the market through the FDA's 510(k) pathway, which clears a device by showing it is substantially equivalent to one already on the market. A smaller share, around 15%, goes through the more rigorous De Novo pathway for genuinely novel devices with no existing predicate. Full premarket approval, which requires clinical trial data, remains rare for diagnostic AI, though patient-advocacy groups have pushed for it to be used more often for tools that materially influence major treatment decisions.
Where AI is actually showing up in care
Two companies illustrate how far specific use cases have progressed. One radiology AI platform, used across roughly 2,000 hospitals and processing tens of millions of cases a year, received a foundation-model clearance covering more than a dozen CT-scan conditions with reported sensitivity and specificity both above 95%. A separate stroke-detection platform, deployed at over 1,700 hospitals, has been credited with cutting the time to treatment for a dangerous type of stroke by roughly half an hour and reducing disability rates at 90 days after treatment.
Beyond imaging, the FDA has granted "Breakthrough Device" status — an accelerated review track for tools addressing serious unmet needs — to AI systems covering everything from Alzheimer's-related retinal scanning to kidney-disease progression and automated chest X-ray reporting, the last of which is designed to draft preliminary radiology reports for a physician to review rather than replace their judgment.
The limits of a fast-moving pathway
The speed of clearances has drawn scrutiny alongside the enthusiasm. Because the dominant 510(k) pathway relies on comparison to existing devices rather than mandatory prospective clinical trials, critics argue some tools reach hospitals with less real-world validation than their clinical role would suggest. For now, virtually every cleared AI diagnostic tool operates as decision support rather than an autonomous diagnostician — a radiologist, cardiologist or pathologist still reviews and signs off on the final call, and that human-in-the-loop design is likely to remain the regulatory baseline for the foreseeable future.