<p>The US Food and Drug Administration has approved three AI diagnostic tools in the past 30 days — the highest monthly rate since the agency began reviewing AI-powered medical devices. The approvals span radiology, ophthalmology, and dermatology, reflecting the breadth of clinical areas where pattern-recognition AI has demonstrated clinical-grade performance.</p>
<h2>What Got Approved</h2>
<p>The three approvals include an AI system for detecting early-stage diabetic retinopathy from fundus photographs with 94.7% sensitivity, an AI-assisted mammography reader that reduces false negative rates by 18% compared to single radiologist review, and a dermatology tool that identifies melanoma from smartphone photos at a level comparable to board-certified dermatologists.</p>
<p>All three received De Novo authorization — the FDA pathway for novel technologies without a predicate device — rather than the less stringent 510(k) clearance, indicating the agency is building durable regulatory frameworks for AI diagnostics rather than fast-tracking them.</p>
<h2>Clinical Deployment</h2>
<p>Approval is the beginning, not the end. Large hospital systems are integrating AI diagnostic tools cautiously, typically in augmentation roles (AI flags, human confirms) rather than autonomous roles. Liability frameworks for AI-assisted diagnosis are still evolving, and most institutions wait for malpractice insurers and legal counsel to catch up with the technology before changing clinical workflows.</p>
<h2>Global Context</h2>
<p>The EU's AI Act classifies medical diagnostic AI as high-risk, requiring conformity assessment and technical documentation before market deployment. This creates a regulatory divergence where US approvals don't automatically transfer to European markets, adding complexity and cost for companies targeting both.</p>