Regulatory intelligence · Atlas v1.3
Selected FDA case studies showing what entered regulation, what evidence supported it, what remains unproven, and how authorized systems are monitored after market entry.
10 representative case studies selected from 30 source-checked FDA submission records. This is not an exhaustive device list, product endorsement, purchasing guide, legal advice, or clinical advice.
Case studies
Each case separates FDA-list inclusion, disclosed AI architecture, validation, change control, and postmarket evidence.
10 case studies
Alphatec Spine, Inc. · Monitoring
An intraoperative monitoring system with an authorized classifier-change plan and a separately scoped software and interface recall.
Why it mattersHigh-value example of an AI-related surgical monitoring submission with a PCCP and a separately scoped postmarket safety record.
Brainomix Limited · Imaging
Locked deep-learning segmentation and volumetry for intracranial hyperdensities, with controlled retraining and threshold changes permitted under a PCCP.
Why it mattersCurrent high-value example of an explicit AI/ML imaging PCCP whose allowed modifications, validation controls and non-adaptive boundary are visible in public FDA material.
Mazor Robotics Ltd. · Planning
Two locked AI planning functions inside a robotic spine platform, with expert plan acceptance kept separate from placement accuracy and patient outcomes.
Why it mattersHigh-value case linking AI planning evidence to a mature robotic platform while keeping postmarket software-version risk separate from submission-level clearance claims.
Surgical Theater, Inc. · Navigation
A validated 3D U-Net vertebral-segmentation component inside a wider AR/VR planning and navigation interface, with a prior-version recall kept distinct.
Why it mattersStrong case for separating a validated AI segmentation component from the broader navigation and AR interface, while preserving prior-version postmarket attribution.
Mighty Oak Medical · Planning
A spine-planning suite whose newest clearance concerns correction simulation while its machine-learning segmentation and PCCP come from earlier submissions.
Why it mattersStrong lineage example separating what changed in the current clearance from AI evidence and change-control authority inherited from earlier submissions.
JLK, Inc. · Imaging
A dual ICH and LVO notification system in which the new clinical validation addressed LVO while the ICH component was inherited unchanged.
Why it mattersHigh-value example of notification-only labeling and of evidence inherited from a previously cleared unchanged component.
Darmiyan, Inc. · Prognosis
A prognostic De Novo case showing how demographic underrepresentation can lead to a required postmarket performance study.
Why it mattersHigh-value comparator for De Novo evidence, subgroup limitations, human-factors findings and postmarket evidence requirements.
BRAINSCOPE COMPANY, INC · Assessment
A prospective De Novo example in which both prespecified primary performance endpoints failed, yet narrow labeling and controls supported a grant.
Why it mattersCritical counterexample showing that pathway grant, prospective evidence and failed prespecified endpoints can coexist when benefit-risk, labeling and special controls are considered together.
Oculogica, Inc. · Assessment
An eye-movement concussion aid with missed performance goals, a revised clinical reference standard, and explicit limits on negative and positive results.
Why it mattersHigh-value De Novo case showing that a grant is not the same as meeting every pivotal endpoint, and that limitations and labeling are part of the authorization logic.
Augmedics, Ltd. · Navigation
An augmented-reality spine-navigation clearance that appears on FDA’s AI list although its public summary does not identify a learning architecture.
Why it mattersUseful example of cross-panel review, AI-list disclosure uncertainty and the need to attribute a postmarket action to the affected component rather than the entire platform.
No cases match these filters.
Reading rule
How AI reaches clinical regulation · Evidence matrix · Methodology · Update ledger · FDA GenAI consultation draft · Selection reviewed Sep 9, 2026 · Next scheduled review Oct 9, 2026