1,451 AI-Powered Medical Products Approved in the U.S.: Europe Lacks Even a List

5 min read

The U.S. Food and Drug Administration (FDA) has approved a total of 1,451 AI-powered medical products by the end of 2025—295 of them in the past year alone. Europe, by contrast, lacks even a comparable public database. While U.S. hospitals routinely integrate AI diagnostics into their operations, German hospitals rely on isolated solutions and await the implementation of the AI Act, whose high-risk system requirements won’t take effect until 2027. A snapshot of the current landscape.

295 approvals in a single year: what the FDA’s numbers reveal

Since 2017, the FDA has maintained a public, accessible database of all AI-based medical products approved for market release. The numbers show a clear trend: from 29 approvals in 2017 to 295 in 2025—a record that far surpasses the previous annual high of 223.

A full 76% of all approved AI devices are in radiology, where pattern recognition across large image volumes works particularly well. GE HealthCare tops the list with 120 approvals, followed by Siemens Healthineers with 89. That two of the top three manufacturers have European roots only underscores the regulatory asymmetry between the U.S. and the EU.

Yet not every approval automatically translates to clinical utility. A study published in JAMA Network Open (April 2025) examined 903 FDA-approved AI medical products and delivered sobering results: 24.1% lacked any clinical validation study whatsoever. Only 2.4% had undergone randomized controlled trials, while 8.1% were validated through prospective studies. Roughly 90% of devices, therefore, relied on retrospective data—or had no clinical evidence at all.

This means the FDA’s 1,451 approvals are impressive as a market momentum indicator but offer limited value as a quality benchmark. For hospital buyers evaluating an AI system, FDA approval alone isn’t enough. What matters are prospective studies in comparable patient populations and transparent, reproducible clinical validation within their own care setting.

Europe’s Blind Spot: No Database, No Transparency

Europe has no public equivalent. Neither the European Medicines Agency (EMA) nor Germany’s Federal Institute for Drugs and Medical Devices (BfArM) maintains an accessible database of CE-marked AI medical products. The European Database on Medical Devices (EUDAMED), designed precisely to ensure transparency, has been under development for years. Some modules will become mandatory in May 2026, but a comprehensive overview of all AI-enabled products on the European market won’t be available for several more years.

Independent observers estimate that at least 219 CE-marked AI medical products exist as of January 2025. The true figure could be higher, but no one knows for sure. For hospital buyers tasked with deciding which AI systems to evaluate, this is a real problem.

DACH Hospitals Already Using AI – But in Siloed Solutions

Despite regulatory gaps at the European level, some hospitals in the DACH region are already deploying AI diagnostics in live settings. The results are promising, though they remain isolated.

Asklepios, one of Germany’s largest private hospital operators, has rolled out the Israeli AI system Aidoc across more than 25 facilities. The software analyses CT scans in real time, detecting strokes, intracranial haemorrhages, and spinal fractures—funded through the Hospital Future Act (KHZG). Its key advantage? Aidoc prioritises the reporting queue, ensuring critical cases reach radiologists immediately rather than getting stuck in the backlog.

The largest prospective study to date on AI in diagnostics also hails from Germany. The Vara-PRAIM study, published in *Nature Medicine*, examined over 463,000 screening mammograms. The outcome: a 17.6% increase in breast cancer detection, coupled with a simultaneous reduction in radiologists’ workload. It’s the most extensive prospective study worldwide on this topic.

“The question is no longer whether AI works in diagnostics. The question is how we move it from studies into routine practice—regulatorily, organisationally, and economically.”
— Prof. Alexander Berens, University of Tübingen (press release, 2025)

The issue? Each of these implementations is a standalone project. There’s no national standard for integrating AI into hospital information systems, no uniform interface, and no central registry showing which hospital uses which system. While the Innovation Fund of the Federal Joint Committee (G-BA) supports systematic AI adoption in radiology through the xR.AI project, Germany is still far from widespread rollout.

The result? German hospitals wanting to introduce AI diagnostics must navigate the entire process alone—from product selection to integration with their hospital information system (KIS) and radiologist training. Legal liability also remains unresolved at both national and European levels. If an AI system misses a finding and the treating physician relies on it, who bears responsibility? The Medical Device Regulation (MDR) governs manufacturer liability, but medical due diligence still applies. In practice, this uncertainty leads many radiologists to avoid using AI-generated reports as a decision-making basis, treating them at most as an additional clue.

AI Act from 2027: What MedTech companies will face

The EU’s Artificial Intelligence Regulation (AI Act) came into force in August 2024, but medical products benefit from a transitional period: comprehensive requirements for high-risk systems will only apply from 2 August 2027. AI systems that qualify as medical devices under the MDR or the In Vitro Diagnostic Medical Devices Regulation (IVDR) will automatically be classified as high-risk AI.

What does this mean in practice? Manufacturers will need to implement a dedicated AI risk-management system, meet data-quality and data-governance requirements, prepare technical documentation that goes beyond MDR standards, and design systems that enable human oversight. The effort required is substantial—particularly for smaller companies.

A silver lining: in November 2025, the European Commission proposed a comprehensive simplification aimed at reducing overlap between the AI Act and MDR/IVDR. Instead of duplicate documentation and conformity assessments, an integrated procedure is planned. Approval is expected by summer 2026.

For German MedTech SMEs—93% of the industry consists of companies with fewer than 250 employees—this is no theoretical concern. In recent years, the MDR has already led some manufacturers to withdraw products from the European market because certification costs exceeded revenues. If the AI Act adds similar bureaucracy, industry consolidation will accelerate.

At the same time, regulation creates a competitive edge in quality: medical AI products compliant with both MDR and the AI Act will enjoy a trust advantage over devices approved only by the FDA. European manufacturers that clear this dual hurdle can leverage it as a differentiator in the global market—but first, they must survive the financial and organisational challenges of the transition period until 2027.

Gain speed or miss the opportunity

The global AI healthcare market reached an estimated 21.66 billion US dollars in 2025. It is projected to exceed 110 billion US dollars by 2030—a compound annual growth rate of 38.6% (MarketsandMarkets). PwC Strategy& estimates Europe’s potential at 203 billion US dollars by 2030.

Germany boasts strong players in AI for MedTech, including Siemens Healthineers, Brainomix, Vara, and Merantix. Clinical research is internationally competitive—as the PRAIM study clearly demonstrates. What’s missing is the bridge between research and routine use: a European AI database to ensure transparency; a harmonised approval process that doesn’t stifle SMEs; and a hospital landscape that views AI not as a subsidised project, but as a standard tool.

The FDA has approved 1,451 AI medical products in eight years—and learned to adapt its processes along the way. Europe, meanwhile, has spent the same period drafting regulations whose impact won’t be felt until 2027. Those who fail to recognise this delay as a strategic risk haven’t grasped the market’s dynamics.

Frequently Asked Questions

Image source: Pexels / MART PRODUCTION (px:7089623)