Vara, headquartered in Berlin, and Medical Horizons, Italy’s leading distributor of AI-based medical devices, are announcing a partnership that brings Vara’s screening AI to Italian mammography programs at a defining moment for the country.

Italy’s landmark moment

In December 2025, Italy’s National Institute of Health (Istituto Superiore di Sanità) issued updated national guidelines formally recommending AI as a triage tool within organized mammography screening. With this, Italy became the first European country to issue a national-level conditional recommendation in favor of AI-based triage to determine whether mammograms require single or double human reading.

It is a landmark moment for evidence-based AI in radiology, and it shifts the question from whether AI belongs in screening to how well it gets deployed.

What Vara is built for

Vara is designed for mammography, screening and diagnostic. We currently support approximately 50% of Germany’s national breast cancer screening programme, processing around 200,000 cases per month across seven years of continuous deployment. It is the largest AI deployment in any organized screening program worldwide.

The clinical evidence comes from the PRAIM study, published in Nature Medicine in January 2025: 12 centres, 119 radiologists, 463,156 women, routine clinical conditions. +17.6% cancer detection, up to -15% false recalls, -59% radiologist workload reduction. All across major scanner types, with zero exclusion criteria. Optimising for one outcome at the expense of another is the more common finding in the literature; PRAIM showed all three improving at the same time.

Beyond evidence, what Vara brings to every program is seamless IT integration and continuous real-time AI performance monitoring, the operational backbone that lets screening programs adopt AI without compromising on patient safety.

What Medical Horizons adds

Medical Horizons has over 30 years of experience working across Italian healthcare institutions, including Campus Bio-Medico University Hospital in Rome. The team has deep relationships with the regional health authorities and screening organizations that will be moving the new national guidelines from recommendation to clinical practice.

For Italian programs that want to adopt Vara, Medical Horizons provides the local presence, regulatory familiarity, and clinical network needed to deploy AI responsibly at scale.

Why this matters

Italy’s screening programs are now in the early window between guideline and implementation. The decisions made in this period (which AI, which deployment model, which evidence base) will shape how AI is used in Italian breast cancer screening for years to come.

The case for AI in screening rests on real-world evidence at scale, not pilot studies or marketing claims. Italy has set a high bar by anchoring its recommendation in clinical evidence. The partnership between Vara and Medical Horizons is built to meet that bar: bringing the largest body of prospective real-world evidence in mammography AI into Italian screening through a partner who knows the system from the inside.

Italy has set the European standard. We are honored to help answer the question of how well AI gets deployed alongside Medical Horizons and Italy’s screening community.

Vara, headquartered in Berlin, and deepc, based in Munich, are two of Germany’s most recognised names in AI radiology. Today, we are announcing a partnership that makes Vara available to more screening programmes and diagnostic centres across Europe through the deepcOS® platform.

What Vara is built for

Vara is designed for mammography, screening and diagnostic. We currently support approximately 50% of Germany’s national screening programme, processing around 200,000 cases per month across seven years of continuous deployment.

The clinical evidence comes from the PRAIM study, published in Nature Medicine in January 2025: 12 centres, 460,000 women, routine clinical conditions. +17.6% cancer detection, -43% reading time, -56.7% radiologist workload reduction, all three improving at the same time. Optimising for one at the expense of another is the more common finding in the literature. Since publication, continued development has exceeded even those benchmarks.

What sets Vara apart in that evidence base is how it performs with prior imaging. An independent evaluation across 10,348 screening participants in Sweden compared commercially available AI vendors on positive predictive value. Without priors, Vara scored 9.3%. With one prior, 11.4%. With three priors, 18.0%—above the radiologist benchmark of 10.1%, and a result no other vendor in the evaluation matched or supported. The ability to use multiple priors is a technical capability most competitors do not offer.

What deepc adds

deepcOS® is vendor-neutral AI deployment infrastructure operating across 30+ countries. It handles the integration complexity that makes scaling AI across heterogeneous clinical and IT environments difficult: different PACS systems, different workflows, different regulatory contexts.

For programmes and providers that want to adopt Vara, deepcOS® provides an additional path into existing clinical environments with no change to how radiologists work.

Why this matters

For most programmes, the remaining question is practical: how do we access this, and how do we integrate it without a multi-year IT project. The deepcOS® partnership extends Vara’s reach to providers across Europe who want evidence-based AI for mammography without the integration overhead.

Europe’s screening programmes face growing case volumes, radiologist capacity under pressure, and the need to improve detection without increasing recall rates. The evidence that AI addresses all three exists. Friction in access and integration has slowed adoption. The deepcOS® partnership addresses that directly, extending Vara’s reach to providers across Europe who want evidence-based mammography AI through infrastructure they may already use.

Until now, there has been no AI on the market authrised to do what a second radiologist does — read a mammogram independently.

That changed on 15 October 2025. Vara received a new CE certificate for its breast-imaging AI. It is the only AI on the market authorised to operate as an independent second reader, for both screening and diagnostic use. It marks Vara’s entry into the European market.

Why independent reading matters

Most AI in breast screening operates as decision support — a tool that highlights findings for a radiologist to review. An independent second read is fundamentally different. The AI reads the mammogram on its own, the same way a second radiologist would.

This matters because maintaining a human second read is often not economically feasible, particularly in diagnostic workflows. Many programmes simply can’t staff it. An AI that is authorised and clinically validated for that role fills a gap that has been widening for years.

The evidence behind it

This CE mark didn’t happen in a vacuum.

Earlier this year, Nature Medicine published the PRAIM study — initiated by Vara and conducted across 12 screening centres in Germany. It enrolled 460,000 women with no exclusion criteria, making it the largest prospective, real-world study of AI in healthcare.

The results: statistically significant superiority in cancer detection rate and meaningful workload reduction for radiologists.

PRAIM is what made approval for independent second reading possible. Not a retrospective analysis. Not a single-site pilot. A population-wide, prospective study across a national programme.

What regulators recognised beyond performance

In approving Vara’s AI for independent second reading, regulators also recognised something else: Vara’s autonomous, prospective AI monitoring system.

This is the real-time quality assurance infrastructure we’ve written about before — continuous, case-level monitoring of every AI prediction, connected to clinical outcomes. Regulators saw it as setting a new standard for responsible deployment.

Evidence gets you approved. Monitoring is what makes deployment safe.

What’s available, and where

The CE-marked product is available in Europe starting today, as an independent second read or decision support system — including analysis of mammograms from prior screening rounds.

Providers can integrate Vara directly in their PACS viewer, or via the Vara platform. That platform currently serves 40% of the German national screening programme, processing approximately 150,000 mammograms monthly and more than 1.5 million annually.

What this means going forward

Stefan Bunk, co-founder and CTO of Vara:

“Our ambition has always been to pioneer data-driven AI at a nationwide scale. The industry has long moved beyond retrospective evidence. Now, by pairing population-wide, real-world results with automatic real-time monitoring, we’re offering a credible, scalable path for AI in healthcare.”

Europe’s screening capacity challenge isn’t theoretical. It’s here. The question is whether AI can meet the clinical, regulatory, and operational bar required to help.

With the largest prospective evidence base in healthcare AI, the only CE mark for independent second reading, and real-time monitoring built into every deployment — Vara’s answer is yes.

Vara builds AI and cloud workflow tools for breast imaging, helping radiologists detect cancer earlier while reducing unnecessary recalls and reading time

For years, the case for AI in radiology rested on retrospective data and small, controlled settings. That changed on 7 January 2025, when Nature Medicine published the PRAIM study — the first nationwide deployment of AI in Germany’s National Screening Program.

PRAIM enrolled 500,000 participants across Germany. It was conducted by Vara and the University of Lübeck. It is the largest prospective, real-world study of AI in healthcare to date.

What PRAIM measured

The study integrated AI into live screening workflows and tracked what happened to two things: cancer detection and radiologist workload.

The findings:

• AI integration improved cancer detection rates by 17.6%, without increasing false positives

• Radiologists spent 43% less time interpreting AI-tagged normal examinations

• In a simulated scenario where radiologists did not interpret AI-labelled normal mammograms — 56.7% of all scans — cancer detection rate still improved by 16.7%, while recalls fell by 15%

These are not lab results. They come from a live national screening programme, across multiple centres, with no exclusion criteria.

Why this study is different

Previous AI studies in radiology have typically relied on retrospective datasets or controlled research environments. PRAIM was prospective and deployed in routine clinical practice — the same conditions under which AI would need to perform if adopted at scale.

That distinction matters. Retrospective studies can show that AI performs well on historical data. Prospective studies show whether it performs well when radiologists actually use it, in real workflows, on real patients.

Professor Alexander Katalinic, Head of the Institute of Social Medicine and Epidemiology at the University of Lübeck and lead investigator:

“The results are extremely positive and have exceeded our expectations. We can now demonstrate that AI significantly improves the cancer detection rate in screening for breast cancer.”

What this means in practice

Germany screens millions of women annually. If PRAIM’s results were applied across the full programme, the implications are straightforward: more cancers caught earlier, fewer unnecessary recalls, and a meaningful reduction in the reading workload that screening programmes are struggling to staff.

The 56.7% workload finding is particularly relevant. More than half of all scans were AI-labelled normal — and even without a radiologist reviewing those, detection rates still improved. For programmes facing chronic radiologist shortages, that is not a theoretical efficiency gain. It is a practical path forward.

Context

Vara’s AI is integrated into 40% of Germany’s national screening centres, processing over 80,000 mammograms monthly. The company has published in Nature Medicine and Lancet Digital Health, and holds what is now the largest prospective evidence base for AI in any radiology specialty.

Stefan Bunk, CTO of Vara:

“The PRAIM study is a testament to the vast potential of AI in improving breast cancer screening. We are confident that the results will elevate the discussion about AI-assisted mammography screening, paving the way for wider adoption in health systems globally.”

PRAIM does not close the conversation about AI in screening. But it shifts it from “does AI work?” to “how do we deploy it responsibly?” — a question that requires a different kind of evidence, and a different kind of infrastructure.