At the second edition of the Healthtech Adria Conference, held 2–3 October 2025 at the University of Split School of Medicine, AI4Health.Cro joined a packed room of clinicians, founders, innovators, and ecosystem builders to talk about one thing that matters: moving from promising AI demos to tools that survive real clinical workflows.
Co-organised by NetHub and the University of Split School of Medicine, the two-day event gathered 300+ participants and 30 speakers from Croatia and abroad, with a strong focus on digital health, medical technologies, and biopharma—where adoption depends on evidence, integration, and trust.
“AI shouldn’t be used everywhere”
During a roundtable on AI in healthcare, Dr. Anja Barešić, AI4Health.Cro project coordinator, put the core principle on the table:
“AI shouldn’t be used everywhere. Every innovation must prove it’s better than humans for a clearly defined task.”
Dr. Barešić shared the stage with Jayden K., Ivan Biliškov, PhD, and Jovan Stevovic, pushing the discussion away from hype and toward the practical choices that determine whether a product is safe, useful, and adoptable.
Founder-grade takeaways from the roundtable
What came through clearly (and repeatedly) is that “great model” is not the finish line:
-
Start from clinical needs, not feature ideas. The gap between what startups want to build and what clinicians actually need is still one of the biggest failure modes.
-
Don’t do AI “because we can.” Novelty doesn’t translate to adoption unless it solves a specific, owned problem.
-
Educate clinicians to unlock better questions. When doctors understand what AI is (and isn’t), they demand evidence and benchmarks—and they’re right to.
-
Translate metrics into clinical logic. F1 and ROC don’t map to care decisions by default; clinicians need strengths/limitations, failure modes, and “for which patients.”
-
Own your comparators. Define the right benchmark, report clearly, and let medical experts judge fitness-for-use.
AI4Health.Cro: building the path from idea to impact
Anja also shared what AI4Health.Cro is seeing “on the ground”:
-
“In the last 2.5 years, we’ve supported 30+ users on a personalised digitalisation path from idea to market.”
-
Seven AI/health-tech solutions are currently being tested in Croatian hospitals and other healthcare facilities, with structured feedback from clinicians to ensure products fit real workflows.
This is exactly where AI4Health.Cro focuses: validation, testing, and implementation readiness—not just model performance.
Where AI4Health.Cro helps teams move faster (and safer)
AI4Health.Cro’s Test-before-Invest approach is designed for early validation in real environments, so teams can learn early, de-risk development, and avoid expensive “wrong assumptions.”. The hub also supports companies and public entities with services spanning testing, ecosystem connections, and go-to-market readiness in healthcare.