AI in Healthcare: Breakthrough Progress Meets Dutch Obstacles

Hey everyone, after a small break since the last blog, I thought it would be nice to write another one. This time, I'm tackling AI in healthcare, which I think is one of the most exciting areas out there. We're talking tools that could save lives, cut down on mistakes, and make doctors' jobs easier. But here's the thing: sometimes you have to be a bit critical of a system to push for real change. In the Netherlands, where I live, work, and study, the healthcare sector is in my opinion too slow in adopting AI, and it's frustrating to see such huge potential going untapped. Let's break it down step by step.

First off, the big issue at hand is pretty simple: AI is exploding with breakthroughs in medicine globally, but in the Netherlands, adoption is painfully slow. I've seen this firsthand. During my bachelor's thesis on dementia and AI, I interviewed clinicians who were hesitant about these systems, even though they admitted they didn't know much about them. And from my four years working in elderly care at Zuyderland, I experienced the daily grind of inefficiencies. Like outdated scheduling, piles of paperwork, and routes for caregivers that wasted hours. Back then, I didn't know how to fix it, but now, after studying AI, I see how smart algorithms could optimise all that, freeing up staff to focus on actual patient care. Yet, despite all the hype, Dutch hospitals and care facilities are still mostly in pilot mode or ignoring AI altogether. For example, a 2024 survey showed that only about one-third of radiology centres here have deployed AI tools, even though adoption doubled in two years. It's like we're stuck in neutral while the rest of the world races ahead.

But let's back up and talk about why this matters: AI isn't just "helpful", it's often straight-up better than humans at certain medical tasks. Here is a list of some of the most interesting things I found during the research for this blog post:

- Superior Diagnostic Accuracy: Google's AMIE, an AI optimised for medical reasoning, achieved a final diagnosis accuracy of 35.4% compared to 13.8% for doctors with over a decade of experience. When listing top possible diagnoses, AMIE included the correct one 55.4% of the time, surpassing physicians at 34.6%.

- Optimal Treatment Recommendations: In a virtual clinic trial, AI recommendations were rated "optimal" in 77% of cases, outperforming physicians who scored 67%.

- Excelling in Pathology: AI models analysed over 152,000 biopsy slides with 96% sensitivity and 93% specificity, significantly outpacing the average clinician's performance.

- Skin Cancer Detection: AI achieved 92.5% sensitivity for melanoma detection, far exceeding general doctors at 64.6% and matching expert dermatologists.

- Cardiology Precision: AI detects heart arrhythmias with over 98% accuracy, compared to cardiologists' range of 73-83%.

- Improved Screening Compliance: In ophthalmology, AI increased diabetic eye screening compliance from 22% to 100% in a trial.

- Consistent Outperformance: Meta-analyses from 2023-2025 confirm AI matches or surpasses experts in tasks like reading X-rays, ECGs, and solving complex case studies from the New England Journal of Medicine, where a Microsoft AI solved 8 out of 10 cases compared to doctors' 2 out of 10.

These stats aren't just numbers; they're proof that AI is reshaping healthcare with unmatched accuracy and potential to save lives.

So, if AI is this powerful, why the reluctance in the Netherlands? I completely get why clinicians in the Netherlands are hesitant about AI. It's not just about new tech, it's about trusting something that feels like a mysterious black box, especially when lives are at stake. Picture yourself as a doctor, years deep into your career, used to relying on your hard-earned expertise. Suddenly, there's this system spitting out diagnoses or treatment plans, but no one can clearly explain how it got there. That's unsettling. A recent survey from 2025 echoed this, revealing that 62% of European clinicians worry about AI making mistakes. Patients share that scepticism, wondering if a machine can truly understand their pain. I've seen this firsthand in elderly care, where I worked, clinicians pouring their hearts into patient care, wary of anything that feels like it might distance them from that human connection.

Then there's the fear that AI could reshape their roles entirely. I've talked to older doctors who barely had exposure to AI during their training, and to them, it feels like an intruder. It's not hard to understand why they'd rather stick to the methods they know. A 2025 study of Dutch GPs found that those who understood AI were far more open to it, while others clung to tradition, almost as a reflex. I don't blame them; it's human to feel uneasy when technology threatens to upend your world. On top of that, the practical hurdles are real. Integrating AI into hospitals is a costly, technical beast, with no guaranteed payoff yet. GDPR's strict data privacy rules make it tough to share patient info for training AI, and let's not forget the mess of fragmented data, handwritten notes, and siloed records, which can render these tools useless. In elderly care, I saw how inefficient systems bogged everything down. AI could streamline scheduling or predict ICU needs, but without trust from the team, it just gathers dust. Many pilot programs fizzle out, not because people are lazy, but because budgets dry up or the tools don't fit into daily workflows. It's a tangle of human fears, knowledge gaps, and systemic roadblocks.

And I'm not here to just point fingers at the hesitancy around AI in Dutch healthcare; I genuinely believe we can bridge this gap with solutions that address clinicians' concerns and build trust. As someone who's seen the challenges up close, I understand how daunting it is to trust a system you don't fully grasp, especially when lives are on the line. But I'm optimistic we can turn this around in the Netherlands with the right approach. The key is education, not to drown clinicians in technical jargon, but to help them understand enough about AI to feel confident in its potential. When doctors see how these systems work, without needing to code or dive into algorithms, they can start to trust them. This trust will pave the way for adoption, where AI doesn't replace doctors but reshapes their roles. In the coming years, I see doctors evolving from solely making diagnoses to acting as vital translators between AI systems and patients, explaining results with empathy and clarity. This partnership can boost patient trust, reduce medical errors, and elevate healthcare quality. Not for every aspect of care, of course, but for critical tasks like diagnostics and imaging where AI shines.

We also need to advocate for smarter government involvement. Right now, the Ministerie van Volksgezondheid, Welzijn en Sport (VWS) is moving too slowly, stuck in a cycle of inefficiency and wasted resources. They're commissioning endless consultant reports, like the PwC study on the EU AI Act, when what we need is input from AI specialists who live and breathe this technology. The new AI Act, while important for safety, adds regulatory layers that slow progress, unlike Singapore's clear guidelines or the US's swift FDA approvals. Only 13 of 33 Dutch hospitals had dedicated AI budgets in 2022, and that's not enough. The government could accelerate change by funding shared data platforms, offering incentives for AI adoption, and mandating AI training in medical education. Instead, we see complacency amid staff shortages and rising costs, which is frustrating when Dutch healthcare is world-class. We can't let ourselves fall behind neighbours like Sweden or Germany, who are already using AI for cancer screening.

And here's where I turn to myself and the AI community: we need to step up. Too often, we back off when we meet resistance or hesitancy, but this is a massive shift for healthcare professionals, and we need to be patient. Dutch universities and startups are already making strides, think of the UMC Groningen's AI for drafting patient letters. Projects like the national AI for Imaging (AIFI) pilot are creating shared infrastructure to make adoption easier. By investing in hands-on workshops that show clinicians how AI can spot rare diseases faster or flag tumours humans might miss, we can break the distrust cycle. It's about showing, not just telling, that AI complements human skills, handling data crunching while doctors bring empathy and oversight. With more collaboration, like the Dutch AI Coalition's ELSAlabs, and a commitment to supporting institutions through this transition, we can make AI a trusted partner in Dutch healthcare sooner than we think.

Ultimately, AI in healthcare isn't about replacing doctors; it's about empowering them to perform better. The Netherlands has the talent and the need (think ageing population and burnout), so let's harness it. I'm hopeful we'll get there with more education, investment, and a nudge from the top. But it won't happen without some tough love and faster action. If we keep ignoring the potential, we're missing out on a healthier future for everyone.