As the CCO of an AI company, I should probably tell you that hospitals are embracing AI in droves. The reality is much more practical.
Hospitals want fewer complications, less pressure on their staff and better oversight of care quality. AI becomes interesting when it can demonstrably contribute to those goals.
“We don’t need another AI pilot.”
Over the past year, I have heard this in several hospitals, from executives to surgeons. And honestly, I completely understand. Behind that response is a clear request: help us solve a concrete problem and show us that it can be done safely. For me, that is where it begins.
No one loses sleep over an algorithm
When I ask hospital executives what is on their minds, I hear about staff shortages, full wards and operating schedules under pressure. About the challenge of delivering appropriate care under the Dutch Integrated Care Agreement (IZA), and how to maintain the same quality with fewer people. And I hear about complications. Take an infection after surgery. For the patient, it means a longer recovery, additional tests and sometimes readmission. For the hospital, it means a bed occupied for longer, additional care and even more pressure on an already stretched schedule.
That is where the need lies. As a supplier, you need to understand how your solution fits into that picture. Start talking about your model, your dataset or your AUC, and you have lost the conversation before it has even begun.
“The word AI makes the conversation harder, not easier”
This may seem counterintuitive. You might expect “AI” to open doors. In my experience, it first raises questions. Can I trust this? Who is liable if the model gets it wrong? What does the AI Act mean for us? And how much extra work will this create for my staff? These are fair questions, and I am glad hospitals ask them. They show that the AI label alone sells nothing. It is a promise you have to deliver on, and healthcare rightly sets a high bar.
Five perspectives on one decision
A hospital rarely leaves this decision to one person. There are often five groups around the table, each with its own perspective:
Hospital leadership: Does this fit our strategy, and what value does it bring to patients and the organisation?
Clinicians: Do I trust this result, and how do I act on it at the bedside?
IT: Does it work within our EHR, without an extra screen, another login or additional maintenance?
Procurement and legal: Is it certified, who carries which risks, and what about the GDPR and the AI Act?
Infection prevention and quality teams: Does it help us keep track of outcomes?
Together, these answers determine whether a solution earns a place in the hospital. Clinical value, technical integration and organisational feasibility all need to come together.
That is why I also see my role as CCO as that of a translator. I connect the questions of hospital leadership and frontline staff with what the technology can offer. That requires careful listening, clear expectations and an understanding of what a department needs.
What gives hospitals the confidence to move forward
Certification is the baseline. CE marking under the MDR tells a hospital that a product has demonstrated safety and performance for its intended use. Without that foundation, there is no conversation.
It has to work where care happens. A tool outside the EHR gets forgotten. Information within the EHR, available when a clinician makes a decision, gets used.
Hospitals look at each other. The first hospitals to take the step deserve considerable respect. They make it easier for the next hospital to say yes, because there is someone they can call.
People make the difference. Internal clinical champions, clear agreements on how to act on a result, and support that continues beyond go-live.
And yes, those concerns are valid. Pilot fatigue exists. So does alert fatigue. A solution should ease the pressure on clinical teams, not add to it.
“The best AI is the AI you do not notice.”
That means fitting into existing workflows, without giving clinicians another system to check or another stream of alerts to manage. I believe AI will truly become standard care when no one calls it an AI project anymore. When it becomes part of how a department works, just like a lab result or a vital sign: information that helps identify which patient needs attention.
That is the standard we set for ourselves with PERISCOPE®. It predicts the risk of postoperative bacterial infections, from surgical site and urinary tract infections to pneumonia and bloodstream infections, using data already available in the EHR. No additional data entry, no separate system.
So, why does a hospital choose AI?
Honestly? It doesn’t. A hospital chooses better care, less pressure on its staff and confidence that a solution is safe. AI earns its place only when it delivers on all three. Exactly as it should be.
Written by:
