I would like to extend a well-deserved thank you to our first customers, OLVG and Maasstad hospitals, for their invaluable role in advancing AI in critical care.
In Dutch healthcare, it's tempting for hospitals to launch AI development projects on their own. These initiatives are often backed by resources and receive attention through publications and PR. However, these projects rarely move beyond the pilot phase—if they even reach that stage. Erasmus MC has shown that only 2% of algorithms developed (and published) for the ICU are ever validated in practice, and structural implementation remains an even rarer feat. Scaling AI solutions across multiple hospitals is particularly rare. Developing and testing an initial version of an AI solution is one hundred times less work than the task of scaling it.
That’s why we are so incredibly grateful to OLVG and Maasstad for boldly embracing Pacmed Critical as the first "not invented here" AI solution in critical care. And not just any solution—this is one that supports life-and-death decisions for the most critically ill patients in the most complex environment: the ICU. Here, intensivists, as "superspecialists," work with vast amounts of medical data, and our AI integrates in real-time, processing thousands of data points from the PDMS.
Being the first to implement any software means dealing with the product's earliest (and often roughest) version, along with the most challenging implementation efforts. This is especially true for AI software, which continuously improves based on usage and scaling.
Thanks to OLVG and Maasstad, our software can now be deployed across the Netherlands to support ICUs in treating more patients with reduced pressure on staff. Over the past two years, we've made remarkable strides:
- We’ve reduced implementation time and complexity by fourfold.
- Extracting historical data (in a privacy-compliant way) is now much less complex for hospitals.
- We can now integrate with all EHRs and PDMS in the Netherlands through FHIR, making integration simpler and more reliable, with the potential for wider application.
- Calibration for aligning our algorithms with hospital data models and policies now takes 50% less time.
- Our product has seen major advancements:
- Increased predictive power of algorithms, with better handling of input parameters.
- The ability to write back into EHR/PDMS systems.
- Integration with nursing workflows (now CE certified for use by nurses).
- Improved explainability of our algorithms, aligning with clinical reasoning at the organ system level.
- Refined data monitoring, ensuring predictions are only made when data is complete and reliable.
The most significant milestone: Pacmed Critical is no longer a point solution. We have expanded our software with a second module and are actively working on further expansions.
We’re excited to soon share with Santeon and others the lessons we’ve learned in scaling and co-creating AI for better clinical decision-making and making healthcare more accessible.
A heartfelt thank you once again to OLVG and Maasstad. Together, we've overcome major challenges as pioneers. Hospitals driving scaling and creating innovation at the bedside deserve all the praise and attention.
I'd also like to express immense gratitude to:
- Amsterdam UMC: for their substantial contribution to the development, deployment, and validation of our AI decision support, and for having the foresight to start this journey in 2016.
- LUMC, Radboud, and ETZ: for their roles in research and validation.
- St. Antonius: for their contributions to identifying steps for responsible AI scaling and playing a key role in product expansion.
- Catharina and MST: who are currently implementing Pacmed Critical, with a significant focus on research and scalability to ensure AI can be deployed across the Netherlands.
- Santeon: for their collaboration in overcoming obstacles for scalable, reusable AI deployment, and their support in making FHIR integrations possible.
- CZ, Zilveren Kruis, and VGZ: for their trust, support, and contributions, without which this collaboration wouldn’t have been possible.
We look forward to what’s next—together with Santeon and the rest of the healthcare community. This is a critical phase for AI in healthcare. Technological advances are rapidly accelerating, and the pressure on healthcare professionals and hospitals is increasing. AI is receiving a lot of attention, but it also brings hype and unrealistic expectations. A healthcare transformation is needed, but technology alone won’t drive it, even one as powerful, promising, and continuously improving as AI. What AI can do, however, is accelerate and enable this transformation—if done safely and responsibly, in line with the values of our unique and complex healthcare system. And, most importantly, with healthcare leading the way.
We believe this is only possible through collaboration—and we’ve seen that in practice. The first scaling of our clinical AI decision support for future-proof hospital care is a perfect example.
Thank you again to all who have made this possible.
Wouter Kroese
Co-founder & CEO, Pacmed
