Artificial intelligence (AI) isn’t just for tech giants or chatbots- it can save lives. Jeroen de Ridder, Professor of Bioinformatics in Molecular Medicine, demonstrates this through his research. He explores how intelligent algorithms can detect patterns in large volumes of molecular data. The goal: to recognize diseases such as cancer earlier and treat them more effectively.
I have always been curious about the inner workings of things and their potential applications” says Jeroen. As a child, he preferred taking toys apart rather than playing with them. That curiosity led him to study Electrical Engineering at TU Delft, where he became fascinated by technology and mathematics.
During a course on Pattern Recognition everything fell into place. From that moment on, he knew what direction he wanted to take and began his scientific journey into machine learning: teaching computers to draw conclusions from large datasets.
“Why is something that comes naturally to a child – such as recognizing an elephant – so difficult for a computer? My four-year-old daughter sees a trunk and immediately shouts: ‘Elephant!’ She doesn’t need to take measurements or compare surfaces. She just sees it,” Jeroen explains.
“But a computer doesn’t ‘recognize’ things that way. It has to calculate. It must infer patterns from numbers. And then, that seemingly simple difference between an elephant’s trunk and a camel’s hump suddenly isn’t so simple.”
On Wednesday, May 14, Jeroen gave his inaugural lecture as Professor of Bioinformatics in Molecular Medicine, centered around a key question: how can we use AI to better recognize and treat the ‘molecular fingerprints’ of disease?
AI has become something of a buzzword these days, applied in countless ways. So why did Jeroen- with his technical background – choose to focus on medicine? “Biological processes are incredibly complex. Yet, with data analysis, you can gain surprising insights without ever setting foot in a lab. During my PhD research, from 2005 to 2010, I was amazed at how patterns could be discovered in biomedical data,” he says.
Jeroen conducted his PhD work at a time when biomedical pattern recognition was rapidly advancing. In the late 1990s, the ‘microarray’ was introduced: a laboratory instrument, a specialized kind of chip, that allows researchers to measure the activity of thousands of genes simultaneously. For the first time, scientists could map tumors on a large scale – not by zooming in on them under a microscope, but by creating so-called ‘gene expression profiles’: identifying which genes are switched on or off in a tumor cell, and which ones are releasing the messenger molecules that a cancer cell needs to produce essential proteins.
The result was a collection of massive datasets containing genetic information from hundreds of patients. Thanks to this, scientists quickly discovered that leukemia wasn’t just one disease, but a collection of distinct subtypes—each with its own molecular characteristics and behavior, or in other words: its own ‘molecular fingerprint’.
Since then, technology has evolved at lightning speed. Researchers now use datasets and pattern recognition not only to understand cancer but also to better predict its prognosis. New measurement techniques, such as next-generation sequencing, make it possible to analyze DNA in intricate detail.
At the same time, there has been an explosion of data on how the smallest components of the human body behave: from gene activity to DNA structure, from protein binding to chemical reactions between molecules. “This data goldmine requires even smarter analytical methods. And that’s exactly where AI comes in,” says Jeroen.
In 2016, Jeroen moved to the University Medical Center Utrecht (UMC Utrecht) after leading his own research group at TU Delft. “I was excited by the idea of working closer to the ‘wet lab’, in a hospital setting. What especially appealed to me was the opportunity to contribute more directly to clinical practice – and to make a real impact on healthcare.”
Jeroen now leads a bioinformatics lab at UMC Utrecht, where researchers use data science to tackle medical challenges.
This data goldmine requires even smarter analytical methods. And that’s exactly where AI comes in.
Sturgeon is one of the most compelling and recent examples of how Jeroen and his team work. This algorithm was developed in collaboration with the Princess Máxima Center and Amsterdam UMC.
Thanks to Sturgeon, the type of brain tumor can now be identified during surgery, based on DNA pattern recognition. Normally, this process takes at least a week—and the patient might even require a second surgery. But with Sturgeon, the neurosurgeon can determine how much of the tumor needs to be removed during the operation, preventing the need for another invasive procedure.
The technology was developed, tested, and implemented at record speed. Today, Sturgeon is routinely used during brain surgeries at the Princess Máxima Center, and it has already attracted international attention.
AI typically requires large datasets to perform well. But what if you’re working with a rare disease that affects only 50 patients? “That’s a big challenge,” says De Ridder. His research group’s solution is to first train models on general molecular knowledge, then fine-tune them for a specific problem. This approach—known as self-supervised pre-training—is at the core of the research project FoundationDX, funded by the European Research Council.
Within FoundationDX, AI models first learn how a cell is structured on a molecular level using ‘healthy’ data. Only after that do they begin analyzing ‘diseased’ data—the molecular profiles of tumors from a specific patient group. “This way, you can still make reliable predictions on the basis of limited data. And that’s essential for rare tumors, where millions of examples simply don’t exist.”
Jeroen emphasizes that collaboration is essential to his work. “Bioinformatics is by definition a cross-disciplinary field. You have to speak the language of biologists, clinicians, and programmers alike. Getting from idea to implementation in just two years, like we did with Sturgeon, is remarkable. That only works when everyone -from pathologist to surgeon and from data scientist to lab technician -works together toward the same goal: providing the best possible care.”
According to Jeroen, the Utrecht campus is the perfect place for this kind of collaboration. “UMC Utrecht, Utrecht University, the Hubrecht Institute, and the Princess Máxima Center form a unique ecosystem. Here, you can go from a brainstorming session with a surgeon straight into an AI meeting with PhD students. That’s where the best ideas arise – often over a cup of coffee.”
Jeroen is still driven by curiosity and wants to inspire the next generation of researchers to take a similar approach. That’s why, as a professor, he invests heavily in education.
Jeroen: “We’re training the next generation – people who not only know how to apply AI, but also understand what it means for healthcare. That comes with a great responsibility. Stay curious. Keep asking questions. And keep collaborating. Because real innovation happens at the intersection of disciplines.”
“The desire to make a difference. We conduct fundamental research, but always with the aim of solving real-world problems. Whether it’s about better diagnostics, personalized treatments, or reducing healthcare costs – if data can help you make the right decisions, you can truly have an impact.”
At the Utrecht Science Park, several institutions are working passionately to better understand and treat cancer. Together, they form the collaborative network called Utrecht Cancer.
More than 1,200 researchers from UMC Utrecht, the Faculties of Veterinary Medicine and Science at Utrecht University, the Hubrecht Institute, and the Princess Máxima Center are combining their knowledge, innovation, and experience to improve treatment, quality of life, and survival rates for adults, children, and even animals with cancer. Utrecht Cancer also actively fosters collaboration with other public and private partners.
