Accelerating the healthy ageing transition with data science

Increased life expectancy, coupled with lower birthrates, is going to put enormous pressure on our healthcare system in the coming decades. By 2030, all of the Baby Boomers, the largest generational demographic, will be aged 65 or older, which means a substantial increase in the older (and retired) adult population. The simple math; more people will be in need of healthcare, and less people will be able to provide it. So how can we make sure healthcare remains accessible and affordable for everyone?

Multi-generational health data 

Healthy aging is a big part of the solution, but also one of the toughest nuts to crack. Because in order to fully understand the complex interactions between genetics, lifestyle, diet and environmental factors, you’ll need high quality data, and lots of it. Fortunately, the Northern Netherlands is also home to Lifelines, one of the biggest biobanks and multi-generational cohort studies in the world, following 167,000 participants across three generations.  

With Lifelines, we collect health data and biosamples from healthy individuals throughout their lives, which we’ve been doing for 20 years now

Director Kaat van de Vyver at Lifelines General 

‘And what makes this project really interesting and unique, is that it’s a different approach to medical or life science. Instead of looking for ways to cure someone who is already sick, we’re enabling researchers to study what's causing sickness in a group of people by identifying markers and biomarkers at the earliest possible stage, and also comparing these markers to social and environmental factors, for example. This is incredibly valuable data for researchers and also policy makers, who can use it to make evidence-based policy decisions or look at the effectiveness of implemented policies.’ 

Early detection 

To date, around 800 papers have been published using Lifelines data. One of the recent breakthroughs in research involves the early detection of bladder cancer, by measuring the amount of glycosaminoglycan in urine samples. This helps identify people at high risk of developing bladder cancer within seven years, even if they appear healthy at the time of the test. Urine analysis also revealed a dietary salt intake that was higher than the recommended norm, allowing the Dutch Ministry of Health to create evidence-based legislation for the recommended amount of salt in processed foods. 

Another recent study revealed the effect of fast-food on Body Mass Index (BMI). ‘Participants, in particular young adults, living in low-income neighborhoods with at least two fast-food outlets within 1 kilometer (0.6 miles) of their residential address, had a higher BMI than their peers with no fast-food outlets within 1 kilometer’, de Vyver explains.   

Focus on prevention   

As the saying goes, prevention is the best medicine. ‘But our healthcare systems are primarily based on curative medicine’, de Vyver says. ‘And most of our business models are also designed around that.  One of the biggest challenges in transitioning towards preventive healthcare is finding the right business models, because it’s a lot easier to make money by proving you’ve cured someone who’s sick, than proving you prevented someone who’s healthy from getting sick.  

That’s where Lifelines can offer real value, because we can provide that proof, which is why we’re also exploring new collaborations not only with universities and knowledge institutions, but also with organizations in areas such as medtech.’, de Vyver continues. ‘Our high-quality data could be especially useful for developing new and more effective drugs by finding early biomarkers or to test the real-world efficacy of existing drugs across three generations.’ 

Molecular modeling 

Even though prevention is the best medicine, the reality, of course, is that people do get sick as they age. And with increased life expectancy, age-related diseases also increase, as does the risk of developing cancer, for example.

In the Northern Netherlands, data is also the foundation for cutting edge predictive models to provide patients with the best possible treatment, which is exactly what Groningen-based medtech startup Protyon does for lung cancer patients.  

Cancer is a very complex disease and targeted drug therapy is often a first-line treatment choice for certain cancers because it targets specific molecules on cancer cells, minimizing damage to healthy cells and potentially leading to fewer side effects compared to traditional chemotherapy

CEO Rositsa Jordanova at Protyon

‘But as much as $16 billion per year is wasted globally on misaligned drug therapies and for around 40% of patients, these drugs aren’t effective. And when they do work, they stop being effective for 30% of patients who initially responded well to these treatments, because of mutations in protein structures. We use 3D molecular modeling software to predict which of these drugs will be most effective for individual patients.’ 

Like playing Tetris 

So how does Protyon’s software work? ‘We work with protein structures, which is different from other big data projects where DNA or genomic data is used. We take it a step further and look at the proteins, because they are the molecules that do all of the work and interact with the drugs used for treatment’, Jordanova explains. ‘With our modeling software, we try to predict how these proteins will mutate over time and how their shape can be matched with the right drug molecule. So in that sense, it’s a little bit like playing Tetris.’ 

‘And in the end, we can come up with a list of drugs that would be most effective for treatment’, Jordanova continues. ‘This helps clinicians to make informed decisions about the best line of treatment and improves quality of life for patients and increases their life expectancy. And it also lowers the burden on the healthcare system, because it saves costs by eliminating ineffective treatments.’

Building on a decade of research 

Protyon was founded three years ago, as a spin-off of the University of Groningen and the University Medical Center Groningen. ‘But we’re building on a decade of research’, Jordanova says. ‘And of course, to be able to create these molecular structures, we do need a lot of data, for which we use the Protein Data Bank and patient databases, but we’re currently also creating our own database. Eventually, our modeling software will also use AI components, such as machine learning and molecular analysis using digital twins of patient molecules.’ 

Our current focus is on lung cancer, but we’re looking to create protein modelling for other types of cancer in the coming years

CEO Rositsa Jordanova at Protyon

Jordanova continues. ‘And our big vision for the future is that our software could be used as an add-on for any type of drug, allowing for super personalized treatment for everything and everyone.’