Universiteit Leiden

nl en

The rocking researcher: Marjolein Fokkema connects disciplines with algorithms and pop songs

From predicting Alzheimer’s to the growth of organisms: psychologist Marjolein Fokkema’s algorithms can be used in many disciplines. They also provide inspiration for her songs, theatre shows and life lessons. What drives this rocking researcher?

 ‘I want the best of both worlds: complex data and comprehensible results’

An upbeat guitar riff, tight drums and a singer urging us to, ‘Loosen up that model!’ To her students’ surprise, Leiden psychologist Fokkema sometimes starts her lectures with a catchy video clip. Mary-Jo who sings of the beauty and wisdom of algorithms is Fokkema herself – as too is the bearded guitarist. In her somewhat less rock & roll office at the faculty, she explains why this is how she sometimes starts her lectures, ‘I try to get my students excited about algorithms. It wakes them up and hopefully gets them raring to delve into the complex material. They usually respond really positively.’

Mary-Jo & the Support Vector Machines

Due to the selected cookie settings, we cannot show this video here.

Watch the video on the original website or

Fokkema designs machine learning algorithms: smart statistical models that, acting on a series of instructions, speedily detect patterns from data and can make predictions. ‘Psychologists should be able to use them in decisions and researchers to test their theories.’ For example, her models produce decision trees that show expected outcomes in graphs and offer practitioners different scenarios. ‘My flexible algorithms take account of developments over time and weigh up changeable features. I want the best of both worlds: complex data and comprehensible results.’

Predicting Alzheimer’s

Researchers from various disciplines approach Fokkema for this sought-after specialism. In Leiden, for example, she and PhD candidate Wouter van Loon are developing a prediction model that uses data from different sources, such as brain scans, questionnaires and genetic data to predict the likelihood of Alzheimer’s. The research is funded by the Society Artificial Intelligence and Life Sciences (SAILS) interdisciplinary programme.

Growing up

The equally interdisciplinary Growing Up Together in Society (GUTS) project focuses on how this generation of young people are growing up in our complex society. And how they can become engaged and resilient citizens. Psychologists are working with, among others, psychiatrists, neuroscientists and sociologists in the programme.

And with PhD candidate Zino Brystowski, Fokkema is developing a model in which various theories ‘compete’ with each other, says Fokkema, ‘The different disciplines all have their own theories and data sources. We are using our methods to explore which of these provide the best predictions and how they complement each other.’

Poster of Fokkema’s theatre show
Poster of Fokkema’s theatre show

Why young people choose STEM subjects

Other universities are also keen to call on Fokkema’s expertise. Educational psychologists from Dortmund University are researching why pupils choose STEM subjects and the role of gender differences in this. ‘The researchers are incorporating all sorts of data into my model that produce a kind of “if-then” rules and explain why pupils do or don’t choose STEM subjects.’

These insights can help teachers give pupils extra attention at crucial moments when they are likely to turn their backs on STEM. Talking of STEM subjects, Fokkema is developing models with biologists in Leiden to map how microplastics influence the growth of organisms.

Will AI replace practitioners?

Fokkema is also working with psychologists from Radboud University who are researching treatments for disorders. ‘We’re developing a model that takes account of factors such as genetic predisposition, the nature and duration of symptoms and which are most severe. This results in a decision tree with options for suitable treatments.’ Doesn’t that put AI in the practitioner’s seat?

‘The aim is that practitioners can use the results, not that they will replace practitioners’, says Fokkema. ‘We’re exploring methods that may contribute to more personalised care in the future. Decision trees like this help you navigate through a forest of information. Obviously, it’s up to the practitioner to assess the patient properly and carefully weigh up all options. I do understand the concerns about the power of AI. I’m trying to do justice to the complex reality by providing interpretable results.’

Humans versus robots theatre show

Talking of the friction between humans and machines, Fokkema, who also did drama training, will be appearing in theatres with the show Ik blijf wel mens (I’ll still be human). ‘The story is that I’m going to be transformed into a robot and am taking stock before the operation. What is a human and what is a robot? Perhaps we do more justice to human complexity with AI because we systematically weigh up more factors than a practitioner can alone. But the algorithms do have to be smart and flexible.’

Hence her anthem We gotta loosen up that model! ‘Songs like this give me poetic licence. Algorithms have a beauty in themselves but I sometimes stumble upon life lessons or parallels with everyday life. Such as: Be flexible and learn to deal with uncertainty!

Theatre show Ik blijf wel mens.

Text: Linda van Putten

 

This website uses cookies.  More information.