Lisa Marie Rolli: Hotz Medalist on AI, Bioinformatics & Her PhD

Lisa Marie Rolli’s academic journey is one for the storybooks.

What began as a spark of curiosity for chemistry and programming has led her to study Bioinformatics at Saarland University, where she started working as a research assistant in her early semesters. Throughout her Bachelor’s and Master’s education, she focused on reliability and interpretability in AI.

Now a PhD student at Saarland Informatics Campus, Lisa Marie was recently awarded the Günter Hotz Medal for her Master’s thesis on trustworthy machine learning. In this interview, she reflects on her journey from Bachelor’s to PhD, her award-winning thesis, the challenges she overcame, and why trustworthiness in AI matters more than ever.

You’ve completed your bachelor’s, master’s and are currently doing your PhD in Bioinformatics. What made you want to study Bioinformatics, and how has studying Bioinformatics at #SIC influenced your career journey?

“In the last two years of high-school, I participated in a program, which is called “Juniorstudium”. It allowed me to take some university courses before officially having my A-levels.  I didn’t know what I wanted to study, but I thought I could go into a technical field or the natural sciences. My favorite subject in school was chemistry, so I started with a chemistry lecture – which was really nice!  At the same time, a friend of mine had started studying Computer Science, and he convinced me to join the Programming 1 course. It was challenging, but I enjoyed it! It was very interesting to me.

I still liked chemistry, though, so Bioinformatics was the best of both worlds for me. And I liked it so much, I continued with the Master’s program.

Studying at SIC was really nice because of the environment. Everyone helps each other, there’s a strong collaboration culture. I had really good thesis supervisors for my Bachelor’s thesis, who also hired me as a research assistant afterward. They wanted me to publish papers with them about our work, and they were enthusiastic about me doing my Master’s thesis with them. It was a very welcoming, supportive environment, so, I started working here, and now I’m in the PhD program.

At the same time, I would say the program was challenging in a good way. I had to put in a lot of effort, but I liked it because I learned a lot. It was intellectually stimulating.” 

You’ve been recently awarded the Günter Hotz Medal by the FDSI for your master’s thesis, “Increasing Trust in Machine Learning-based Drug Sensitivity Prediction.” What made you pick your research topic and can you tell us more about your thesis?

“During my Bachelor’s, I started working at the Chair of Professor Lenhof. I did my thesis on anti-cancer drug sensitivity prediction, focusing on creating reliable predictions providing user-specified certainty guarantees of prediction correctness.

The goal of anti-cancer drug sensitivity prediction is to develop machine learning models that can predict the efficacy of a drug treatment. These predictions could finally be used in a treatment recommendation system.. Obviously, if you want to use machine learning in medical applications, you can’t trust any model; you have to make sure that the model complies with the highest degree of trustworthiness. For example, the model should fulfill specific certainty guarantees and the model should be understandable to humans, specifically for a medical doctor, so they can see how the efficacy predictions are made.

During my time as research assistant, we focused on the trustworthiness in terms of interpretability and reliability of ML models for anti-cancer drug sensitivity prediction. We noticed that these concepts are largely understudied in the field. 

At the moment, the research in the field mostly focuses on enhancing the prediction performance and not studying how trustworthy the model actually is. 

Since we saw that there was a lack of interpretability research, in my Master’s thesis, we designed a novel approach, which is way more interpretable than the other approaches in the field. We also compared its performance to state-of-the-art deep neural networks and it outperformed them. 

For me, personalized medicine is really interesting, and doing research in this field feels super cool to me! I think trustworthiness in personalized medicine is extremely important. Nothing can be applied without trust. It’s one of the most important aspects of this research area, so that’s why I wanted to contribute to this.”

What were some challenges you’ve faced during your research, and how did you overcome them during your thesis?

“When I started my thesis, I had a different topic. It was about predicting compound synergies, i.e., predicting if the drugs have a positive, antagonistic, or neutral effect on each other. I was very enthusiastic about it. My supervisors asked me to check model assumptions before actually starting, which made sense. I had to check if there was a positive correlation between the similarity of the drugs and the response. That was not the case — the results came out extremely scattered. The plots weren’t displaying any correlation at all, not even a negative one. We tried to adapt the model without changing the thesis topic, but there was a point where none of my ideas worked out. Finally, my supervisors and I did not see another option than changing my subject.  They helped me brainstorm, and eventually, I arrived at my final subject. 

You might have an idea and think it’s really nice, until you try it and it doesn’t work out. There was this other time when we had an idea, but it made the runtime go over the roof. Several hours for prediction is not very feasible, is it? You have to discard ideas to get it to work faster — or for it to work at all. Sometimes you spend days just trying to get something to work. But sometimes, if it constantly fails, you need to accept it and move on to the next idea. 

You have to be brave enough to say, ‘Okay, this isn’t working, I need to start all over again.’

Of course, you always struggle with different things, it’s not only that. Our hard disk crashed at some point during my thesis, and for some reason, our backup died. Luckily, the University’s technical support team recovered an old version I could work with. 

Thankfully, I had a lot of help and support from my supervisor. I had a lot of support from my family and boyfriend as well, they would help get me up again when I felt low. All in all, I think you need that kind of mental support during this process. You just have to keep pushing through and hope it works out.”

What are you currently doing?

“Right now, I’m doing a PhD at the Chair of Professor Volkamer funded by an EU (European Union) project called RADAR. Similar to what I did before, I’m focusing on trustworthy machine learning. The goal of RADAR is to develop new plastic alternatives through sustainable roots and using biomass instead of fossil fuels to create compounds. 

I’m in the Hazard Assessment Team, where I do toxicity predictions for different toxic endpoints in human cells, the environment, and animals. I was especially interested in the project because of its focus on chemistry, which I didn’t get to explore very often during my studies. That really piqued my interest. Plus, I think it’s a really important project, and it would be amazing if it succeeds! I’m happy to be contributing to it.” 

Why did you choose to study at #SIC? What sets #SIC apart for you in terms of research opportunities, as well as academic and professional development?

“I first chose to study here because it was the closest university to where I lived, and I had already heard good things about the Computer Science department. I started studying here after my Juniorstudium and really liked it! Later in my studies, I did some Erasmus exchanges and realized that this university is really great.

Personally, I really like the teaching style here. There are also so many research opportunities — you can get involved really early, even during your Bachelor’s. I started working as a research assistant in my third semester at the Chair of Software Engineering. After my bachelor thesis, I continued as a research assistant at the Chair of Bioinformatics, and I’m still in the same office!

I feel like the professors here genuinely want you to join their research team. They try to support your career goals and provide you a huge variety of opportunities to help you grow professionally. They’re also extremely helpful while you write your thesis, and if you contribute to a project, you’re always credited on the paper. They really push you toward excellence.”  

What advice would you give to young women who want to go into the field of Bioinformatics and to prospective #SIC students?

“In the beginning, the transition from school to university is a big leap. It’s super hard to adapt – it’s a challenge. Go into it knowing that, but also know it will get better. Having your friends and family makes a huge difference too. Talking to people really helps… Try to form study groups, collaborate with others, and don’t go through everything alone. Everyone struggles, but it’s better to struggle in a group rather than alone. 

The first lectures can be difficult, but once you get the hang of them, you manage to get through the rest. I also recommend that everyone take full advantage of the teaching opportunities here — the course tutors want to help you. Don’t get too frustrated if your code or project doesn’t work right away. If you run into bumps in your career, don’t stress over them too much. Everything will turn out okay.

Generally, I would say make the most of this positive environment. If you’re offered a tutoring position for example, I think it’s worth giving it a shot. Chasing these opportunities is really important.

To young women: If you think Bioinformatics is the right choice for you, don’t let anything stand in your way – I can recommend coming here.” 

Interested in the Bioinformatics study program at Saarland Informatics Campus? Get more information here. Applications for the master’s program for Winter Semester 25/26 will be open until May 15th!

To get acquainted with how Saarland University supports young female scientists, take a look at the page of the Gender Equality Office.

Editor:
Saarland Informatics Campus Team
Email: yagmur.akarsu@uni-saarland.de