Please introduce yourself. What is your background?
My name is Loris Constantin, and I grew up in Valais before moving to Lausanne to pursue my studies at EPFL. Over the course of six years, I earned my Bachelor's in Life Science Engineering and later completed a Master's in Computational Biology. I’m deeply passionate about applying the best of computer science to biology, driven by the meaningful goal of improving people’s lives. In fact, I plan to begin a PhD focused on this very intersection.
Outside of my academic pursuits, music has always been a big part of my life. I’ve been playing in a Brass Band for 17 years and also taught myself to play guitar and sing. On the sports side, I enjoy running and climbing, with occasional skiing, and I used to play football as well. I like to think of myself as someone who’s open to new challenges and adventures, though I also place great value on what I have at home.
You have received the SSBE award. Please briefly describe the project.
This project focused on adapting cutting-edge AI technology, initially designed to perform sleep staging from PPG signals recorded at the finger in a lab, for ambulatory use. The motivation behind this was to address the limitations of current sleep staging methods, which, while crucial for diagnosing sleep disorders, are costly, time-consuming, and inconvenient. Patients need to stay overnight in a sleep lab, where multiple signals are recorded and manually analyzed by experts. Our goal was to provide an alternative by using wrist-based PPG data from a smartwatch, allowing sleep staging to be done at home.
We encountered two major challenges: a lack of data to train the algorithm and significant variability between datasets, both of which made it difficult to develop a robust, high-performing model. Despite these difficulties, we successfully created a highly effective algorithm that is now state-of-the-art in terms of performance and robustness for wrist PPG.
What does the SSBE award mean to you?
Receiving this award marks a deeply satisfying conclusion to both my Master’s studies and my overall journey at EPFL. It’s not just a validation of my work, but a testament to the support I’ve received from my professors, colleagues, and everyone who has been part of my growth along the way. This recognition encourages me to keep pushing my limits, serving as a reminder that I’m on the right path. It gives me confidence to tackle future challenges and seek opportunities where I can continue making a meaningful impact. Above all, it strengthens my commitment to contributing positively to the field of biomedical engineering and beyond.
Is this project still active? What are the goals?
The project has reached already a satisfactory state, but we can further move closer to the goal we set for ourselves. Although we have a very convincing night-recordings algorithm, we set out on the one hand to further improve its performance using even more complex techniques, and on the other hand to extend its capacities to 24h monitoring (some preliminary work done in this sense). Indeed, 24h sleep monitoring based on PPG has to potential to be more accurate than current actigraphy-based methods, and it would as well enable to have a look at how well we can predict naps, thus giving a more complete view of a patients’ sleep routine.
Did the award have an impact on your career?
As I won it very recently, I cannot say it has already had an impact, but in the future I think it can certainly bring people to trust my capacities and my commitment. Very surely, I has already had an impact on how I see my work, as I explained before.
What kind of work are you currently doing? How is it related to your studies?
I am currently working in Singapore, and will do so for the next 5 months. The field of work is still linked to AI and data analysis, although for drug design against cancer. As my Master’s thesis, it is very in line with what I studied, although different enough that I expect it will complete my view of the field.
What is special about your current work?
AI for drug design, especially combined to the use of large language models (LLMs) is extremely new, and rather complex. It will be captivating to see how things evolve in such a fresh setting. Beyond the work itself, the opportunity to work in Singapore, a very international country of south-east Asia, allows me to interact with professionals from quite various cultures and is likely to be very enriching.
The Sleep Staging Pipeline begins with PPG data collected overnight from wrist-wearable devices. After some preprocessing, the signal is processed by our deep learning model, which utilizes two ResNet modules for feature encoding and context aggregation. The model then outputs sleep stages for each 30-second window with a median accuracy of 78%, ensuring results that are both unbiased and reproducible. This pipeline has shown to be robust to various other datasets, with promising results on 24-hour recordings.
Where do you see yourself in five years?
In 5 years, I hope I will have achieved to finish the PhD I am planning. We will see at that moment if I still want to work in a research-oriented position, or whether I need a change. As I said, I have very open to new challenges and I am not planning so far ahead. But most likely I will stay in Switzerland.
Is there something that you would like to convey to the SSBE members?
I’m truly honored to have received this award; it means a lot to me and serves as a great source of encouragement. I do wish I could have been there in person for my presentation, but the timing was too tight with my upcoming employment in Singapore. I’m grateful that you were able to arrange for me to present in real-time regardless, and I hope my talk met expectations. In the future, I definitely look forward to attending the SSBE conference in person.