Podcast Applied Science UNBOXED

From Thesis to Publication – How Smartwatches could transform Diabetes Care

What if an everyday smartwatch could save lives by detecting dangerous drops in blood sugar before they happen? Episode 9 of Applied Data Science UNBOXED follows how one master’s student turned a class project into real medical research, using data science to improve life with diabetes.

Podcast: Applied Data Science UNBOXED
Episode 9: From Thesis to Publication – How Smartwatches could transform Diabetes Care
Host: Fabio Sandmeier
Guest: Yasmine Mohamed and José Mancera

Shortcuts:
Motivation | Idea | Challenges | Breakthrough | Supervisor | Takeaway

Host Fabio Sandmeier meets Yasmine Mohamed, graduate of the Master in Applied Information and Data Science, and her supervisor José Mancera, lecturer in data engineering and machine learning. Together, they talk about Yasmine’s master’s thesis journey from idea to publication, the technical and ethical hurdles along the way, and what it means to turn data into something that truly makes a difference.

00:11 – Motivation: Purpose over paperwork

“I don’t want to do something just to graduate. I want to do something impactful.”

With a background and family ties to healthcare, Yasmine set out to build something with real use. Her idea took shape in a start-up module: use wearables and data to support people with chronic conditions. Diabetes became her focus after personal experiences showed how fast and disorienting hypoglycemia can be.

Listen to this part (00:11)

05:26 – The idea: From medical-grade sensors to everyday watches

Prior research often used medical-grade equipment like ECG chest straps to capture physiological changes during hypo events. Yasmine asked a simple question: what if we used off-the-shelf smartwatches instead, devices people already wear while working, swimming, sleeping. The concept was simple, feed synchronised smartwatch signals and CGM glucose data into a model learning patterns that precede low blood sugar.

Listen to this part (05:26)

08:15 – Challenges: Ethics, funding and the long wait

Ambition quickly met reality. To bring the project to life, the team needed funding for the devices, ethical approval to conduct a human study, and clinical partners willing to collaborate. Getting the green light from the ethics committee alone took around six months. Designing the study also came with challenges: for methodological reasons, some participants were temporarily blinded to their CGM readings. Once all approvals were in place, a new challenge began: collecting enough high-quality data from real participants.

Listen to this part (08:15)

12:09 – Breakthrough: First data and the eureka moment

After months of waiting, the first data finally began to flow into Yasmine’s AWS S3 bucket: raw, messy, and full of promise. She spent countless hours cleaning, aligning, and transforming the signals, determined to make sense of the chaos. Then, one day, the breakthrough came. The first models began to recognise patterns, subtle physiological changes that signalled an upcoming drop in blood sugar. It was her long-awaited eureka moment: proof that everyday smartwatches, combined with machine learning, could indeed help detect hypoglycemia before it becomes dangerous.

Listen to this part (12:09)

15:21 – The Supervisor’s View: Guiding, not leading

“You are the tech lead here, the project manager”

With these words, José Mancera set the tone in their first meeting. As Yasmine’s supervisor and lecturer in data engineering and machine learning at the MSc in Applied Information and Data Science, he made it clear that this was her project. From then on, José guided rather than led, while Yasmine took full ownership: coordinating across hospitals, managing timelines, data workflows and communication with doctors and patients. José describes her drive as exceptional: she was constantly knocking on doors, building bridges between the academic and clinical worlds. What impressed him most was her ability to think and act like a true data scientist, combining technical precision with leadership, empathy and persistence.

Listen to this part (15:21)

Get the details for the Master’s programme “MSc in Applied Information and Data Science”

19:18 – Results and outlook: Toward needle-free support

Yasmine’s master’s thesis shows that consumer smartwatches can help detect hypoglycemia in people with type 1 diabetes with promising accuracy, without needles or invasive devices. The work has been published in a scientific journal and now invites others to validate, extend and translate the findings. The project perfectly reflects the spirit of the Masters programme, applying data to create real-world impact. Yasmine has since moved into a research role in health data science, working on detection of tropical diseases such as malaria and dengue, staying true to the mission that started this journey.

Listen to this part (19:18)

Key Takeaway: Impact needs persistence

Applied data science is not just about models, it’s about persistence, patience, and purpose. Turning data into impact takes time: building partnerships, earning approvals, and staying the course. With the right mindset and mentorship, a student project can grow into research that improves lives.