Our team was selected to participate in the “Solve for Healthcare & Life Sciences with Gemma” hackathon, hosted by Google Paris! I had the pleasure of teaming up with my former ETH Zürich lab-mate Patrick Langer, along with Nour Ben Rejeb and Nathan Denier.
We were honored to receive the Impact Award (2nd place) among 126 participants across 26 teams, a great recognition of how our research at ETH Zürich and Stanford University translates into real-world impact.
VitalCue team receiving the Impact Award (2nd place).
VitalCue & WearableGemma: From Smartwatches to to Actionable Insights
Meet Max. Like many of us, Max leads a busy life and has been waking up tired every single day. His smartwatch shows a low sleep score, but Max doesn’t know what to do with that data. It’s a common problem – we’re surrounded by vital information, yet most of us don’t know how to act upon it. Max is worried: what if this is an early sign of something more serious?
The bigger picture reveals a critical gap: 97% of healthcare spending today goes to treating illness, while only 3% is spent on prevention. But prevention is where the real impact lies, both for people’s lives and for lowering costs for our communities. What if we could identify issues before they turn into more serious health problems?
That’s where VitalCue comes in. Our solution connects to Max’s smartwatch, and when he reports symptoms like fatigue and poor sleep, VitalCue starts diving deep into data from wearable devices.
At the heart of VitalCue is WearableGemma, a new architectural variant of Gemma 3 that uses cross-attention mechanisms inspired by DeepMind’s Flamingo. This model is purpose-built to interpret time-series data from wearables like heart rate, heart rate variability, and step counts—all the things your smartwatch already measures today.
Technical Innovation
Using cross-attention, WearableGemma can analyze both an arbitrary number and length of vital signals. Because VitalCue uses data from your smartwatch, it can assess vital patterns over time—not just a single snapshot like in a doctor’s appointment. Combined with user-reported symptoms, it generates personalized insights and recommendations, from improving sleep hygiene to suggesting a doctor’s visit.
In Max’s case, VitalCue detected a pattern: his body was under physical stress late at night, disrupting his sleep cycle. This insight helped Max adjust his sports schedule and regain his energy.
After a fiercely competitive first pitch round, only 10 teams advanced to the finals—each one being incredibly strong. Our solution stood out for bringing health into people’s daily lives with technology they already own.
Hackathon participants at Google Paris.
Recognition and Impact
The jury feedback highlighted that we’re “bringing health to everyone by combining hardware with Google’s open models, helping users catch early signs of health issues and supporting prevention.” This recognition validates our approach of making advanced AI-powered health monitoring accessible through everyday wearable devices.
Huge thanks to the team for the incredible collaboration, and to the organizers at Google DeepMind and Google for hosting such an inspiring event!