Predicting Brain Health Using a Smartwatch

A team from the University of Geneva shows that it is possible to predict brain‑health variations using data collected from smartwatches and smartphones. The study is led by Igor Matias, PhD assistant at the Research Institute for Statistics and Information Science and first author of the research.

To explore this potential, 88 volunteers aged 45 to 77 were monitored for ten months using a mobile application and a smartwatch measuring, among other indicators, heart rate, physical activity, sleep patterns, and air pollution. These passive data were complemented every three months with affective questionnaires and cognitive performance tests.

At the end of the study, an artificial intelligence system developed by the team analysed the collected information to predict cognitive and affective fluctuations. “On average, the error rate was only 12.5%,” explains Igor Matias. Affective states were the easiest to predict (5–10% error), while cognitive performance proved more difficult to estimate. Among the most informative indicators were weather conditions, air pollution, sleep variability and daily heart‑rate patterns.

This research, supervised by Professor Katarzyna Wacand Professor Matthias Kliegel (Faculty of Psychology and Educational Sciences), is part of the Providemus alz joint faculty project. The next phase, already underway, will extend over 24 months to refine the models and evaluate their relevance in real‑life individual contexts.

> Read the University of Geneva press release 

2026