Kyuhwa Lee
NeuroAI Group Leader
Wyss Center, Geneva, Switzerland
I develop time-series models and real-time inference systems to infer human state, intention, and behavior from high-dimensional longitudinal biomedical and behavioral data. My methods characterize how these signals are structured and how their representations change under non-stationarity. I have applied them to neural decoding of visual perception, exoskeleton and exosuit control, and movement restoration in people with Parkinson's disease and spinal cord injury.
News
- Real-time detection of gait and freezing states from subthalamic activity now drives adaptive deep brain stimulation, in our trial published in Nature Medicine. (Jun 2026)
- Guest lecture at Keio University, "Learning stable neural representations under longitudinal non-stationarity", hosted by Prof. Junichi Ushiba. (Jun 2026)
- Plenary speaker at the 6th Korea-Japan-Taiwan Neurorehabilitation Congress in Seoul, on implantable BCI for neurorehabilitation. (Mar 2026)
- MNE-LSL, a real-time framework bringing MNE-Python to online neuroscience through LSL-compatible devices, appeared in the Journal of Open Source Software. (2025)
- Invited by the Swiss State Secretariat for Education, Research and Innovation to speak at the 12th Swiss-Korean Life Science Symposium, Seoul, "Using AI to Decode Movement Intention". (2025)
- IEEE Spectrum covered the exosuit work, "Exosuit Muscle Control Steps Closer to Reality". (Mar 2024)
- Became Group Leader of the NeuroAI Lab, and started the longitudinal drift program with it. (Jan 2024)
- Synapsuit began (role: PI), a Swiss-Korean project on motor restoration with implanted soft electrodes and a wearable exosuit, funded by Innosuisse and KIAT (CHF 1.02M, Swiss side). (Apr 2023)
Beyond research
Some of my works on music production and piano recordings can be found on SoundCloud. I occasionally take photographs, collected long ago at findingaura.net. Both are old, and I have no plans to tidy them.