Related Experiment Video
Updated: Jul 7, 2026
05:54
MRI-guided Focused Ultrasound Thalamotomy for Patients with Medically-refractory Essential Tremor
Published on: December 13, 2017
Decoding pre-movement neural activity from thalamic LFPs for adaptive neurostimulation in tremor patients
Fernando U Rodriguez Plazas1,2, Thomas G Simpson1, Laura Wehmeyer1
1Medical Research Council Brain Network Dynamics Unit, Nuffield Department of Clinical Neurosciences, University of Oxford, Oxford, UK.
Neurocomputing
|July 6, 2026
Summary
Machine learning models can predict upper-limb movements using brain signals before they happen. Personalized, multi-band approaches are key for adaptive deep brain stimulation in tremor disorders.
Area of Science:
- Neuroscience
- Biomedical Engineering
- Machine Learning
Background:
- Deep brain stimulation (DBS) is used for tremor disorders.
- Adaptive DBS requires predicting movement onset.
- Neural oscillations in the thalamus and scalp EEG may hold predictive information.
Purpose of the Study:
- Investigate machine learning feasibility for decoding pre-movement neural signals.
- Predict upper-limb movement onset using thalamic local field potentials (LFPs) and scalp electroencephalography (EEG).
- Advance adaptive DBS for tremor disorders.
Main Methods:
- Recorded thalamic LFPs and scalp EEG from 11 patients undergoing DBS for tremor.
- Utilized machine learning models (logistic regression, gradient-boosted decision trees, CNNs) to differentiate rest from pre-movement states.
- Developed patient-specific decoders.
Main Results:
- Achieved above-chance movement prediction ~430 ms before onset with thalamic LFPs and ~840 ms with EEG.
- Patient-specific models outperformed general models due to inter-patient variability.
- Multiple frequency bands independently contributed to decoding accuracy, emphasizing multi-band analysis.
Conclusions:
- Early neural signals preceding movement can be decoded using machine learning.
- Personalized, multi-band machine learning models are valuable for capturing pre-movement neural activity.
- Findings support the development of tailored adaptive neurostimulation therapies for tremor.