Skip to content
The Lighthouse Lab
Research · Neuroscience

From Neural Dynamics to Recovery

Applying the same geometric tools to neural population activity, toward biomarkers that track recovery after neurological injury.

Neural population activity is usually reduced with PCA, which assumes a flat Euclidean space. Neural dynamics do not obviously live in one. We are extending dimensionality reduction to metric-manifold settings and testing the result against recurrent network models trained on the same behavior.

The applied goal is concrete: a measure of motor recovery that is sensitive enough to guide robot-assisted therapy, and that means the same thing across patients and sessions.

All research