Research Project

Geometry of Post-Stroke Behavior

Uncovering the geometry of post-stroke behavioral impairment and predicting long-term recovery

Aims

' Project in progress.

Description & Main Findings

Project in progress.

Tools & Methods

Nonlinear Dimensionality Reduction
UMAP
Factor Analysis
PCA
K-Means Clustering
Ridge Regression
Cross-Validation
Multivariate Lesion-Symptom Mapping
Python
Git
Apptainer
HPC

Deliverables

Manuscript submitted (2026). Presented at the Cognitive Computational Neuroscience meeting (Amsterdam, 2025), Society for Neuroscience (San Diego, 2025), and the Computational Psychiatry Symposium (Iowa City, 2025).

Nonlinear low-dimensional embeddings predict stroke outcomes Submitted to Human Brain Mapping · 2026

Predicting stroke recovery with nonlinear low-dimensional embeddings of behavioral profiles (extended abstract) Cognitive Computational Neuroscience · Amsterdam · 2025

Predicting stroke recovery with nonlinear low-dimensional embeddings of behavioral profiles (poster) Society for Neuroscience · San Diego · 2025 Computational Psychiatry Symposium · Iowa City · 2025