Predicting Cardiovascular Drift from 3D Running Pose via Spatio-Temporal Graph Attention

A spatio-temporal graph-attention model that infers cardiovascular drift directly from monocular 3D running pose, linking biomechanics to cardiovascular load without wearable instrumentation and providing a physiologically interpretable estimator for endurance monitoring.

With Yuval David (lead author, MSc student) and Alex M. Bronstein. Under review.