← all work
04Jan 2019 — May 2021Lead researcher · IIT Madras

Mapping Rehabilitation Therapy to Lesion

A feedforward neural network that captures the basics of bimanual reaching, into which a simulated stroke is introduced — then used to map which rehabilitation strategy best suits which patient.

KerasCNNBlenderStroke
[ lesion map / recovery heatmap ]
// approach
Build a simple feedforward network capturing the basic features of bimanual reaching.
Introduce a simulated stroke into the network.
Analyze how increasing the complexity of movements used in retraining affects recovery.
Expand the network across movement types and map rehabilitation characteristics to patient characteristics.
// outcomes
Links specific therapy parameters to specific patient and lesion profiles.
Points toward the optimal therapy for a given patient.
// related publications
Interaction of network and rehabilitation therapy parameters in defining recovery after stroke in a bilateral neural network
J. NeuroEng. & Rehab · 2022
// related patents
GRANTED
System For Handling Rehabilitation Strategy Training For Subject
Patent No. 507924 · granted Dec 2024
← previous
next →