Problem
Support seizure-state detection and behavior analysis from visual observations.
Context
The current portfolio identifies pose estimation, seizure-state prediction, and analysis of induced and natural conditions.
My role
Developed system components across pose and prediction workflows.
Constraints
- Biomedical data and privacy requirements.
- Small or specialized datasets may limit generalization.
TODO_REVIEW: Confirm study protocol, collaborators, and publication status.
Architecture
TODO_REVIEW: Add the approved data, pose-estimation, temporal modeling, and
evaluation pipeline.
Technical decisions
TODO_REVIEW: Document model families, temporal windows, labeling process, and
validation design.
Trade-offs
TODO_REVIEW: Document sensitivity/specificity, interpretability, and deployment
trade-offs.
Results
TODO_REVIEW: Add verified results only after approval.
Screenshots
TODO_REVIEW: Add approved visualizations with appropriate privacy review.
Related links
TODO_REVIEW: Add public paper, lab, or project links.