SAILS — Speech & Action In-home Learning System
Simons Foundation (Site PI, 2025–2026); Kennedy Krieger Institute partnership
Into the Wild — automated quantification of autism-relevant behaviors from naturalistic home videos. The study pipeline runs end-to-end: parents are reached through SPARK Research Match, submit questionnaires and home-video uploads, and complete a remote language assessment; videos pass automated metadata checks and human quality review; trained annotators produce ground-truth labels against a shared manual with ICC reliability checks; and VLM-based models score uniform frame samples for video-level behavior predictions. Brings together three theses: Manaal Mohammed's on target-child behavior detection (vocalizations via USC-SAIL / Whisper baseline; locomotion via S3D gated 3D CNN), Lucie Bierent's on socialization-behavior recognition (response to name, non-verbal communication, gesture prediction), and Bruke Wossenseged's thesis work.
- child-development-education
- mental-health-psychiatry
- ml-nlp-knowledge
- speech-voice-biomarkers
Resources
Links
People
Contributors
External collaborators
- Rachel Reetzke (Kennedy Krieger)
- Rebecca Landa (Kennedy Krieger)
- Elizabeth Eiler (Kennedy Krieger)
- Arushi Singh (Kennedy Krieger)
- Jeremy Perrin (Kennedy Krieger)
- Vini Singh (Kennedy Krieger)
Edit this project in src/data/projects.yml. Add a paper in
src/data/publications.yml and tag it with the project slug.
See
PROJECTS.md
for conventions.