PocketOT

Duke Hackathon Submission — Fall 2024

PocketOT helps occupational therapists move faster from narrative → intervention direction.
Clients describe daily barriers across six domains — AI structures those reflections into clinical reasoning signals — and therapists use those signals to draft focused intervention plans more quickly.

---

Key Features

• 6-pillar client reflection assessment
• AI-assisted structuring layer (clusters reflection text → OT-relevant categories)
• Client selects top three priorities to preserve autonomy
• Therapist view showing pillar signals + directional starting point for interventions

---

Tech Stack

• React + Vite
• TypeScript
• Node / Vercel server endpoints (for AI calls)
• OpenAI API for structured output

---

Demo Flow

1. Client completes reflection questions
2. Model structures narrative into clinical reasoning categories
3. Therapist sees top pillar signals
4. Drafts interventions from a clearer starting point

---

Team

• Stephanie Sorto-Moreno — UX Research / Digital Health (Design & Technology Innovation, MENG @ Duke)
• Lewis Othniel — Occupational Therapy Doctorate @ Duke