Skip to main content
Medical Design Program
< Health Design Sprint >

Rapid Prototyping with AI

We are entering a new AI-driven tool development era where domain expertise trumps coding ability. Large language models have democratized software development. If you can articulate the digital tool you want to create, AI can help you build it. Quickly, cheaply, and customized to your vision.

In the OLD ERA
The hard part was building the app.
In the NEW ERA
The hard part is knowing what app to build.

Medical students, with cutting-edge clinical knowledge and direct access to patients and workflows, are uniquely positioned to lead in this next era of healthcare innovation. But they need new forms of training and hands-on experience to develop the necessary knowledge, skills, and (perhaps most importantly) mindsets. The Medical Design Program’s “Health Design Sprint” focuses on all three.

Core Learning + Doing Loop

The course runs on a single repeating workflow that builds muscle memory: Claude Chat → Product Requirements Document (PRD) → App Builder (Lovable or Claude Code) → Fresh Build. Students generate a fresh build every day from their enriched PRD—not iterating on yesterday’s build. By comparing daily builds, they see directly how richer planning yields richer apps.

1
Chat with Claude
Brainstorm, structure thinking, enrich your PRD
2
Add PRD Layer
Each day adds a new dimension: users, design, features
3
Fresh Build
Generate a new app from the enriched PRD
4
Compare & Reflect
What changed? What did the richer PRD give the AI?
← Each version gets richer. By end of Week 1: five versions of the same app.
Key Insight

The valuable skill isn’t coding syntax — it’s strategic thinking about what problems matter and how to build solutions iteratively.

Framework: Knowledge, Skills, Mindsets

Every session develops all three dimensions:

Knowledge

How Large Language Models (LLMs) work, PRD-driven development, design thinking methodology, healthcare data and privacy considerations

Skills

Structured planning with AI, rapid prototyping in Lovable, user research, iterative PRD layering, user testing

Mindsets

Comfort with ambiguity, embrace messiness before structure, AI as partner not oracle, process over polish

Layered PRD Builds

Students build real apps from Day 1. Each day introduces a new design dimension that gets added to the PRD set—and each updated PRD produces a new, more sophisticated app version.

masterplan
user journey
design
features
polish
< Week 1 — Team Challenge-Area Apps
Day 1Three Tools
Day 2Users
Day 3Features
Day 4Design
Day 5Demo Day
> Week 2 — Solo Specialty Apps
Day 6Chat → Agent
Day 7Jigs
Day 8Deep Build
Day 9Polish
Day 10Demos

AI Tools Integration

<>
Think
Claude
<
Store
GitHub
>
Build
Lovable
<<
Present
Gamma

The curriculum emphasizes why to choose specific tools, not just how to use them—building transferable judgment as the landscape evolves.

Course Structure

10-day intensive, virtual-first for equity across sites
Team challenge-area apps (Lovable). Each day adds a PRD layer.
Solo specialty apps (Lovable or Claude Code). Chat → Agent paradigm shift.
No coding experience required. Domain expertise is the primary building material.

This elective builds on the established foundation of the UVA Medical Design Program, which has trained 200+ students in human-centered design methods since 2017. The Health Design Sprint extends that work into the AI era — integrating vibe coding tools while preserving the program’s core commitment to understanding users and problems before building solutions.

The methodology is designed to be transferable — institution-agnostic, documented, and evidence-backed. Any medical school with domain experts, design thinking facilitation, and access to AI tools can run a sprint.

See what students built → or browse sprint archives.

What We Measure

Each sprint is designed as a learning experience for the institution, not just for students:

  • How non-technical domain experts use AI tools in intensive learning contexts
  • Time-to-functional-prototype for domain experts using vibe coding workflows
  • Impact of layered PRD approach on prototype quality and student learning
  • Failure modes, pedagogical pivots, and what scales beyond each cohort

The 10-Year Foundation

The Health Design Sprint builds on a decade of teaching design thinking to medical students through the UVA Medical Design Program.

  • 2017–2022: Monthly workshops and real-world client projects — emergency department redesign, patient experience research, community health partnerships. Proved that medical students can do rigorous design thinking. Iterated the pedagogy across 5+ cohorts and 200+ students.
  • 2023–2025: The AI inflection. Large language models went from curiosity to capability. The design thinking methodology acquired a powerful new output: students could go from insight to working prototype in the same sprint.
  • 2026: The breakthrough. Two weeks, eight students, five production-quality healthcare applications. A fully documented, replicable methodology — the Signal-to-Prototype Loop.

Why This Matters

Medical schools everywhere want to teach with AI but lack proven models. This methodology — stress-tested across 9 cohorts and 200+ students — is designed to transfer.

Small cohort, intensive format, documented outcomes, transferable framework. Success here provides a template that any institution can adapt. The data we gather informs decisions about AI tool access, curriculum design, and the role of domain expertise in the age of generative AI.