AI Dental mentor

Dental Mentor: Gamified AI Training for Tomorrow’s Dentists. Practice fixed Dentistry in 3D simulation, lecture, diagnosis, and lab with an AI tutor, quizzes, & live leaderboards.

Comment

Inspiration

As a dedicated dental educator, I saw that students often struggle to connect textbook theory with real-world clinical practice. Inspired by the engagement of modern video games and the power of AI tutoring, I envisioned AI Dental Mentor: an immersive, gamified platform where learners across all specialties can practice, ask questions, and receive personalized feedback in real time.

What it does

AI Dental Mentor offers four interactive “rooms”:

  1. Dental Office – Simulate 3D tooth preparations with anatomically accurate models and get AI-driven corrections on taper, undercuts, and margins.
  2. Lecture Hall – Chat with a lovable, context-aware AI tutor (using Gemini RAG) for free-form Q&A and dynamic mini-lectures.
  3. Diagnosis & Treatment – Review patient histories, X-rays, and photos; answer MCQs; earn or lose points; and view instant rationale.
  4. Dental Laboratory – Select materials, design crown geometry, and simulate virtual milling, guided by AI recommendations.

How we built it

  • Unity & XR Toolkits: Crafted 3D scenes for desktop, AR, and VR using Unity’s XR Interaction Toolkit.
  • AI Backend: Developed a FastAPI service integrating Google Vertex AI’s Gemini API with a simple RAG vector store to power Q&A, quiz and case generation, and design feedback.
  • No-Code Wiring: Leveraged Bolt to bind every frontend button—quiz generators, “Ask Tutor” panels, document uploads—to backend endpoints without altering the UI.
  • Multiplayer & Leaderboard: Integrated Photon PUN for real-time group sessions, synchronized scores, and a shared leaderboard.

Challenges we ran into

  • Balancing Realism & Performance: High-fidelity tooth meshes initially caused lag. We optimized shaders, reduced polygon counts, and introduced local caching for repeated AI prompts.
  • Seamless AI Integration: Early GPT calls were too slow for interactivity. We refined prompt templates, limited token usage, and implemented lightweight pre-caching of common queries.
  • UI Integrity: Wiring new functionality risked shifting layouts. We used Bolt’s no-code bindings and automated tests to ensure every button still looked and behaved exactly as before.

Accomplishments that we’re proud of

  • Delivered a fully functional four-room gamified platform covering all dental specialties.
  • Achieved real-time AI feedback on student actions, from quiz scoring to mesh-based prep analysis.
  • Enabled AR/VR deployment, letting students summon the tutor in their environment or step into a virtual clinic.
  • Built a secure, role-based system where teachers can upload custom materials and track class performance.

What we learned

  • The importance of modular prompt designfor AI: small, focused Gemini prompts yielded faster, more accurate responses.
  • How to maintain UI/UX consistency when layering on complex AI logic and multiplayer networking.
  • Best practices for asset optimization: anatomically correct models with PBR textures can run smoothly with careful LOD and shader tuning.

What’s next for AI Dental Mentor

  • Expand RAG to include large journal databases (via Pinecone/FAISS) for richer case studies.
  • Adaptive learning pathways: use student performance data and bandit algorithms to tailor question difficulty in real time.
  • Enhanced analytics dashboard for educators: deeper insights on student progress, common pitfalls, and curriculum gaps.
  • Voice interaction: integrate speech-to-text and text-to-speech for hands-free queries and narrated walkthroughs in AR/VR.

Built With

  • ar-foundation
  • c#
  • fastapi
  • openxr
  • photon-pun-2
  • python
  • unity-xr-interaction-toolkit
Share this project:

Updates