PrescribatronJelqinator

Inspiration

"Can you provide your family history of chronic heart disease?"
"I just answered that on the sign-in form."
"Do you recall your Epstein-Barr virus titers from your last five blood tests?
"You're the doctor—I just get my blood drawn."

Healthcare should be fluid and intelligent, not a repetitive Q&A session. PrescribatronJelqinator gives doctors instant, structured access to a patient's entire medical history, allowing them to ask real questions instead of rehashing paperwork.

What it does

PrescribatronJelqinator is a secure, AI-powered medical assistant that allows doctors to:
✅ Instantly retrieve patient data from encrypted hospital records.
✅ Ask context-aware questions (e.g., "Does this patient have a history of anemia?").
✅ Run locally for privacy while maintaining HIPAA compliance.
✅ Summarize clinical notes and cross-reference medical history in real time.

Instead of sifting through records manually, a doctor can simply ask and get an accurate, contextual response based on the patient's real medical history.

How we built it

• Encrypted Patient Database 📂

  • Converts massive CSV-based hospital records into a 16GB+ encrypted SQLite database.
    
  • Uses AES-256 encryption on all sensitive fields.
    
  • Employs TF-IDF retrieval for fast medical record access.
    
  • Integrated with the Stanford STARR medical database for direct access to structured patient data from the Stanford ENT clinic.

• Locally Hosted RAG Model 🧠

  • Runs an Ollama-powered LLM that can answer doctor queries based on retrieved records.
    
  • Uses TF-IDF similarity to fetch the most relevant patient data.
    
  • Ensures zero cloud dependencies for full HIPAA compliance.
    

• Flask API for Secure Access 🔐

  • Uses JWT authentication to control doctor access.
    
  • Enforces TLS encryption to protect queries and responses.
    
  • Allows querying patient records in real time via REST API.
    

Challenges we ran into

• HIPAA Compliance & Encryption – Ensuring that all patient data is encrypted while still being efficiently retrievable.
  
• Running Models Locally – Deploying an LLM-powered system that can operate without cloud dependencies.
  
• Handling Massive Data – Managing 16GB+ patient records efficiently while ensuring real-time queries.

Accomplishments that we're proud of

• Successfully encrypting & indexing a massive patient database while keeping queries fast.
  
• Integrating a local RAG model that can answer questions directly from patient records.
  
• Ensuring full HIPAA compliance while allowing real-time retrieval of medical data.
  
• Making everything run entirely locally, meaning no external servers, no data leaks.
  

What we learned

• Efficient database encryption is possible without sacrificing performance.
  
• TF-IDF combined with LLMs is a powerful way to build real-time, document-aware AI.
  
• Doctors hate repetitive questions—automation can significantly improve medical interactions.
  

What's next for PrescribatronJelqinator

🚀 Cloud Hosting – Deploying to secure, hospital-grade cloud infrastructure.
🏥 EPIC Hospital Access – Integrating directly with EPIC medical record systems.
📝 Clinical Note Understanding – Allowing direct interaction with physician notes for deeper insights.
🎙 Doctor-Patient Voice Transcription – Enabling real-time note-taking based on conversations.

Note: we cannot include any gallery images for HIPAA compliance, but come check out our live-demo booth at the expo!!

Built With

• json