Overview
An AI-powered content management system for medical documentation that understands medical terminology and context
AskMd is a specialised CMS designed for healthcare professionals to create, manage, and search medical documentation. Built at Medhack 2024, where it placed 4th, AskMd uses natural language processing to understand medical context and provide intelligent documentation assistance.
The Healthcare Documentation Problem
Healthcare professionals spend countless hours on documentation, often using systems that don't understand medical terminology or context. This leads to:
- Time wasted on repetitive documentation tasks
- Inconsistent note formatting across providers
- Difficulty finding relevant patient information quickly
- Increased risk of documentation errors
During our 1-week hackathon, we aimed to create a solution that would streamline medical documentation while maintaining accuracy and compliance.
Key Features
- Voice-to-Text Transcription: Dictate notes naturally using medical terminology with high accuracy
- Smart Templates: AI-powered templates that adapt to different note types (SOAP, H&P, discharge summaries)
- Semantic Search: Find patient information by meaning, not just keywords
- Auto-Completion: Intelligent suggestions for diagnoses, medications, and procedures
- Quality Checks: Automated validation for completeness and consistency
How It Works
AskMd leverages cutting-edge AI technologies to create a seamless documentation experience:
- RAG (Retrieval-Augmented Generation): Combines retrieval of relevant medical documents with language model generation to provide accurate, context-aware responses
- Vector Stores: Medical documents and patient notes are embedded and stored in vector databases, enabling lightning-fast semantic search across thousands of documents
- LangChain Integration: Orchestrates complex AI workflows, chaining together document processing, retrieval, and generation steps seamlessly
- Document OCR: Extracts text from scanned handwritten notes and PDFs, making legacy paper-based records searchable
Hackathon Experience
Building AskMd over 1 week at Medhack 2024 was an intense but rewarding experience.
Key challenges included:
- Integrating RAG pipelines and vector stores within tight time constraints
- Ensuring HIPAA-compliant data handling
- Creating an intuitive UX for busy healthcare providers
- Balancing feature completeness with demo polish
Despite the challenges, we successfully demonstrated a working prototype among 100+ teams.
Our Pitch
View Pitch Script
Doctors are drowning in paperwork. Searching through patient notes, scanning PDFs, and digging through EMRs takes valuable time away from patient care.
Our solution, AskMD is an AI-powered assistant that helps doctors instantly find key information from past patient notes. This saves them time they would have spent manually combing through lots of information.
Lots of hospitals still rely on a paper filing system with handwritten notes which is incredibly time consuming to read through manually. AskMD includes a document OCR system so doctors can pull out information from scanned documents in seconds.
Unlike cloud-based solutions, AskMD can run fully offline. This is great as sensitive health information never needs to leave the hospital network. No internet required, no data privacy risks—just a seamless, AI-powered workflow. We also have the ability to run a cloud based solution for less strict settings.
AskMD uses data by integrating with existing EMR systems, making it a simple process to onboard users.
We're targeting hospitals, clinics, and practitioners who work with lots of medical notes. Doctors shouldn't waste time digging through notes when we can do it for them. AskMD provides the data to help doctors make better decisions.
Technology Stack
Frontend
React, TypeScript, NextJS, TailwindCSS
Backend
NextJS
AI/ML
LangChain, RAG, OpenAI
Vector Store
Supabase Vector, OpenAI Embeddings
Database
Supabase
OCR
Tesseract, Azure OCR
Impact & Future Plans
The positive reception at Medhack validated the need for better medical documentation tools. Healthcare professionals who tested the prototype were particularly excited about the voice transcription and smart template features.
While we didn't place in the top 3 and just barely missed out on the chance to pitch our idea to the judges, the experience was invaluable. We learned a tremendous amount about healthcare workflows, AI integration challenges, and the real needs of medical professionals. The project remains a meaningful exploration of how AI can transform healthcare documentation.