Project Overview

• Purpose: MyMIA is an AI-powered clinical assistant that eliminates administrative burden for Malaysian doctors, allowing them to focus on patient care instead of paperwork.
• Users: Malaysian healthcare professionals, primarily doctors in public hospitals and clinics who currently spend 20-40% of their time on manual documentation and system fragmentation.
• Core Problem: Our survey of Malaysian doctors revealed they waste 2.5 hours daily navigating 3-6 disparate systems, manually transcribing data, and completing handwritten documentation - time stolen from patient care.

Proposed Solution

MyMIA provides four core capabilities:

1. Unified Patient Dashboard: Consolidates fragmented patient data into one intelligent interface
2. AI Clinical Assistant: Patient-specific chatbot answering queries about medical history and lab results
3. Intelligent Patient Synthesis: Auto-generates comprehensive patient summaries from scattered records
4. Voice-First Documentation: Converts natural language dictation into structured clinical notes

Workflow

1. Doctor enters patient IC number → MyMIA displays unified patient record
2. AI instantly synthesizes patient summary from historical data
3. Doctor asks contextual questions: “Does this patient have heart issues?”
4. During consultation, voice recording auto-categorizes into clinical notes, diagnosis, and treatment plans
5. All documentation updates in real-time across the unified system

Measurable Impact

Based on our doctor survey:

• Admin Time Reduction: From 2.5 hours to under 1 hour daily
• Clinical Time Gained: 1.5 hours daily returned to patient care - nearly 8 hours per week
• System Efficiency: Eliminates juggling 3-6 different systems through unified platform
• Error Reduction: Voice documentation eliminates manual transcription errors
• Risk Mitigation: Addresses medico-legal documentation concerns through better digital documentation and audit trails

User Experience & Demo Steps

1. Patient Search: Enter IC → instant comprehensive patient view
2. AI Summary: Auto-generated patient overview appears with medical history
3. Contextual Q&A: Ask “Patient’s diabetes control status?” → get instant, cited response
4. Voice Documentation: Speak during consultation → text auto-categorizes into sections
5. Seamless Integration: All updates sync across unified platform

Validation & Feedback

• Primary Research: Surveyed 8 Malaysian doctors across specialties and locations
• Pain Point Validation: 100% confirmed system fragmentation and manual documentation issues
• Solution Alignment: Features directly address top complaints (time waste, duplication, medico-legal risk)

Vision: Purpose-built for Malaysian public healthcare reality (system fragmentation, manual workflows, infrastructure constraints)

Creativity:

• Unified Patient Intelligence: Consolidates fragmented patient records into one searchable platform
• Conversational Patient History: AI chatbot that instantly answers doctor queries about patient history and lab results
• Intelligent Documentation: Voice-to-text that automatically categorizes into clinical notes, diagnosis, and treatment sections
• Patient Summary: AI auto-generates patient summaries from fragmented records enables doctors to understand patient condition quicker
• Impact: Transforms fundamental doctor-patient interaction by reclaiming clinical tim

User Experience

• Intuitive Design: Single IC search → complete patient view
• Wow Factor: Voice-to-structured documentation with Malaysian English processing
• Seamless Journey: Eliminates system switching and manual transcription

Technical Implementation

• Functionality: Working prototype with core AI features deployed
• Innovation: Multi-modal AI (voice, text, contextual understanding)
• Intelligence: Patient-specific chatbot with medical context awareness
• Architecture: Hybrid cloud design for Malaysian infrastructure reality

Scalability & Deployability

Technical Scalability:

• Cloud-native architecture supporting thousands of concurrent users
• API-first design for integration with existing EMR systems
• Multi-language support for Malaysian healthcare diversity

Implementation Pathway:

• Phase 1: Pilot with MoH in selected hospitals
• Phase 2: Integration with Ministry’s EMR rollout
• Phase 3: Nationwide deployment across public healthcare

Sustainability Model:

• Open-source core components
• Malaysian data sovereignty compliance
• Long-term partnership approach with continuous improvement

Technologies

Frontend

• Next.js 15
• JavaScript
• TailwindCSS
• Vercel

Backend

• FastAPI
• Python
• Pydantic
• Haystack
• Dokploy
• Docker
• GCP
• Vertex AI
• Hetzner
• Groq
• Ollama
• GitHub Actions
• GitHub Container Registry (GHCR)

Deployment
Frontend (Vercel)

• Public URL
• Optimized build for Next.js
• Integration with GitHub, CI/CD triggered by webhooks

Backend (Hetzner)

• VPS with medium specs (8 CPUs + 16 GB RAM), applications managed by Dokploy
• Integrated with GitHub, CI/CD triggered by webhooks, dockerized image push to GHCR and fetched by Dokploy for deployment
• Serve RESTful API endpoints to frontend via secure API-KEY authentication and CORS

Generative AI

• Ollama (Development)
• Qwen3-8B (Complex Reasoning)

• Llama3.2-3B (Chatbot)

• Groq (Production)
• Llama-3.3-70B-Versatile (Complex Reasoning)
• Llama-3.1-8B-Instant (Chatbot)
• Gemini 2.5 Flash-Lite (Speech-to-text)

Prior to the hackathon, our team set up the GitHub repository, backend, and frontend framework, designed the initial webpage layout, AI Summary, implemented a demo login, and conducted a Google Form survey to identify doctors’ administrative pain points.

During the hackathon, we focused on resolving error messages, integrating an API for conversation transcription, Voice-to-text features, AI Chatbot, and refining the UI for a cleaner, more user-friendly experience.