AI-Driven Medical Observation & Decision-Support Platform 🏥
An intelligent clinical support system for real-time patient monitoring, clinical reasoning assistance, and medical education - built under strict ethical governance for students, clinicians, and hospitals.
🎉 MedObsMind is LIVE and accessible now!
| Platform | URL | Status |
|---|---|---|
| 🌐 GitHub Pages | https://sharmapank-j.github.io/MedObsMind | ✅ LIVE NOW |
| ⚡ Netlify | Deploy to Netlify → | 🚀 One-Click Deploy |
| 🌍 Custom Domain | https://medobsmind.in | 🔄 Ready (DNS Setup) |
| 🔌 API Endpoint | https://api.medobsmind.in | 🔄 Ready (After Backend Deploy) |
| 👨💼 Admin Portal | https://admin.medobsmind.in | 🔄 Ready (After Backend Deploy) |
Quick Access: Click Here to Visit MedObsMind →
Deploy Your Own:
MedObsMind is an AI-powered medical observation and clinical support system designed to improve patient safety, enhance clinical reasoning, and support medical education. It combines real-time vitals monitoring, intelligent alerts, and evidence-based decision support to help healthcare professionals detect patient deterioration early and make informed clinical decisions.
"An India-trained, rule-first, safety-locked medical AI system that continuously observes ICU patients, explains clinical risks using a medical LLM, and always keeps human clinicians in full control."
MedObsMind follows a strict hierarchy:
- Rules Fire First - Clinical protocols (NEWS2, qSOFA, threshold alerts) trigger FIRST
- LLM Explains After - Medical AI explains why alert fired and provides context
- Human Decides Always - Clinician has final authority on all clinical decisions
What MedObsMind Does:
- ✅ Continuous 24/7 patient monitoring
- ✅ Transparent clinical reasoning with explanations
- ✅ Evidence-based alerts and recommendations
- ✅ Educational support for medical students
What MedObsMind Does NOT Do:
- ❌ Autonomous diagnosis (always advisory)
- ❌ Prescribe medications (human decision required)
- ❌ Override clinician judgment (human has final say)
- ❌ Black-box decisions (all reasoning is explainable)
Governance Structure:
- d³media (Parent Authority) - Ethics, rights, oversight, and safety audits
- d²media (Operational Layer) - Technology deployment and operations
- Domain Classification - High-impact medical AI requiring dual supervision
This hierarchical governance ensures that MedObsMind operates with the highest standards of medical ethics, patient safety, and regulatory compliance.
MedObsMind is built around four integrated core modules:
Real-time patient monitoring and trend analysis
- Vital signs tracking (HR, BP, SpO₂, RR, Temperature)
- Automated trend analysis and pattern detection
- Anomaly detection with intelligent filtering
- Time-series visualization and historical comparisons
- Multi-patient dashboard for ward/ICU monitoring
Evidence-based decision support and alerts
- Differential diagnosis suggestions (advisory only)
- Red-flag alerts for critical conditions
- Severity scoring support (NEWS2, MEWS, qSOFA)
- Lab-vitals correlation and interpretation
- Risk stratification (sepsis, shock, deterioration)
- Explainable AI - shows reasoning for all suggestions
Hospital workflow optimization and documentation
- OPD (Outpatient) flow management
- Ward round assistance and checklists
- ICU task prioritization and handovers
- Automated round summaries
- Shift handover documentation
- Clinical note templates and auto-completion
Case-based learning and clinical reasoning aligned with NMC standards
- Interactive case studies with AI explanations
- Clinical reasoning walkthroughs (why/how)
- AETCOM module integration - Ethics, communication, professionalism
- NMC competency tracking - K/KH/S/D progression
- Exam correlation with real-world scenarios
- Bedside learning companion for students
- Mistake analysis and learning feedback
- MBBS curriculum alignment (CBME)
- Digital portfolio for competency assessment
AETCOM Features:
- Ethics scenario simulations (resource allocation, consent, end-of-life)
- Communication skills practice (breaking bad news, family counseling)
- Reflective learning prompts and journals
- Cultural sensitivity training for Indian healthcare context
See AETCOM & NMC Integration for details
-
MBBS Students & Interns
- Bedside learning with AI guidance
- Case-based clinical reasoning practice
- Safe environment to learn from mistakes
- Exam preparation with real scenarios
-
Residents & Clinicians
- Real-time decision support during rounds
- Evidence-based recommendations
- Workload management and prioritization
- Continuous professional development
-
Hospitals & ICUs
- Multi-patient monitoring dashboards
- Early warning system for deterioration
- Quality metrics and safety analytics
- Resource optimization
-
Rural / Low-Resource Setups
- Offline-capable operation
- Low-bandwidth optimized
- Affordable deployment
- Multilingual support (coming soon)
- ICU Early-Warning System - Continuous monitoring with intelligent alerts
- Ward Deterioration Detection - Catch declining patients before crisis
- Emergency Triage Support - Risk-based prioritization
- OPD Decision Support - Evidence-based suggestions for common presentations
- Rounds Assistance - Structured approach to patient assessment
- Handover Quality - Complete patient summaries for shift changes
- Student Bedside Learning - Interactive learning during clinical rotations
- Case-Based Teaching - AI-explained real patient scenarios
- Clinical Reasoning Development - Step-by-step diagnostic thinking
- Remote Monitoring - Telemedicine integration
- Community Health - Primary care decision support
- Research Platform - De-identified data for clinical studies (with approval)
MedObsMind operates under strict ethical principles:
-
No Autonomous Diagnosis
- AI provides suggestions, not conclusions
- Human clinician makes all final decisions
- System is assistive, never authoritative
-
Human Override Always
- Clinicians can override any AI suggestion
- All overrides logged for learning
- Clinical judgment is paramount
-
Transparency & Explainability
- Every alert shows reasoning
- Confidence levels clearly displayed
- Sources cited for recommendations
-
Safety & Bias Audits
- Regular safety audits by d³media
- Bias detection and mitigation
- Performance monitoring across demographics
- Adverse event tracking and analysis
-
Data Ethics
- Patient privacy by design
- Minimal data collection
- No data selling ever
- HIPAA/DISHA compliance
- Secure de-identification for research
-
Continuous Oversight
- d³media ethical oversight
- d²media operational monitoring
- Regular clinical validation
- User feedback integration
"Free for learning, affordable for care, never at the cost of patient safety"
Students & Education
- Free access for MBBS students and interns
- Subsidized pricing for medical colleges
- Education mode always free
- Research access with approval
Hospitals & Clinics
- SaaS model: Per-bed or per-facility pricing
- Tiered plans (OPD, Ward, ICU)
- Flexible licensing for resource-limited settings
- Government/NGO special pricing
Research Collaborations
- Approved collaborations only
- Ethics committee review required
- No commercial data selling
- Open science contribution
Prohibited
- ❌ No selling of patient data
- ❌ No advertising to patients
- ❌ No autonomous diagnostic claims
- ❌ No unlicensed medical advice
- ✅ Real-time vitals input (manual + device-ready)
- ✅ Trend graphs (HR, BP, SpO₂, RR, Temp)
- ✅ Rule-based alerts (NEWS2, MEWS)
- ✅ Patient summary auto-generation
- ✅ Basic patient management
- 🔄 AI risk prediction (sepsis, shock, arrest)
- 🔄 Lab + vitals correlation
- 🔄 Shift-wise doctor notifications
- 🔄 Explainable AI (why alert triggered)
- 🔄 Historical trend analysis
- 📅 ICU workflow assistant
- 📅 Drug dose safety checks
- 📅 Voice input for rounds
- 📅 Offline-first mode (India-specific)
- 📅 Multi-hospital deployment
Backend
- FastAPI (Python) - High-performance async API
- PostgreSQL - Primary database with FHIR-ready schema
- Redis - Alert queuing and caching
- SQLAlchemy - ORM with async support
AI/ML
- Rule-based alerts (NEWS2, MEWS)
- Time-series ML (XGBoost for deterioration prediction)
- LLM for clinical summaries (local + API hybrid)
- Explainable AI for transparency
Frontend
- Android (Primary) - Native app for bedside use
- Web Dashboard (React) - Hospital overview and analytics
Standards
- FHIR-ready data model
- HIPAA-compliant architecture
- Modular, hospital-agnostic design
MedObsMind/
├── backend/ # FastAPI backend
│ ├── app/
│ │ ├── models/ # SQLAlchemy models
│ │ │ ├── patient.py # Patient demographics
│ │ │ ├── vitals.py # Vital signs observations
│ │ │ └── alert.py # Clinical alerts
│ │ ├── api/ # API endpoints
│ │ │ ├── patients.py # Patient management
│ │ │ ├── vitals.py # Vitals recording
│ │ │ └── alerts.py # Alert management
│ │ ├── services/ # Business logic
│ │ │ ├── alert_engine.py
│ │ │ └── risk_scoring.py
│ │ ├── ml/ # ML models & scoring
│ │ │ ├── news2.py # NEWS2 calculator
│ │ │ ├── mews.py # MEWS calculator
│ │ │ └── predictor.py # ML predictions
│ │ └── core/ # Configuration
│ ├── tests/ # Unit tests
│ └── requirements.txt
├── app/ # Android application
│ └── src/main/
│ ├── java/com/medobsmind/app/
│ └── res/
├── web/ # React dashboard (future)
├── docs/ # Documentation
│ ├── API.md # API documentation
│ ├── DEPLOYMENT.md # Deployment guide
│ └── MEDICAL_SAFETY.md # Safety guidelines
└── README.md
- Vitals: HR, BP, SpO₂, RR, Temperature (continuous/periodic)
- Lab Values: CBC, metabolic panel, arterial blood gas
- Clinical Notes: Doctor observations and assessments
- Scores: NEWS2, MEWS, SOFA, APACHE-lite
- Privacy-First: No raw images initially, HIPAA-ready
🔍 Explainable AI - Clear reasoning for all alerts and suggestions
📝 Audit Logs - Complete trail for every alert and action
🏥 On-Device + Private Hosting - Data sovereignty options
✅ Validated Algorithms - NEWS2, MEWS based on clinical guidelines
- JWT authentication with role-based access control
- Encrypted data at rest and in transit
- HIPAA-compliant data handling
- Complete audit logging
- Regular security audits
Visit the live platform: https://sharmapank-j.github.io/MedObsMind 🎉
The website is already deployed and accessible! You can:
- ✅ View the platform overview
- ✅ Explore features and capabilities
- ✅ See the user interface
- ✅ Learn about the technology
One-Line Production Deploy:
curl -fsSL https://raw.githubusercontent.com/Sharmapank-j/MedObsMind/main/scripts/deploy.sh | sudo bashSee DEPLOYMENT_GUIDE.md for complete deployment instructions.
# Navigate to backend directory
cd backend
# Create virtual environment
python -m venv venv
source venv/bin/activate # On Windows: venv\Scripts\activate
# Install dependencies
pip install -r requirements.txt
# Setup environment
cp .env.example .env
# Edit .env with your database credentials
# Run database migrations
alembic upgrade head
# Start the server
uvicorn app.main:app --reloadAPI will be available at: http://localhost:8000
API Documentation: http://localhost:8000/docs
# Build debug APK
cd app
./gradlew assembleDebug
# Install on device
./gradlew installDebugSee ANDROID_BUILD.md for detailed instructions.
- Backend README - Backend setup and API details
- Android Build Guide - Android app build instructions
- API Documentation - Interactive API docs (when running)
- Free for Students - Educational access
- Subscription for Hospitals - Per-bed or per-facility pricing
- Government / NGO Deployment - Subsidized access
- ICU Module - Premium add-on with advanced features
- Backend structure with FastAPI
- Core database models (Patient, Vitals, Alerts)
- NEWS2 scoring engine implementation
- Basic API endpoints
- Android vitals entry interface
- Complete vitals API with trends
- Real-time alert system
- Notification service
- Doctor dashboard mockup
- AI risk prediction models
- Lab integration
- Case timeline view
- Pilot testing at 1 hospital
- Advanced ML models
- ICU workflow tools
- Multi-hospital deployment
- Compliance certifications
MedObsMind has extensive documentation (130+ KB) covering every aspect:
- Complete Vision (21 KB) - End-to-end explanation
- Core identity, governance, design philosophy
- Rule-first approach: Rules → LLM → Human → Decision
- All 5 modules, data flow, hardware specs, roadmap
- LLM Architecture (19 KB) - Medical LLM design
- "Explain don't decide" principle
- LLaMA-3 8B, LoRA fine-tuning, RAG layer
- Safety guardrails, confidence scoring, deployment
- India Context (27 KB) - India-specific considerations ⭐ UPDATED
- Population parameters, disease patterns, drug dosing
- Infrastructure reality, language support, cultural factors
- Geographic & Endemic Diseases: 20+ diseases mapped by 6 regions, 4 seasons
- Northeast: Malaria, JE, Scrub Typhus | Eastern: Kala-azar, Cholera, Filariasis
- Northern: High Altitude Illness, Goiter | Western: Fluorosis, Heat Stroke, Snake Bite
- Southern: Dengue, Leptospirosis | Central: Malaria (high burden)
- Prevention strategies (vector, water, environmental, zoonotic)
- Public health integration (IDSP, NVBDCP, NTEP)
- Disease elimination programmes (Kala-azar, Filariasis, Malaria, TB)
- Regulatory compliance (ICMR, CDSCO, DPDP Act)
- AETCOM, NMC & NBEMS Integration (34 KB) - Medical education & govt schemes ⭐ UPDATED
- AETCOM module integration (Ethics, Communication, Attitude)
- NMC competency framework (K/KH/S/D progression)
- NBEMS postgraduate training (DNB/DrNB/FNB, CanMEDS framework)
- Indian Government Healthcare Schemes: PMJAY, ABDM, NHM, e-Sanjeevani, Jan Aushadhi (8 central schemes)
- State Health Insurance Schemes: 12 major schemes (Maharashtra, Tamil Nadu, Karnataka, AP/Telangana, Rajasthan, Kerala, West Bengal, Delhi, Gujarat, Odisha, Chhattisgarh, Punjab)
- Education mode features for MBBS students and residents
- Ethical AI in medical education and public healthcare
- AI Architecture (12 KB) - Edge + Cloud design
- Feature Matrix (7 KB) - Features by user type
- ICU MVP Roadmap (12 KB) - 3-phase plan
- Cost-Impact Model (13 KB) - Financial & social impact
- Governance Framework (14 KB) - Ethics, oversight, and NMC compliance
- Backend README - API documentation
- Android Build Guide - App build instructions
- Working Apps Inventory - Repository status
- Implementation Status - Current progress and roadmap
- Contributing Guidelines - How to contribute
- Code of Conduct - Community standards
- License - MIT with medical disclaimer
This is currently a solo development project for Indian hospitals. Contributions welcome for:
- Medical algorithm validation
- UI/UX improvements
- Testing and documentation
- Hospital-specific integrations
See CONTRIBUTING.md for detailed guidelines.
© 2026 MedObsMind. All rights reserved.
- Live Website: https://sharmapank-j.github.io/MedObsMind ⭐ VISIT NOW
- Custom Domain: https://medobsmind.in (After DNS setup)
- API Docs: https://api.medobsmind.in/docs (After backend deploy)
- Email: support@medobsmind.in
- GitHub Issues: Report bugs or request features
- GitHub Repository: https://github.com/Sharmapank-j/MedObsMind
Build the most affordable, effective ICU monitoring system globally, with India leading the way in ethical AI healthcare.
Create a nationwide network of intelligent ICU monitoring that prevents thousands of avoidable deaths annually through early detection.
Develop technology in India, for India, that becomes the gold standard for resource-efficient healthcare AI worldwide.
This software is for clinical decision SUPPORT only.
- NOT approved for autonomous medical decisions
- Requires trained healthcare professional oversight
- All alerts and suggestions are advisory
- Clinical judgment always supersedes system recommendations
- Consult local regulations before clinical use
MedObsMind - Intelligent patient monitoring for better clinical outcomes. 🏥❤️
MedObsMind was born from a medtech hackathon with a simple yet powerful question: How can we make medical AI accessible to everyone in India, regardless of connectivity?
- 💡 Hackathon Concept: Identified the critical need for offline, privacy-first medical AI
- 🔬 Prototype Development: Built an on-device medical LLM trained on Indian medical data
- 🚀 Testnet to Reality: Evolved from testnet deployment to real-world applications
- 🌟 MedObsMind Today: Serving students, professionals, and healthcare providers nationwide
MedObsMind addresses the unique challenges of Indian healthcare through:
- Trained on Indian medical journals, research papers, and clinical guidelines
- Understanding of diseases, treatments, and scenarios specific to Indian population
- Contextually aware of local healthcare challenges and resource constraints
- 100% offline functionality - No internet required
- Built for India's diverse connectivity landscape
- Works perfectly in rural and remote areas with zero network coverage
- Addresses the digital divide in healthcare access
- Complete on-device processing - your data never leaves your device
- No cloud dependencies, no data transmission
- Medical information stays private and secure
- HIPAA considerations built into the architecture
- Students learn through safe simulations and game-like scenarios
- Practice medical procedures in virtual environments
- Make mistakes in simulated settings, not in real life
- Real-life scenarios without real-life risks
- Always accessible, even in the most remote areas
- Empowers rural health workers with medical knowledge
- Bridges the urban-rural healthcare gap
- Healthcare support wherever it's needed most
- Synthesized from Indian medical knowledge with ethical considerations
- Algorithmic fairness and bias mitigation
- Transparency in AI decision-making
- Human oversight and patient-centered design
Integrated with surgical cameras to help doctors identify anatomical structures that might be missed during complex procedures, providing real-time visual guidance and safety alerts.
Features:
- Real-time structure recognition
- Visual guidance during surgery
- Safety alerts and warnings
- Works with standard surgical cameras
Students practice procedures in realistic simulations, learning through trial and error in safe virtual environments before real-world application.
Features:
- Realistic medical scenarios
- Interactive learning modules
- Instant feedback and corrections
- Gamified learning experience
- Step-by-step procedure guidance
Healthcare professionals get on-demand assistance for diagnosis support, treatment protocols, and drug interactions - all offline and private.
Features:
- Clinical decision support
- Drug interaction database
- Treatment protocol guidance
- Evidence-based recommendations
- Differential diagnosis assistance
Empowering rural health workers with medical knowledge and guidance, even in areas with zero connectivity, bridging the urban-rural healthcare gap.
Features:
- Offline access to medical knowledge
- Multi-language support (Hindi, Tamil, Telugu, Bengali, etc.)
- Community health guidance
- Telemedicine support tools
- Basic diagnostic assistance
- Large Language Medical Model (LLMM) specialized for medical understanding
- Trained on comprehensive Indian medical literature and case studies
- Optimized for on-device inference with model compression techniques
- Supports multiple Indian languages
- Android Application: Native Android app with TensorFlow Lite / ONNX Runtime
- Web Interface: Progressive Web App for desktop and mobile browsers
- Cross-platform: Planned iOS support
- TensorFlow Lite for on-device ML inference
- ONNX Runtime for model flexibility
- Room Database for local data storage
- CameraX for AI camera integration
- Material Design 3 UI framework
Complete Android project structure included for building APK.
# Clone the repository
git clone https://github.com/Sharmapank-j/MedObsMind.git
cd MedObsMind
# Build debug APK
./gradlew assembleDebug
# Install on connected device
./gradlew installDebugFor detailed Android build instructions, see ANDROID_BUILD.md
This repository includes a serene, trustworthy landing page designed to communicate MedObsMind's core values of privacy, transparency, and ethical AI practices.
- Soft Gradient Design: Calming color palette with gradient background (#e6f7ff → #f0f9ff)
- 3D Visual Effects: Eye-catching card animations with subtle 3D transforms on hover
- Animated Brainwave SVG: Subtle, professional animations that convey care and observation
- Story Section: Complete journey from hackathon to production
- Vision Section: Detailed explanation of our mission and values
- Use Cases Section: Real-world applications with visual examples
- Technology Section: Technical details about on-device AI
- Mobile-First Design: Fully responsive from 320px to 4K displays
- WCAG AA Compliant: Accessible to all users with proper semantic HTML, ARIA labels, and keyboard navigation
# Start a local server
python3 -m http.server 8080
# Navigate to
http://localhost:8080/index.html- HTML5 (semantic markup)
- CSS3 (Grid, Flexbox, animations, transforms)
- Vanilla JavaScript (no frameworks)
- SVG graphics for scalable icons and visualizations
- Primary Gradient: #e6f7ff → #f0f9ff (soft, calming background)
- Accent Teal: #2a9d8f (trust, calmness, medical professionalism)
- Dark Text: #264653 (readability)
- White Cards: #ffffff with subtle shadows
- Calming and trustworthy
- Professional yet accessible
- Privacy and security emphasized
- Indian context celebrated
✅ 100% On-Device Processing - Complete privacy, no data transmission
✅ Works Offline - No internet connectivity required
✅ Indian Medical Context - Trained on Indian medical data and scenarios
✅ Multi-Language Support - Available in multiple Indian languages
✅ Accessibility - Reaches underserved and rural areas
✅ Educational Tool - Safe learning environment for students
✅ Professional Grade - Clinical decision support for healthcare providers
✅ Ethical AI - Transparent, fair, and bias-mitigated algorithms
✅ Free & Open - Accessible to all healthcare stakeholders
- Landing page with complete story and vision
- Android project structure
- Basic UI/UX design
- Complete TensorFlow Lite model integration
- Core medical query functionality
- ONNX Runtime implementation
- Model optimization for mobile devices
- Multi-language tokenizer
- Offline knowledge base
- AI camera assistance module
- Educational simulation platform
- Professional dashboard
- Rural health worker interface
- iOS application
- Desktop applications
- Model updates mechanism
- Community feedback system
We welcome contributions from developers, medical professionals, educators, and healthcare workers!
Areas where you can contribute:
- Medical data curation and validation
- Model training and optimization
- UI/UX improvements
- Documentation
- Testing and quality assurance
- Translations and localization
© 2026 MedObsMind. All rights reserved.
- Email: support@medobsmind.com
- GitHub Issues: Report bugs or request features
- Documentation: Wiki
- Built for Indian healthcare with love and dedication
- Inspired by the need for accessible, private medical AI
- Developed from hackathon concept to production-ready application
- Committed to bridging healthcare gaps across India
MedObsMind - Making medical intelligence accessible, private, and contextually relevant for every Indian, everywhere. 🇮🇳
A professional medical assistant platform powered by MedObsMind, a Large Language Medical Model (LLMM) specialized in medical informatics.
Dsquare Med-assist is the platform that provides access to MedObsMind LLMM - a specialized Large Language Medical Model designed specifically for healthcare and medical informatics applications.
- Interactive chat interface for medical consultations powered by MedObsMind LLMM
- Multiple performance modes (Maximum Accuracy, Balanced, Fast Response, Detailed Analysis, Standard)
- System prompts optimized for medical informatics
- Real-time message display with user and LLMM differentiation
- Speech-to-Text: Speak your medical questions directly to MedObsMind
- Text-to-Speech: MedObsMind LLMM responses are spoken back to you
- Hands-free operation for medical professionals
- Real-time voice recognition
- Real-time video analysis by MedObsMind LLMM for medical contexts
- Camera integration for visual medical information
- LLMM-powered interpretation of visual data
- Toggle between front and rear cameras
- Mute/unmute controls
- Easy switching between different performance modes
- Persistent settings storage
- Performance modes:
- Maximum Accuracy: Most accurate for complex medical queries
- Balanced: Recommended for general use (default)
- Fast Response: Faster response times
- Detailed Analysis: In-depth medical analysis
- Standard: Standard medical assistance
INTERNET: For AI API communicationRECORD_AUDIO: For voice input/outputCAMERA: For video interpretationMODIFY_AUDIO_SETTINGS: For audio control
- Start Chat: Launch Dsquare Med-assist to begin text-based chat with MedObsMind LLMM
- Voice Input: Tap the microphone button to speak your question to MedObsMind
- Video Consultation: Tap the video button to start live video interpretation with MedObsMind LLMM
- Configure Settings: Access settings menu to adjust MedObsMind performance mode
MedObsMind is a Large Language Medical Model (LLMM) - a specialized language model trained and optimized specifically for:
- Medical informatics
- Healthcare applications
- Clinical decision support
- Medical terminology and research
- Visual medical interpretation
- Evidence-based medical information
Unlike general-purpose language models, MedObsMind LLMM is purpose-built for the medical domain, ensuring higher accuracy and relevance for healthcare-related queries.
- Android SDK: Min SDK 24, Target SDK 34
- UI Framework: Material Design Components
- Speech: Android SpeechRecognizer and TextToSpeech APIs
- Camera: Camera2 API for live video
- Storage: SharedPreferences for settings persistence