AI-powered health prediction platform for clinical biomarker analysis
🔗 Live Demo: mira-health-app.streamlit.app
MIRA is a full-stack health prediction web application that combines Artificial Intelligence, Machine Learning, and Predictive Analytics to assess patient health risk from clinical biomarker data. Built as part of the MIRA — Medical Intelligence Robotic Automation platform, it automates health risk classification and generates AI-powered clinical assessments using Groq's LLaMA 3.3 70B model.
⚠️ Disclaimer: MIRA is a demonstration application for technical evaluation. It is not a certified medical device and must not be used for clinical diagnosis. Always consult a qualified healthcare professional.
| Feature | Description |
|---|---|
| 🔬 AI Health Assessment | LLaMA 3.3 70B via Groq API generates clinical risk assessments from biomarker data |
| 📊 Risk Classification | Automatic Low / Moderate / High risk scoring based on standard clinical thresholds |
| 🗄️ Full CRUD | Create, Read, Update, Delete patient records with persistent SQLite storage |
| 🔍 Patient Search | Live search by name or email across all records |
| ✅ Input Validation | Email format, DOB range, and numeric biomarker range validation |
| 📱 Responsive UI | Premium dark-theme interface with custom HTML/CSS built on Streamlit |
| 🔒 Secure | No API keys committed — environment-based secret management |
┌─────────────────────────────────────────────────┐
│ MIRA Stack │
├──────────────┬──────────────────────────────────┤
│ Frontend │ Streamlit + Custom HTML/CSS │
│ Backend │ Python 3.10+ │
│ Database │ SQLite (persistent local storage)│
│ AI Engine │ Groq API — LLaMA 3.3 70B │
│ Validation │ Custom Python utilities │
└──────────────┴──────────────────────────────────┘
| Biomarker | Unit | Normal Range | High Risk Threshold |
|---|---|---|---|
| Fasting Glucose | mg/dL | 70 – 99 | ≥ 126 (Diabetic range) |
| Haemoglobin | g/dL | 12.0 – 17.5 | < 8.0 (Severe anaemia) |
| Total Cholesterol | mg/dL | < 200 | ≥ 240 (High) |
MIRA-Medical-Automation/
│
├── app.py # Main Streamlit application — UI, routing, pages
├── database.py # SQLite CRUD operations — create, read, update, delete, search
├── groq_client.py # Groq LLaMA 3.3 70B API integration — health prediction engine
├── utils.py # Input validation — email, DOB, numeric range checks
├── requirements.txt # Python dependencies
├── .gitignore # Excludes secrets, DB files, pycache
└── .streamlit/
└── secrets.toml # API key config — NOT committed to GitHub
- Python 3.10 or above
- A free Groq API key → console.groq.com
git clone https://github.com/Rosesharma13/MIRA-Medical-Automation.git
cd MIRA-Medical-Automationpip install -r requirements.txt
Create the secrets file:
mkdir .streamlit
Add your key inside .streamlit/secrets.toml:
GROQ_API_KEY = "your_groq_api_key_here"
streamlit run app.py
Open → http://localhost:8501
- Push this repo to GitHub
- Go to share.streamlit.io and sign in with GitHub
- Click New app → select this repository → set main file to
app.py - Go to Settings → Secrets and add:
GROQ_API_KEY = "your_groq_api_key_here"
- Click Deploy — live URL generated instantly
- No API keys or credentials are stored in this repository
- Patient data remains in local SQLite — not transmitted to any external server
- Only anonymised biomarker values (no PII) are sent to Groq for AI analysis
.streamlit/secrets.tomland*.dbfiles are excluded via.gitignore
Rose Sharma
- 🌐 Portfolio: rosesharma13.github.io
- 💼 LinkedIn: linkedin.com/in/rose-sharma13
- 📧 Email: rosesharmaa132003@gmail.com