ai_action_summit_hackathon_an2i

A comprehensive medical prescription analysis system that leverages AI to enhance patient care and medical decision-making.

Our Approach

Our system implements a multi-faceted approach to medical analysis and prescription safety:

1. Real-time Conversation Analysis

   • Automated extraction of medical context from doctor-patient dialogues
   • Dynamic generation of relevant medical queries
   • Intelligent follow-up question suggestions

2. Prescription Safety System

   • Automated anomaly detection in prescriptions
   • Cross-referencing with patient medication history
   • Real-time alerts for potential drug interactions
   • Analysis of patient medication adherence patterns

3. Comprehensive Reporting

   • Side-by-side display of medical reports and prescription analysis
   • Detailed symptom and pathology documentation
   • Clear treatment plans with safety considerations
   • Intelligent summarization of medical findings

Generated Reports

Access our sample medical reports and analysis:

• Technical Report
• Business Plan

Flask API Service

The project includes a robust Flask API that provides various medical analysis endpoints:

API Endpoints

• /detect-prescription-anomalies - Analyzes prescriptions for potential conflicts
• /extract-ordonnance - Extracts structured data from medical prescriptions
• /summarize-ordonnances - Generates summaries of multiple prescriptions
• /search-medical-articles - Searches for relevant medical literature
• /fetch-article-abstract/{pmid} - Retrieves abstracts of medical articles
• /generate-search-summary - Summarizes medical research findings
• /generate-follow-up-questions - Generates relevant follow-up questions
• /extract-pertinent-points - Extracts key medical points from conversations
• /generate-search-propositions - Suggests relevant medical search queries
• /generate-report - Generates comprehensive medical reports

The API runs on port 5000 by default and accepts JSON payloads for POST requests.

Streamlit Dashboard

The project includes a user-friendly Streamlit dashboard that provides a graphical interface to interact with the API services. The dashboard offers:

Features

• Interactive navigation sidebar for different functionalities
• Form-based input for medical data
• Real-time API response display
• Error handling and user feedback
• Medical article search interface with abstract viewing
• Report generation interface
• Prescription anomaly detection with visual alerts

Dashboard Pages

1. Prescription Analysis

   • Anomaly detection
   • Ordonnance extraction
   • Prescription summarization

2. Medical Research

   • Article search with configurable results
   • Abstract viewing
   • Search summary generation

3. Medical Documentation

   • Follow-up question generation
   • Pertinent point extraction
   • Report generation

The Streamlit dashboard connects to the Flask API at http://localhost:5000.

Development

Notes

• commit messages should follow conventional commits spec: https://www.conventionalcommits.org/en/v1.0.0/ (see sections: Summary, Examples)
• work must be merged from feature branches into main branch using Pull Requests. Once your code is ready to be reviewed, please mark me as a reviewer, which will notify me you're ready fir code review

How-to's

Install

1. install uv: https://docs.astral.sh/uv/getting-started/installation/#standalone-installer
2. run

   uv sync --all-extras --dev

3. Before running scripts add the current directory to the PythonPath:

export PYTHONPATH=/home/azzedine/Projects/ai_action_summit_hackathon_an2i:$PYTHONPATH

Running the Services Locally

1. Start the Flask API:

python3 -m workspace.src.api

2. Start the Streamlit Dashboard:

streamlit run workspace/app/main.py

Docker Deployment

The project can be run using Docker Compose, which sets up all required services automatically.

Prerequisites

1. Install Docker and Docker Compose on your system
2. Copy the environment variables file:

   cp .env.example .env

3. Update the .env file with your credentials:
   • API_KEY: Your API key
   • PROVIDER_URL: Your provider URL
   • MongoDB credentials (if using MongoDB):
     • USERNAME_MONGODB
     • PASSWORD_MONGODB
     • INSTANCE_MONGODB
     • REGION_MONGODB
     • PATH_TLS_CERTIFICATE_MONGODB

Services

The Docker Compose setup includes the following services:

• API Service (Port 5000): Flask API server
• Streamlit Dashboard (Port 8501): Web interface
• Redis (Port 6379): Message broker for Celery
• Celery Workers:
  • LLM Worker: Handles language model tasks
  • API Worker: Handles API-related tasks
• Flower (Port 5555): Celery monitoring interface (credentials: admin/admin)

Running with Docker Compose

1. Build and start all services:

   docker compose up --build

2. Access the services:

   • Streamlit Dashboard: http://localhost:8501
   • Flask API: http://localhost:5000
   • Flower Dashboard: http://localhost:5555

3. Stop all services:

   docker compose down

Lint (check & fix)

1. open a new terminal in the root of this repo and run:

uv run ruff check --fix
uv run ruff format
uv run mypy .

Run tests

1. open a new terminal in the root of this repo and run:

uv run pytest