An adverse drug events explorer Streamlit app powered by the OnSIDES database, Haystack, and MCP-Alchemy.
This application leverages large language models and the OnSIDES database to provide a natural language interface for exploring adverse drug events. MedOryx can help you:
- Query adverse drug events (ADEs) for specific medications
- Find drugs that cause particular side effects
- Compare adverse events across drug classes
- Explore serious ADEs from drug labels' Boxed Warnings section
- Access comprehensive adverse event data extracted from FDA drug labels
- OnSIDES: Comprehensive database of adverse drug events extracted from drug labels
- MCP-Alchemy: MCP Server for Text-to-SQL
- Haystack: AI orchestration framework
- Streamlit: Web app framework
- Python 3.11+
- Anthropic API key
- uv - Modern Python package manager
-
Clone this repository:
git clone https://github.com/oryx1729/med-oryx-ade cd med-oryx-ade -
Install dependencies with uv:
uv pip install . -
Set your Anthropic API key as an environment variable:
export ANTHROPIC_API_KEY=your_api_key_here -
Set the SQLAlchemy database URL for MCP-Alchemy:
export DB_URL=postgresql://user:password@localhost/dbname
-
Start the application:
uv run streamlit run app.py -
Open your browser and navigate to the provided URL (typically http://localhost:8501)
-
Enter your questions about adverse drug events in the chat interface
The agent can answer questions like:
- What are the most common side effects of metformin?
- Show me all cardiovascular adverse events associated with statins.
- Which antidepressants have the fewest reported gastrointestinal side effects?
- List all drugs that can cause severe liver toxicity.
- Compare the adverse events of amoxicillin and azithromycin.
- What pediatric-specific adverse events are reported for methylphenidate?
- Show me drugs with black box warnings for QT prolongation.
- Which antipsychotics have the highest risk of weight gain?