MeDI (MATRIX Indication List)

Overview

MEDI (MATRIX Indication List), now maintained through the medic repository, provides structured drug-indication and drug-contraindication relationships for drug repurposing research. The original Every Cure repository is archived and points users to the maintained medic repository.

Data Content

MEDI includes:

• Drug indications extracted from DailyMed product labels
• Drug contraindications identified from safety information
• Mappings to standardized disease ontology terms (MONDO)
• Ground truth datasets for training predictive models

Extraction Methodology

The dataset is generated through:

• Automated extraction from FDA DailyMed using LLMs
• Integration with EMA (European Medicines Agency) and PMDA (Pharmaceuticals and Medical Devices Agency) indications data
• Systematic normalization and mapping to MONDO ontology terms
• Quality control and validation processes

Use Cases

MEDI serves as:

• Training data for drug repurposing prediction models
• Reference dataset for validating computational drug repositioning approaches
• Source of structured drug-disease relationships for knowledge graphs
• Ground truth for evaluating machine learning models in pharmacology

Technical Implementation

The project uses:

• Jupyter Notebooks for data processing pipelines
• Automated data ingestion from DailyMed, EMA, and PMDA sources
• MONDO ontology integration for disease term standardization
• Version-controlled releases on GitHub

Developed By

Every Cure organization, with primary development by Marcello DeLuca and contributors.

Access

The dataset and processing code are freely available through the GitHub repository with regular versioned releases.