Millions of individuals worldwide suffer from diseases for which there are no available treatments. While the Food & Drug Administration (FDA) has approved roughly 3,000 drugs to address a corresponding number of diseases, there remain an additional 9,000 diseases without a single approved therapy. Given that numerous diseases share common underlying mechanisms of action, and individual drugs can target multiple mechanisms, the existing pool of 3,000 FDA-approved drugs holds the promise of addressing the 9,000 diseases that currently lack therapeutic options. 
MATRIX (Machine Learning/Artificial Intelligence-Enabled Therapeutic Repurposing in eXtended uses) aims to develop computational methodologies for identifying the FDA-approved drugs most likely to treat diseases with inadequate treatment options, and to identify and validate top candidates for drug repurposing using these methodologies.

MATRIX Phase 1 develops the first comprehensive scoring system that queries the world’s biomedical knowledge of “all drugs vs all diseases” to predict the efficacy for every drug to treat every human disease. The resulting information on the pharmaco-phenome will be made available open-source, allowing researchers to view the probability of efficacy across the entire landscape of FDA-approved drugs and human diseases.  

Phase 2 advances at least 30 top repurposing opportunities to preclinical and clinical validation, prioritizing the evidence necessary to establish therapeutic validity. This enables faster, more cost-efficient progression to clinical use than standard discovery pathways. Scientific validation of AI models informs technical development through closed-loop learning, continuously improving MATRIX predictive performance.