A recent study from Harvard Medical School introduced TxGNN, an innovative AI tool designed to identify potential drug candidates for over 17,000 rare and neglected diseases.
The research, published in Natural Medicine Sept. 25, found that TxGNN is nearly 50% more effective at identifying drug candidates and 35% more accurate in predicting contraindications than existing methods.
The model links conditions to existing drugs, both FDA-approved and experimental, by leveraging data on DNA, clinical notes and gene activity. The model also offers insights into potential side effects, potentially speeding up the lengthy drug discovery process.
TxGNN has been made available for free, encouraging its adoption by clinician-scientists in the search for new therapies.