The new tool outperformed leading AI models for drug repurposing by being nearly 50% better at identifying drug candidates and 35% more accurate in predicting contraindications. (Source: Image by RR)

TxGNN Pinpoints Drug Candidates From Existing Medicines for Over 17,000 Diseases

There are over 7,000 rare and undiagnosed diseases affecting 300 million people globally, with only 5-7% of these conditions having an FDA-approved drug, leaving most untreated. A new AI tool called TxGNN offers hope by identifying drug candidates from existing medicines for over 17,000 diseases, many of which lack treatments. Developed by Harvard Medical School scientists, the tool is available for free to clinicians and significantly outperforms existing AI models in identifying both drug candidates and potential side effects or contraindications.

TxGNN’s strength lies in its ability to repurpose existing drugs, which have known safety profiles and regulatory approval. Nearly 30% of FDA-approved drugs have gained new uses after initial approval, though this process often relies on chance discoveries. As reported in medicalxpress.com, TxGNN strategically predicts additional uses for drugs by identifying shared mechanisms between diseases, offering a more efficient path to finding new treatments for rare and neglected diseases.

The AI model’s training involved large amounts of data, such as DNA information, gene activity, and clinical records. It was validated using 1.2 million patient records and successfully identified drug candidates for various diseases, including those it had not encountered during training. In tests, the model’s recommendations aligned with current medical knowledge, showcasing its potential to support clinicians in generating novel treatment ideas for poorly understood diseases.

While the model’s drug recommendations will require further evaluation, particularly regarding proper dosing and timing, its unprecedented capacity to identify potential treatments for rare and neglected conditions represents a major breakthrough in drug repurposing. The ability to expedite the discovery process for these conditions, which often receive little attention or funding, offers new hope for patients and clinicians alike. Researchers believe that TxGNN can significantly reduce the time and cost associated with traditional drug discovery methods, making it possible to explore a wider range of therapeutic options for diseases that have long gone untreated.

read more at medicalxpress.com