Go there, Hugh Laurie. There’s a new doctor’s house in town.
Harvard Medical School Associate Professor Marinka Zitnik and her lab announced the development of TxGNN, an artificial intelligence model that uses neural networks to suggest new therapeutic applications for existing drugs.
The model, published by Zitnik and his co-authors last month in Nature Medicine, allows researchers to choose from about 8,000 drugs to treat more than 17,000 diseases, including some that are poorly understood.
She likened the model to Gregory House, the fictional protagonist of the Fox medical drama of the same name, who has an uncanny ability to diagnose patients with abnormal conditions.
TxGNN “can transfer knowledge from well-studied diseases to those with sparse data, enabling the identification of drug candidates without the need for large-scale clinical trials or new drug development,” Zitnik said in an email. I wrote it in “This reduces costs and time to market, and provides a much-needed means to advance treatments for neglected diseases.”
“Similar to the TV series Dr. House, these models approach medical challenges by combining various medical facts and clues that may not be obvious or directly related. ” she added.
Unlike other AI models, whose inner workings are often unknown even to developers, TxGNN includes an “explanation module” that allows users to see the medical rationale behind its recommendations.
Explanatory modules “can significantly bridge the gap between research and practical application,” wrote Chao Yang, a researcher at Vanderbilt University whose research focuses on similar topics.
Yang said models like TxGNN could far exceed humans’ ability to process and analyze medical data to discover new treatments.
This allows “complex knowledge to be easily transferred into uncharted territory with minimal model training and small amounts of labeled data, an outcome that is nearly impossible with traditional methods.” Yang wrote.
Michael W. Maloney, a scientist at the University of Chicago, said in an email that he is “very optimistic about the use of AI in drug discovery and drug repurposing,” adding that TxGNN “is a technology that leverages the intuition and capabilities of human expert researchers.” He will play an active role in the field in which he will be promoted.” There will be a shortage. ”
Still, he wrote, he maintains “a level of vigilance that I would imagine would be shared among the drug discovery and development community.”
Maloney wrote that he was concerned that the AI model could produce false positives.
“This risk may be exacerbated by irresponsible actions by physicians,” Maloney wrote. “We don’t want AI to do doctors’ homework.”
As models like hers become more prominent, Zittnik writes, “regulatory frameworks will need to adapt to this new paradigm.”
“Current regulations may not fully account for drug candidates generated by AI, so regulatory bodies such as the FDA and EMA need to establish clear guidelines for using AI in drug development. “There is a need,” she wrote.
