AI-powered drug repurposing enables new treatments for rare, undiagnosed diseases

Scientists at Harvard Medical School are using artificial intelligence (AI) to solve the problem of treating rare and neglected diseases. They have developed a new method called TxGNN that could help identify new drug candidates for the more than 7,000 rare and undiagnosed diseases. Details of the method will be published in a Natural medicine Article titled “A fundamental model for clinically oriented drug repurposing.”

So far only 5–7% of rare and neglected diseases have an FDA-approved drug. The rest remain either untreated or undertreated. AI-based applications like TxGNN can help address this challenge by driving the discovery of new therapies based on existing drugs and providing hope to both patients and physicians.

According to its developers, TxGNN is the first method specifically designed to identify drug candidates for rare diseases and conditions with no treatment. It identified drugs from existing drugs for more than 17,000 diseases, many of which have no treatment yet. This represents the largest number of diseases yet treated by a single AI model. And that's just a start, the researchers say, because the model could be applied to even more disease types than those used for this study.

So how does it work? The tool is a graph model with two central functions. The first identifies treatment candidates along with possible side effects, and another function explains the reasoning behind the decision. It was trained on large amounts of data, including genomic data, cell signaling information, gene activity data, clinical notes, and more. The researchers tested and refined the model by having it perform various tasks and validated its performance using 1.2 million patient records.

The team then asked the model to select drug candidates for different diseases. They also asked the model to make some additional predictions. For example, they asked the tool to predict patient characteristics that would prevent the selected drug candidates from being used to treat certain populations. In another task, the scientists tasked the tool with identifying small molecules that effectively block the activity of some proteins involved in disease-causing pathways and processes.

In another test, the researchers asked the model to identify drugs for three rare diseases that were not included in its training dataset. The diseases were a neurodevelopmental disorder, a connective tissue disease, and a disease that causes water imbalance. They then compared the model's treatment suggestions with clinical knowledge about the effects of the suggested drugs. The results showed that TxGNN's recommendations were not only consistent with current medical knowledge, but that the tool also provided the rationale behind its decision.

The Harvard team is making its tool available for free and hopes that it will be used by clinicians and scientists. Of course, any therapies identified using the model will need to be re-evaluated in terms of dosage and timing of administration before being used in patients. The team is already working with several rare disease foundations to find potential treatments for the diseases that interest them.

“We hope to use this tool to identify new therapies for the entire spectrum of diseases. But when it comes to rare, ultra-rare and neglected conditions, we believe this model could help close or at least narrow a gap that leads to serious inequalities in health care,” says Dr. Marinka Zitnik, lead researcher and assistant professor of biomedical informatics at the Blavatnik Institute at Harvard Medical School.