AI model identifies existing drugs that can be repurposed to treat rare diseases

There are over 7,000 rare and undiagnosed diseases worldwide. Although each condition affects only a small number of people, collectively these diseases have a significant human and economic impact, affecting some 300 million people worldwide.

Yet, with only 5-7% of these diseases having an FDA-approved drug available, they remain largely untreated or undertreated.

Developing new drugs is a daunting challenge, but a new artificial intelligence tool may propel the discovery of new therapies from existing drugs, offering hope to patients with rare and neglected diseases and the clinicians who treat them.

The AI ​​model, called TxGNN, is the first developed specifically to identify drug candidates for rare diseases and conditions with no treatment.

It has identified drug candidates among existing drugs for more than 17,000 diseases, many of which have no existing treatments. This represents the largest number of diseases that a single AI model can address to date. The researchers note that the model could be applied to even more diseases beyond the 17,000 it worked on in initial experiments.

The work, described on September 25 in Natural medicinewas conducted by scientists at Harvard Medical School. The researchers have made the tool freely available and want to encourage clinician-researchers to use it in their search for new therapies, particularly for diseases for which treatment options are non-existent or limited.

“With this tool, we aim to identify new therapies across the spectrum of diseases, but when it comes to rare, ultra-rare and neglected diseases, we anticipate that this model could help close, or at least reduce, a gap that creates serious health disparities,” said lead researcher Marinka Zitnik, assistant professor of biomedical informatics in the Blavatnik Institute at HMS.

“This is precisely where we see the promise of AI in reducing the global burden of disease, in finding new uses for existing drugs, which is also a faster and more cost-effective way to develop therapies than designing new drugs from scratch,” added Zitnik, who is an associate faculty member at the Kempner Institute for the Study of Natural and Artificial Intelligence at Harvard University.

The new tool has two main functions: one identifies treatment candidates and possible side effects, and the other explains the rationale for the decision.

In total, the tool identified drug candidates from nearly 8,000 drugs (both FDA-approved drugs and investigational drugs currently in clinical trials) for 17,080 diseases, including conditions for which there are no treatments. It also predicted which drugs would have side effects and contraindications for specific conditions, something the current drug discovery approach identifies primarily through trial and error in early clinical trials focused on safety.

Compared to leading AI models for drug repurposing, the new tool was found to be on average 50% better at identifying drug candidates. It was also found to be 35% more accurate at predicting which drugs would have contraindications.

Benefits of using already approved drugs

Repurposing existing drugs is an attractive way to develop new treatments because it relies on drugs that have been studied, have well-understood safety profiles, and have gone through the regulatory approval process.

Most drugs have multiple effects that extend beyond the specific targets for which they were originally developed and approved. But many of these effects remain unknown and understudied during initial testing, clinical trials, and evaluations, and only emerge after years of use by millions of people. Indeed, nearly 30% of FDA-approved drugs have acquired at least one additional therapeutic indication after initial approval, and many have acquired dozens of additional therapeutic indications over the years.

This approach to drug repurposing is haphazard at best. It relies on patient reports of unexpected beneficial side effects or physicians’ intuition about whether a drug is appropriate for a condition for which it was not intended, a practice known as off-label use.

“We tend to rely on luck and chance rather than strategy, which limits drug discovery to diseases for which drugs already exist,” Zitnik said.

The benefits of drug repurposing extend beyond untreated diseases, Zitnik noted.

“Even for the most common diseases with approved treatments, new drugs could offer alternatives with fewer side effects or replace ineffective drugs for some patients,” she said.

What makes the new AI tool better than existing models?

Most current AI models used for drug discovery are trained on a single disease or a few conditions. Rather than focusing on specific diseases, the new tool has been trained to use existing data to make new predictions. It does this by identifying features that are common to multiple diseases, such as common genomic aberrations.

For example, the AI ​​model identifies shared disease mechanisms based on common genomic foundations, allowing it to extrapolate from a well-understood disease with known treatments to a poorly understood disease with no treatments.

This capability, the research team said, brings the AI ​​tool closer to the kind of reasoning a human clinician might use to generate new ideas if they had access to all the pre-existing knowledge and raw data that the AI ​​model has but that the human brain can’t access or store.

The tool was trained on large amounts of data, including information about DNA, cell signaling, gene activity levels, clinical notes, and more. The researchers tested and refined the model by asking it to perform various tasks. Finally, the tool’s performance was validated on 1.2 million patient records and asked to identify drug candidates for various diseases.

The researchers also asked the tool to predict which patient characteristics would make the identified drug candidates contraindicated for certain patient populations.

Another task was to ask the tool to identify existing small molecules that could effectively block the activity of certain proteins involved in disease-causing pathways and processes.

In a test designed to assess the model’s ability to reason like a human clinician, the researchers asked the model to find drugs for three rare diseases it had not seen in its training: a neurodevelopmental disorder, a connective tissue disease and a rare genetic disease that causes fluid imbalance.

The researchers then compared the model’s drug treatment recommendations with current medical knowledge about how the suggested drugs work. In each example, the tool’s recommendations matched current medical knowledge.

In addition, the model not only identified drugs for all three diseases, but also provided the reasons behind its decision. This explanatory feature allows for transparency and can increase physician confidence.

The researchers caution that any therapies identified by the model would require additional assessment of dosage and timing. But, they add, with this unprecedented capability, the new AI model would accelerate drug repurposing in ways that were not possible until now. The team is already collaborating with several rare disease foundations to help identify possible treatments.